The Linnean shortfall in action: amphibian surveys reveal undocumented species in the Nguru Mountains, Tanzania

The Nguru Mountains in the central Eastern Arc Mountains (EAM) consist of three main protected areas (Mkingu, Kanga and Magotwe), which form an important hotspot for endemic species of vertebrates and plants in Tanzania. However, the herpetofauna of the Nguru are relatively poorly understood, and rapid deforestation highlights the urgent need for comprehensive biodiversity studies. Using standardized survey methods, we conducted amphibian surveys in Nguru Mountains between May 2022 and November 2024, and together with literature and databases search, we provide an annotated amphibian checklist. Using molecular barcoding, we identify 42 lineages of amphibians in the Nguru Mountains, 15 of which are candidates for species new to science. Twenty-eight lineages are endemic to the Eastern Arc Mountains of Tanzania, eight of them being restricted to Nguru. Treating these new records as species would rank the Nguru Mountains third out of the 12 Eastern Arc Mountain blocks for the number of endemic amphibians, after Udzungwa and Uluguru. However, the identified lineages await formal description, and the partial sampling and taxonomic issues detected here, are common to all Eastern Arc Mountains. This study therefore highlights the reality of the Linnean shortfall even in recognized biodiversity hotspots, and how taxonomic work is required to remedy this. The impact of deforestation on Nguru amphibians is yet to be quantified, however this study helps to form a baseline for future studies.

Over the last two decades our understanding of amphibian diversity across the fragmented forests of the Eastern Arc Mountains (EAM) of Tanzania has improved, leading to better understanding of each mountain (e.g. Loader et al., 2004; Menegon et al., 2008; Ngalason and Mkonyi, 2011; Lyakurwa et al., 2019, 2024; Liedtke et al., 2022), the description of many new species (e.g. Channing et al., 2005; Loader et al., 2006, 2010a, 2010b, 2015; Menegon et al., 2007, 2011a, 2022; Poynton et al., 2008; Greenwood et al., 2020; Lawson et al., 2023a), and increasing awareness of the threats to species and their habitats (e.g. Channing et al., 2005; Liedtke et al., 2022; Lawson et al., 2023b; Sewell et al., 2024). However, these efforts have not been applied equally to all mountain blocks (Rovero et al., 2014; Lyakurwa et al., 2024), and even within individual blocks, some forest patches have been sampled repeatedly, while others have been overlooked (e.g. Loader et al., 2011; Menegon et al., 2011b; Liedtke et al., 2022; Lyakurwa et al., 2024). Not surprisingly, the better surveyed EAM blocks like East Usambara, Udzungwa, and Uluguru, rank highly for biodiversity metrics, and as a consequence are also prioritised for conservation (Burgess et al., 2007; Rovero et al., 2014; Meng et al., 2016; Lyakurwa et al., 2019; Ract et al., 2024). A number of other mountain blocks like Malundwe, Mahenge, Nguru, Nguu, Rubeho and Pares have received much less attention.

The inclusion of molecular barcoding approaches in recent surveys of EAM forests have identified large numbers of undescribed, often cryptic amphibian species (Loader et al., 2010a, 2013, 2015; Liedtke et al., 2016; Greenwood et al., 2020; Menegon et al., 2020; Lawson et al., 2023a). These studies show that cryptic species tend to be endemic to each block (Rovero et al., 2014; Gereau et al., 2016; Lyakurwa et al., 2024). It stands to reason therefore that the lesser-known blocks likely harbour similar, undocumented diversity. This is disconcerting, as high rates of habitat loss and decline in amphibians in the EAM have been reported (Channing et al., 2006; Menegon et al., 2020; Tonelli, 2021; Liedtke et al., 2022; Lyakurwa et al., 2025). The high number of cryptic species awaiting discovery, confronted with rapid rates of habitat loss in these under-explored areas may therefore result in poorly targeted conservation actions.

The Nguru Mountain block forms part of the central landscape of the EAM and is among the most significant blocks in terms of size and number of endemic vertebrate species (Burgess et al., 2007; Gereau et al., 2016; Lyakurwa et al., 2024). The block covers a total area of 1,700 km 2, though only 340 km 2 is forested (Figure 1; Burgess et al., 2007). The landscape extends from 380 to 2400 m a.s.l and is characterised by a mosaic of lowland, submontane and montane forests (Pócs et al., 1991; Lovett and Pócs, 1993; Supplementary Figure 1). With the current forest area coverage already at only 32.3% of the historical extent, the area is continuing to experience rapid forest loss, mostly due to logging, fire and encroachment by small scale farms (Hamunyela et al., 2020; Supplementary Figure 1). Woodlands and dry forests are also present on the lowlands and drier slopes of the mountains (Pócs et al., 1991; Lovett and Pócs, 1993; Romdal, 2001).

Three main forest patches are gazetted in Nguru: Mkingu Nature Forest Reserve, Kanga and Magotwe Forest Reserves (Figure 1; Lovett and Pócs, 1993; Doggart et al., 2017). Mkingu Nature Forest Reserve (referred to as Mkingu in this paper) was formed recently by combining what used to be Nguru South and Mkindo Forest Reserves and forms the main block of the Nguru Mountains (Figure 1; Doggart et al., 2017). Kanga Forest Reserve (herein referred to as Kanga) is just 10 km north of the main block, with Magotwe Forest Reserve (referred to as Magotwe herein) on the foothill (between Kanga and Mkingu; Figure 1 and Supplementary Figure 1). The original forest cover extended to the lowland areas, but currently, only very small intact lowland forest exists (Romdal, 2001). Magotwe, which is protected as a forest reserve, used to form a significant portion of the lowland forest (Lovett and Pócs, 1993), but has been severely impacted by anthropogenic activities including crop cultivation and settlements (Doggart and Loserian, 2007; Supplementary Figure 1). A number of small forests managed by village authorities are also present (e.g. Lusungulu, Mtibwa, Mabundi Mtwange, Mkangala, Mbogo and Disalasala; Figure 1; Lovett and Pócs, 1993), but these are generally small and/or in poor condition compared to the main forests (Lovett and Pócs, 1993), which are under the management of the central government. The Nguru Mountains also support large number of people, with an estimated population in 2022 of 211,682 people in the immediate vicinity, mostly small-scale farmers (URT, 2024), with one village found within Mkingu (Doggart et al., 2017). These numbers have substantially increased since the 2012 census (136,599 people; URT, 2013).

The first published herpetological exploration in Nguru was by Emmrich (1994) in the eastern lowland of Mkingu, resulting in range extensions of species known from other EAM blocks. Since then, more herpetological surveys were conducted in the Mkingu and the Kanga region, mostly between 2000 to 2012, leading to descriptions of new, endemic species including, Arthroleptis nguruensis, A. kidogo, Callulina hanseni, C. kanga and C. meteora (Blackburn, 2009; Menegon et al., 2008, 2011a; Poynton et al., 2008; Loader et al., 2010b). Yet, large portion of these forests are still poorly known and previous surveys highlighted the presence of a number of species that could not be identified with certainty (Owen et al., 2007; Menegon et al., 2008). It is evident that our understanding of the amphibian diversity of the Nguru Mountains is not complete. The knowledge gaps highlighted above i.e the discrepancy in the actual number of described species (Linnean shortfall) and the paucity of species distribution data (the Wallacean shortfall) can be mitigated in various ways, including new approaches for quantifying species richness, simplified species description process, integrative taxonomic approaches, long term data collection, targeted survey efforts, improved species distribution modeling, use of historical records, biodiversity databases and citizen science platforms (Fouquet et al., 2007; Brito, 2010; Padial et al., 2010; Beck et al., 2013; Gower et al., 2013; Tolley et al., 2016; Nogueira et al., 2019; Vences, 2020; Ngwava et al., 2021; Liedtke et al., 2022; Dufresnes et al., 2023; Carné and Vieites, 2024; Lyakurwa et al., 2025).

The objective of this study was to compile an up-to-date, annotated checklist of Nguru Mountain amphibians through dedicated fieldwork, DNA barcoding and a revision of existing amphibian studies, and in so doing, highlight the extent of the Linnean shortfall likely affecting the Ngurus as well as other important forests of the Tanzanian EAM.

This study is a result of seven dedicated surveys to Mkingu Nature Forest Reserve (Mkingu) and Kanga Forest Reserve (Kanga) between May 2022 – November 2024 (Supplementary Table 1) and compilation of data from literature and biodiversity databases. Nine sites were established in Mkingu (Figure 1) covering different elevation bands (900–2010 m.a.s.l.), which are usually associated with different vegetation types (Lovett and Pócs, 1993; Lovett, 1996, 1998; Lovett and Thomas, 1988), and sampled at least two times, using drift fence with bucket pitfall traps and time constrained surveys. This therefore marks six additional sites to the six that were surveyed previously by Menegon et al. (2008). Two bucket pitfall lines each with seven buckets and a total of 30 m long transparent drift fence were set in a Y-shape at five sites (Figure 1) following McDiarmid et al. (2012) between May and June 2023. Each trap was monitored twice per day for a period of seven consecutive nights, making a total of 98 trap nights per site and 490 trap nights for the entire study. Time constrained audio-visual surveys were conducted in all nine sites by at least two people for two hours between 19:30 to 22:00 h and at different times during the day following Heyer et al. (1994) making a total of 600 person-hours (400 during the night and 200 during the day). One site was also established in Kanga in November 2024 (Figure 1) and sampled by 4 people using time constrained searching for a total of 70 person-hours (40 during the night and 30 during the day).

Species identifications were made using field guides (Channing and Rödel, 2019), type descriptions and molecular barcoding (see details below). Representative specimens for each species were euthanised via an overdose of MS222 (buffered bath) or topical application of benzocaine, and a small sample of liver and thigh muscle were taken and preserved in 99.8% ethanol for DNA analysis. The specimens were then fixed in 10% formalin and later transferred to 70% ethanol for permanent storage at the Department of Zoology and Wildlife Conservation, University of Dar es Salaam, Tanzania for future verifications. For samples that were barcoded, total DNA was extracted using the Zymo Quick-DNA Miniprep Plus Kit/DNeasy blood and tissue kit. Amplification of 16S rRNA gene and sequencing followed standard protocols for amphibians (e.g. Lawson, 2010). The selection of 16S rRNA was made due to the wide use of this gene in most amphibian species barcoding projects which have generated large comparative database (e.g. Vences et al., 2005; Fouquet et al., 2007). Sequences were aligned (forward and reverse complements) and edited using Geneious prime 2022.1.1 software (https://www.geneious.com). The resulting sequences were then compared to the ones available in GenBank using nucleotide BLAST (https://blast.ncbi.nlm.nih.gov/Blast.cgi) to confirm species identity. Identification was considered a match when BLAST % identity was above 97.5 (Fouquet et al., 2007). Any sequences showing greater divergences were tentatively considered to represent cryptic taxa. We make distinctions between affinity, or confer to indicate varying levels of divergence from the closest known species (see Sigovini et al., 2016). Confer (cf.) was used in cases where the percentage identity was above 95% but less than 97.5%. Affinity (aff.) was used when the percentage identity was below 95% to the closest match. When the divergence was very high (below 93%) or multiple similarly matching species were recovered, lineages were identified only at the genus level and designated as ‘sp’. In this way, we aim to generate a better understanding of the true biodiversity of the amphibian fauna of the Ngurus. It is not the objective of this study to quantitatively delimit, or formally describe new species as such an integrative taxonomic approach would require more in-depth molecular, morphology and bioacoustic analysis, and ideally more detailed information on the natural history of organisms.

We also queried biodiversity databases and citizen science platforms (VertNet; www.vertnet.org and iNaturalist; www.inaturalist.org), and Google Scholar between 18 October 2024 and 15 Jan 2025 using a combination of ‘Amphibia/Anura/frog’ query terms and Nguru/Mkingu/Mkindo/Kanga/Magotwe (for literature search) or the reserves’ boundary (for VertNet and iNaturalist) to find any record of these taxa from the area. Any additional species records not recovered in our own surveys were vetted and included in the final species list.

This study has generated a comprehensive data combining morphology, molecular (mitochondrial) divergences, and ecological and geographic distributions to assign amphibian records to distinct operative taxonomic units/lineages for Nguru Mountains. This information, along with vouchers and sequence data examined is summarised in the Supplementary Table 2.

The literature and biodiversity databases search recovered a total of seven herpetological surveys conducted in Nguru Mountains between 1988 – 2012, registering 40 lineages of amphibians in total (Table 1). The seven additional surveys conducted for this study (in 2022 – 2024) registered 39 lineages, two of which were not recorded by the previous surveys (Nectophrynoides sp 3 and Phrynobatrachus acridoides), but failed to record Sclerophrys gutturalis, Phrynobatrachus natalensis and Chiromantis xerampelina which were recorded by previous surveys (Table 1; Supplementary Table 2). Except Nectophrynoides sp 3, all these species are usually associated with lowland and more open country habitats. The previous and current surveys combined registered 42 lineages of amphibians from 2 orders and 13 families (Table 1, Figures 2, 3, Supplementary Table 2). All surveys have been concentrated in Mkingu except one which considered both Mkingu and Kanga and another one which was conducted only in Kanga (Supplementary Table 1). There was no documented amphibian survey in Magotwe during this period. No single species was recorded throughout all surveys, but Arthroleptis affinis and Nectophrynoides cf. tornieri were recorded in 11 surveys, and Afrixalus cf. uluguruensis and Arthroleptis xenodactylus in ten surveys (Table 1). The rest were recorded in fewer than nine surveys (Table 1). Fifty-eight iNaturalist and 48 VertNet records were also retrieved, all from Mkingu, but did not add any new records.

We generated 82 mitochondrial DNA sequences representing 26 of the 42 identified lineages (Supplementary Tables 2, 3). Twenty-six more sequences were recovered from published sources, bringing the total number of lineages represented by genetic data to 37 (out of the 42). Only five species (Afrixalus stuhlmanni, Chiromantis xerampelina, Sclerophrys brauni, S. gutturalis and Xenopus victorianus) are lacking genetic data, but have at least one preserved specimen available (Supplementary Table 2). These species are lowland taxa and widely distributed across various habitats in Africa, except S. brauni which is restricted to the Eastern Arc Mountain forests of Tanzania, and A. stuhlmanni which is restricted to the Tanzanian coast. We designate at least fifteen (15) lineages as candidate species new to science, as they showed high genetic divergences from the closest currently known species (see Table 1, Supplementary Tables 2, 3). Twelve (12) of these species are identified in this study as either cf., aff. or sp. Four (4) more lineages (Arthroleptis affinis, Arthroleptis stenodactylus, Probreviceps macrodactylus, Nectophrynoides cf. tornieri) showed very high divergences even within the Nguru Mountains, especially from different elevation sites or forest patches, indicating at least two species from each of these lineages might be existing within Nguru (Supplementary File 2). Ten species had no genetic data in GenBank to compare with our sequences, despite all of these species having their type localities from Tanzania. Twenty-eight lineages are endemic to Tanzania, eight of them being only known from Nguru Mountains (see Table 1 and Supplementary Table 2). Taxonomically, the family Arthroleptidae is most represented (12 lineages), followed by Bufonidae (7), Brevicipitidae and Hyperoliidae (6 each) and Phrynobatrachidae (3). All other families are represented by 1 lineage each (Table 1; Supplementary Table 2). We also noted alarming cases of forest clearing throughout the sampled locations (Supplementary Figure 1). A number of endemic species, e.g. Callulina hanseni, Arthroleptis nguruensis, A. xenodactylus and A. kidogo, were observed in recently cleared patches.

The work outlined describes two years of dedicated field surveys in the Nguru Mountains and includes the first incorporation of 16S DNA barcoding together with literature and biodiversity database searches for this important biodiversity hotspot. We document 42 amphibian lineages, 8 of which are likely endemic to the Nguru Mountains. This study highlights the ability of DNA barcoding to alleviate the Linnean shortfall and the potential for species detection (Carné and Vieites, 2024). Among the 42 amphibian lineages, at least 15 are undescribed, cryptic taxa, potentially adding to Nguru endemics, highlighting the rich and unique amphibian diversity of this mountain block. Furthermore, we document a range extension of Arthroleptis xenodactylus, a species considered to be restricted to East Usambara Mountains (Channing and Rödel, 2019; Menegon et al., 2020; Frost, 2024). We find that, undescribed taxa are taxonomically skewed, with genera Nectophrynoides (likely at least 5 new species) and Arthroleptis (likely at least 2 new species) showing the highest cryptic diversity. Our work therefore highlights where future taxonomic work is required.

The refining of species identifications in our study and therefore the diversity of the Ngurus is largely due to taxonomic work carried out since 2008 (e.g. Poynton et al., 2008; Loader et al., 2010a, 2010b, 2015) and our use of 16S barcodes to detect cryptic species. Use of DNA barcoding has increased our understanding of herpetological diversity in other global hotspots tremendously (e.g. Nagy et al., 2012). Until superior high-throughput sequencing becomes more available in the region, mitochondrial barcoding should therefore be applied more widely across the Eastern Arc Mountains to better refine the herpetological records and enable more meaningful comparisons across locations. While species barcoding is a useful approach for documenting genetic diversity, sequencing more loci and integrative taxonomic approach is required to establish whether the single gene barcode differences amount to biologically meaningful species-level differences (Padial et al., 2010; Emerson, 2025). Integrative taxonomy together with more sampling across geographical ranges needs to be done for the candidate species highlighted in this study to accurately confirm their validity and allow them to be described. Nonetheless, there is a growing movement towards the consensus that biodiversity can and must be acknowledged even without the formal description of species (Dufresnes et al., 2023).

With this updated species list, the Nguru Mountains stand out for the large number of amphibian species, ranking the fourth most diverse block, after Udzungwa, Uluguru and East Usambara, and the second after Uluguru when correcting for forested area (see Lyakurwa et al., 2024). The number of Eastern Arc endemic species of the Nguru (28) exceed those of other EAM blocks except Udzungwa (31) and Uluguru (29) (Lyakurwa et al., 2024). However, it is important to point out that species numbers are contingent on survey efforts (Rovero et al., 2014) and there continue to be differences in survey effort across the region. The Nguru and other mountains in the EAM are constrained by both Linnaean and Wallacean shortfalls (see Lyakurwa et al., 2024), and the true amphibian diversity of each block is yet to be fully determined. The uneven patterns of cryptic taxonomic richness in our studies are also mirrored in other regions where molecular barcodes have been utilised. For example, studies from the Udzungwa Mountains reported large numbers of undescribed species, especially Arthroleptis and Nectophrynoides (Tonelli et al., 2017; Tonelli, 2021). Focused taxonomic study on these groups would clearly be beneficial and to establish a barcode database for thorough comparisons.

Sampling in this study expanded from areas previously surveyed (Emmrich, 1994; Menegon et al., 2008), but large areas still remain unexplored. More efforts to sample lowland areas that link EAM forests to Zanzibar-inhambane forests would be important. These habitats might contain unique coastal species as seen in other areas (e.g. Uluguru, Udzungwa and East Usambara). Lowland forests are rapidly disappearing (see Hamunyela et al., 2020 and Supplementary Figure 1) and an opportunity to observe their diversity might be lost given the current rate of deforestation. Magotwe Forest Reserve, formerly a large lowland forest in Nguru Mountains (Lovett and Pócs, 1993), is almost entirely deforested through settlement and farming, and for which no amphibians have been documented. Long term monitoring in disturbed sites is important to inform how endemic species respond to anthropogenic disturbances.

Diversity within and among closely related species across the elevation gradient or isolated patches in the Nguru mountains show some interesting patterns. Despite being separated only by ~10 km of non-forested valley, Kanga and Mkingu each harbour unique species. Some species that were sampled, e.g. Nectophrynoides cf. tornieri and Probreviceps macrodactylus, display high genetic divergence between these two locations, which might warrant them to be recognised as distinct species. This micro-endemism is also shown by a number of described species (Supplementary Table 2) and documented more widely across other taxa (e.g. plants; POWO, 2025). Relatively large genetic divergence was also observed between different elevations (e.g. Arthroleptis affinis and A. stenodactylus). Clearly taxonomic studies of Nguru amphibians are needed and until this work is completed the Linnean shortfall we outline in this study will impact directing appropriate conservation actions for amphibians across this biodiversity hotspot.