Philippa Kaur

Mueller, R. C. et al. An emerging view of the diversity, ecology, and function of Archaea in alkaline hydrothermal environments. FEMS Microbiol. Ecol. 97, fiaa246 (2020).
Google Scholar 
López-López, O., Cerdán, M.-E. & González-Siso, M.-I. Thermus thermophilus as a source of thermostable lipolytic enzymes. Microorganisms 3, 792–808 (2015).PubMed 
PubMed Central 

Google Scholar 
Sahay, H. et al. Hot springs of Indian Himalayas: Potential sources of microbial diversity and thermostable hydrolytic enzymes. 3 Biotech 7, 118 (2017).PubMed 
PubMed Central 

Google Scholar 
Patel, A. K., Singhania, R. R., Sim, S. J. & Pandey, A. Thermostable cellulases: Current status and perspectives. Bioresour Technol 279, 385–392 (2019).CAS 
PubMed 

Google Scholar 
Decastro, M.-E., Rodríguez-Belmonte, E. & González-Siso, M.-I. Metagenomics of thermophiles with a focus on discovery of novel thermozymes. Front. Microbiol. 7, 1521–1521 (2016).PubMed 
PubMed Central 

Google Scholar 
Meslé, M. M. et al. Isolation and characterization of lignocellulose-degrading geobacillus thermoleovorans from Yellowstone National Park. Appl. Environ. Microbiol. 88, e0095821 (2022).PubMed 

Google Scholar 
Verma, P., Yadav, A. N., Shukla, L., Saxena, A. K. & Suman, A. Hydrolytic enzymes production by thermotolerant Bacillus altitudinis IARI-MB-9 and Gulbenkiania mobilis IARI-MB-18 isolated from Manikaran hot springs. Int. J. Adv. Res. 3, 1241–1250 (2015).CAS 

Google Scholar 
Wu, B. et al. Microbial sulfur metabolism and environmental implications. Sci. Total Environ. 778, 146085 (2021).ADS 
CAS 
PubMed 

Google Scholar 
Lavrentyeva, E. V. et al. Bacterial diversity and functional activity of microbial communities in hot springs of the Baikal Rift Zone. Microbiology 87, 272–281 (2018).CAS 

Google Scholar 
Miller Scott, R., Strong Aaron, L., Jones Kenneth, L. & Ungerer Mark, C. Bar-Coded pyrosequencing reveals shared bacterial community properties along the temperature gradients of two alkaline hot springs in Yellowstone National Park. Appl. Environ. Microbiol. 75, 4565–4572 (2009).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Sharp, C. E. et al. Humboldt’s spa: Microbial diversity is controlled by temperature in geothermal environments. ISME J. 8, 1166–1174 (2014).CAS 
PubMed 
PubMed Central 

Google Scholar 
Stefanova, K. et al. Archaeal and bacterial diversity in two hot springs from geothermal regions in Bulgaria as demostrated by 16S rRNA and GH-57 genes. Int. Microbiol. 18, 217–223 (2015).CAS 
PubMed 

Google Scholar 
Hou, W. et al. A comprehensive census of microbial diversity in hot springs of Tengchong, Yunnan Province China using 16S rRNA gene pyrosequencing. PLoS ONE 8, e53350 (2013).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Sahm, K. et al. High abundance of heterotrophic prokaryotes in hydrothermal springs of the Azores as revealed by a network of 16S rRNA gene-based methods. Extremophiles 17, 649–662 (2013).CAS 
PubMed 

Google Scholar 
Purcell, D. et al. The effects of temperature, pH and sulphide on the community structure of hyperthermophilic streamers in hot springs of northern Thailand. FEMS Microbiol. Ecol. 60, 456–466 (2007).CAS 
PubMed 

Google Scholar 
Meyer-Dombard, D. R. & Amend, J. P. Geochemistry and microbial ecology in alkaline hot springs of Ambitle Island, Papua New Guinea. Extremophiles 18, 763–778 (2014).CAS 
PubMed 

Google Scholar 
de Leon, K. B., Gerlach, R., Peyton, B. M. & Fields, M. W. Archaeal and bacterial communities in three alkaline hot springs in Heart Lake Geyser Basin, Yellowstone National Park. Front. Microbiol. 4, 10 (2013).
Google Scholar 
Boomer, S. M., Noll, K. L., Geesey, G. G. & Dutton, B. E. Formation of multilayered photosynthetic biofilms in an alkaline thermal spring in Yellowstone National Park, Wyoming. Appl. Environ. Microbiol. 75, 2464–2475 (2009).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Wang, S. et al. Greater temporal changes of sediment microbial community than its waterborne counterpart in Tengchong hot springs, Yunnan Province, China. Sci. Rep. 4, 7479 (2014).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Sun, Y., Liu, Y., Pan, J., Wang, F. & Li, M. Perspectives on cultivation strategies of archaea. Microb. Ecol. 79, 770–784 (2020).PubMed 

Google Scholar 
Brock, T. D. Life at high temperatures. Science 158, 1012 (1967).ADS 
CAS 
PubMed 

Google Scholar 
Christiansen, R. L. The Quaternary and Pliocene Yellowstone Plateau volcanic field of Wyoming, Idaho, and Montana. Professional Paper (2001).Rowe, J. J., Fournier, R. & Morey, G. Chemical analysis of thermal waters in Yellowstone National Park, Wyoming, 1960–65. USGS https://doi.org/10.3133/b1303 (1973).Article 

Google Scholar 
Fournier, R., Thompson, M. J. & Hutchinson, R. A. The geochemistry of hot spring waters at Norris Geyser Basin, Yellowstone National Park. International symposium on water-rock interactions (1992).Podar, P. T., Yang, Z., Björnsdóttir, S. H. & Podar, M. Comparative analysis of microbial diversity across temperature gradients in hot springs from Yellowstone and Iceland. Front. Microbiol. 11, 1625 (2020).PubMed 
PubMed Central 

Google Scholar 
Pala, C. et al. Environmental drivers controlling bacterial and archaeal abundance in the sediments of a Mediterranean lagoon ecosystem. Curr. Microbiol. 75, 1147–1155 (2018).CAS 
PubMed 
PubMed Central 

Google Scholar 
Foyer, C. H., Noctor, G. & Hodges, M. Respiration and nitrogen assimilation: Targeting mitochondria-associated metabolism as a means to enhance nitrogen use efficiency. J. Exp. Bot. 62, 1467–1482 (2011).CAS 
PubMed 

Google Scholar 
Ershanovich, V. N. et al. Nitrogen assimilation enzymes in Bacillus subtilis mutants with hyperproduction of riboflavin. Mol. Gen. Mikrobiol. Virusol. 2005(3), 29–34 (2005).
Google Scholar 
Offre, P., Spang, A. & Schleper, C. Archaea in biogeochemical cycles. Annu Rev Microbiol 67, 437–457 (2013).CAS 
PubMed 

Google Scholar 
Cabello, P., Roldán, M. D. & Moreno-Vivián, C. Nitrate reduction and the nitrogen cycle in archaea. Microbiology 150, 3527–3546 (2004).CAS 
PubMed 

Google Scholar 
Graupner, M., Xu, H. & White, R. H. The pyrimidine nucleotide reductase step in riboflavin and F(420) biosynthesis in archaea proceeds by the eukaryotic route to riboflavin. J. Bacteriol. 184, 1952–1957 (2002).CAS 
PubMed 
PubMed Central 

Google Scholar 
Chernyh, N. A. et al. Dissimilatory sulfate reduction in the archaeon “Candidatus Vulcanisaeta moutnovskia” sheds light on the evolution of sulfur metabolism. Nat. Microbiol. 5, 1428–1438 (2020).CAS 
PubMed 

Google Scholar 
Castelle, C. J. & Banfield, J. F. Major new microbial groups expand diversity and alter our understanding of the tree of life. Cell 172, 1181–1197 (2018).CAS 
PubMed 

Google Scholar 
Williams, T. A. et al. Integrative modeling of gene and genome evolution roots the archaeal tree of life. Proc. Natl. Acad. Sci. U.S.A. 114, E4602–E4611 (2017).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Guy, L. & Ettema, T. J. G. The archaeal ‘TACK’ superphylum and the origin of eukaryotes. Trends Microbiol. 19, 580–587 (2011).CAS 
PubMed 

Google Scholar 
Wang, Y., Wegener, G., Hou, J., Wang, F. & Xiao, X. Expanding anaerobic alkane metabolism in the domain of Archaea. Nat. Microbiol. 4, 595–602 (2019).CAS 
PubMed 

Google Scholar 
Hedlund, B. P. et al. Uncultivated thermophiles: Current status and spotlight on ‘Aigarchaeota’. Curr. Opin. Microbiol. 25, 136–145 (2015).CAS 
PubMed 

Google Scholar 
Reichart, N. J. et al. Activity-based cell sorting reveals responses of uncultured archaea and bacteria to substrate amendment. ISME J. 14, 2851–2861 (2020).CAS 
PubMed 
PubMed Central 

Google Scholar 
Hua, Z.-S. et al. Genomic inference of the metabolism and evolution of the archaeal phylum Aigarchaeota. Nat. Commun. 9, 2832 (2018).ADS 
PubMed 
PubMed Central 

Google Scholar 
Beam, J. P. et al. Ecophysiology of an uncultivated lineage of Aigarchaeota from an oxic, hot spring filamentous “streamer” community. ISME J. 10, 210–224 (2016).CAS 
PubMed 

Google Scholar 
Gonsior, M. et al. Yellowstone hot springs are organic chemodiversity hot spots. Sci. Rep. 8, 14155 (2018).ADS 
PubMed 
PubMed Central 

Google Scholar 
Gibson, M. L. & Hinman, N. W. Mixing of hydrothermal water and groundwater near hot springs, Yellowstone National Park (USA): Hydrology and geochemistry. Hydrogeol. J. 21, 919–933 (2013).ADS 
CAS 

Google Scholar 
Campbell, K. M. et al. Sulfolobus islandicus meta-populations in Yellowstone National Park hot springs. Environ. Microbiol. 19, 2334–2347 (2017).PubMed 

Google Scholar 
Thiel, V. et al. The dark side of the mushroom spring microbial mat: Life in the shadow of chlorophototrophs. I. Microbial diversity based on 16S rRNA gene amplicons and metagenomic sequencing. Front. Microbiol. 7, 919 (2016).PubMed 
PubMed Central 

Google Scholar 
Caporaso, J. G. et al. Global patterns of 16S rRNA diversity at a depth of millions of sequences per sample. Proc. Natl. Acad. Sci. U.S.A. 108, 4516–4522 (2011).ADS 
CAS 
PubMed 

Google Scholar 
Parada, A. E., Needham, D. M. & Fuhrman, J. A. Every base matters: Assessing small subunit rRNA primers for marine microbiomes with mock communities, time series and global field samples. Environ. Microbiol. 18, 1403–1414 (2016).CAS 
PubMed 

Google Scholar 
Apprill, A., McNally, S., Parsons, R. & Weber, L. Minor revision to V4 region SSU rRNA 806R gene primer greatly increases detection of SAR11 bacterioplankton. Aquat. Microb. Ecol. 75, 129–137 (2015).
Google Scholar 
Thompson, L. R. et al. A communal catalogue reveals Earth’s multiscale microbial diversity. Nature 555, 457–463 (2017).ADS 

Google Scholar 
Eloe-Fadrosh, E. A., Ivanova, N. N., Woyke, T. & Kyrpides, N. C. Metagenomics uncovers gaps in amplicon-based detection of microbial diversity. Nat. Microbiol. 1, 15032 (2016).CAS 
PubMed 

Google Scholar 
Edgar, R. C. Search and clustering orders of magnitude faster than BLAST. Bioinformatics 26, 2460–2461 (2010).CAS 
PubMed 

Google Scholar 
Edgar, R. C. UNOISE2: Improved error-correction for Illumina 16S and ITS amplicon sequencing. BioRxiv https://doi.org/10.1101/081257 (2016).Article 

Google Scholar 
Murali, A., Bhargava, A. & Wright, E. S. IDTAXA: A novel approach for accurate taxonomic classification of microbiome sequences. Microbiome 6, 140 (2018).PubMed 
PubMed Central 

Google Scholar 
Parks, D. H. et al. A complete domain-to-species taxonomy for Bacteria and Archaea. Nat. Biotechnol. 38, 1079–1086 (2020).CAS 
PubMed 

Google Scholar 
Katoh, K. & Standley, D. M. MAFFT multiple sequence alignment software version 7: Improvements in performance and usability. Mol. Biol. Evol. 30, 772–780 (2013).CAS 
PubMed 
PubMed Central 

Google Scholar 
Stamatakis, A. RAxML version 8: A tool for phylogenetic analysis and post-analysis of large phylogenies. Bioinformatics 30, 1312–1313 (2014).CAS 
PubMed 
PubMed Central 

Google Scholar 
Letunic, I. & Bork, P. Interactive Tree Of Life (iTOL) v5: An online tool for phylogenetic tree display and annotation. Nucleic Acids Res. 49, W293–W296 (2021).CAS 
PubMed 
PubMed Central 

Google Scholar 
Matsen, F. A., Kodner, R. B. & Armbrust, E. V. pplacer: Linear time maximum-likelihood and Bayesian phylogenetic placement of sequences onto a fixed reference tree. BMC Bioinform. 11, 538 (2010).
Google Scholar 
Chong, J., Liu, P., Zhou, G. & Xia, J. Using MicrobiomeAnalyst for comprehensive statistical, functional, and meta-analysis of microbiome data. Nat. Protoc. 15, 799–821 (2020).CAS 
PubMed 

Google Scholar 
Chambers, M. C. et al. A cross-platform toolkit for mass spectrometry and proteomics. Nat. Biotechnol. 30, 918–920 (2012).CAS 
PubMed 
PubMed Central 

Google Scholar 
Pluskal, T., Castillo, S., Villar-Briones, A. & Oresic, M. MZmine 2: Modular framework for processing, visualizing, and analyzing mass spectrometry-based molecular profile data. BMC Bioinform. 11, 395 (2010).
Google Scholar 
Patiny, L. & Borel, A. ChemCalc: A building block for tomorrow’s chemical infrastructure. J. Chem. Inf. Model. 53, 1223–1228 (2013).CAS 
PubMed 

Google Scholar 
Chong, J., Wishart, D. S. & Xia, J. Using MetaboAnalyst 4.0 for comprehensive and integrative metabolomics data analysis. Curr. Protoc. Bioinform. 68, e86 (2019).
Google Scholar 
Liu, G., Lee, D. P., Schmidt, E. & Prasad, G. L. Pathway analysis of global metabolomic profiles identified enrichment of caffeine, energy, and arginine metabolism in smokers but not moist snuff consumers. Bioinform. Biol. Insights 13, 1177932219882961–1177932219882961 (2019).PubMed 
PubMed Central 

Google Scholar 
Xia, J. & Wishart, D. S. MetPA: A web-based metabolomics tool for pathway analysis and visualization. Bioinformatics 26, 2342–2344 (2010).CAS 
PubMed 

Google Scholar 
Huber, W. et al. Orchestrating high-throughput genomic analysis with Bioconductor. Nat. Methods 12, 115–121 (2015).CAS 
PubMed 
PubMed Central 

Google Scholar 
Rohart, F., Gautier, B., Singh, A. & Lé Cao, K.-A. mixOmics: An R package for ’omics feature selection and multiple data integration. PLoS Comput. Biol. 13, e1005752–e1005752 (2017).ADS 
PubMed 
PubMed Central 

Google Scholar  More

The richness of Melanesian FrogsApproximately 7.2% (534 out of 7404) of Earth’s recognised frog species occur in Melanesia, a region comprising < 0.7% of the world’s land area. Frog richness in Melanesia, and especially on New Guinea and nearby land-bridge islands (471 species), is higher than in any other tropical insular region (Fig. 1a). New Melanesian frog species have been described at an average rate of nearly 13 species/year since 2000, and the recognised frog fauna has grown by > 50% in that timeframe (Fig. 1b). The authorship of new species has been concentrated, with six authors featuring on 20 or more descriptions since 2000, and one or more of these six authors on every species description since 2000. A small number of species descriptions has included genetic data (31 species), although a higher number of Melanesian frog species have at least one sequence available on GenBank (~38%, or approximately 200 species). This taxonomic work has revealed or emphasised many evolutionary novelties (Fig. 2): multiple apparently independent derivations of extremely miniaturised vertebrates22,23,24, including some of the world’s smallest known tetrapods23,25,26; multiple derivations of complex parental care in different genera27,28; frequent evolutionary shifts between terrestrial, arboreal and scansorial lifecycles22,29; the most extreme sexual size dimorphism yet documented in anurans30; drastic ontogenetic colour change31; a radiation of canopy-dwelling treefrogs32 that show extensive finger webbing and parachuting behaviour convergent with unrelated frog lineages in Asia and the Neotropics; and treefrogs with erectile noses33,34. Taxonomic work has also elucidated novel concentrations of range-restricted endemic taxa, especially in the Milne Bay Region at the far eastern edge of New Guinea21.Fig. 1: Temporal trends in the documentation of the Melanesian frog fauna.a Species accumulation curves for species-rich ( >100 species) insular frog biotas (Species lists from AmphibiaWeb as of 1 October 2021). Separate accumulation curves are given for the entire fauna of Melanesia (including New Guinea), and the fauna of New Guinea and nearby predominantly land-bridge islands. b Species accumulation curve for frogs within Melanesia. Bar at end indicates predicted number of species in each major family based on known, but as yet undescribed candidate species.Full size imageFig. 2: Melanesian frog species described within the last 15 years illustrating the ecological and morphological diversity of the fauna.a Paedophryne titan and b Choerophryne gracilirostris – examples of lineages that have undergone convergent minaturisation; c Choerophryne alpestris – a fossorial species within a largely scansorial lineage; d Xenorhina macrodisca – scansorial species within a largely fossorial lineage; e Cornufer custos and f Oreophryne oviprotector – independent derivations of complex parental care; g Litoria pallidofemora – extensive digital webbing for parachuting; and h Litoria pinocchio – sexually dimorphic and erectile rostral spikes. Photographs F. Kraus (a), S. Richards (b–g), and courtesy of T. Laman (h).Full size imageFrog species richness in Melanesia is highly concentrated into just three families, with Pelodryadidae (137 recognised species, estimated ~200) and especially Microhylidae (317 species, estimated >400) dominating. Melanesian Pelodryadidae are phylogenetically interdigitated with relatives in Australia, suggesting multiple dispersals between the two regions35. In contrast, ancestors of the direct-developing microhylids colonised Melanesia from Asia via trans-marine dispersal likely only once36, radiated across open ecological niches37, and are now the most species-rich insular radiation of frogs in the world. The third major family comprises an ecologically diverse radiation of the direct-developing Ceratobatrachidae (57 species, estimated 66) largely associated with island-arc terranes of East Melanesia and the Philippines, indicating a long history of insular diversification and trans-marine dispersal38. The predominance of direct-developing frogs in Melanesia (~70% of species) mirrors insular faunas in Madagascar (~34%), Sri Lanka (~67%) and the Greater Antilles (~87%). The other four frog families in Melanesia are all relatively species poor (2, 3, 4, and 13 species) (Fig. 1a), centred in New Guinea, and include lineages originating in Asia (Ranidae, Dicroglossidae) or Australia (Myobatrachidae, Limnodynastidae).The described diversity of Melanesian amphibian species remains an underestimate. Survey work and investigation of museum collections by the co-authors identified ~190 additional candidate species distributed across 16 different genera, mostly from Papua New Guinea, suggesting a total richness of over 700 frog species (Fig. 1a, Supplementary Table 1). This estimated percentage of undescribed diversity (~25%) mirrors estimates for the New Guinean flora (~18–22%)7. The majority of candidate species are concentrated in the two most diverse families (Microhylidae and Pelodryadidae), although genetic, morphological, and acoustic evidence indicate the diversity of Melanesian Ranidae is also underestimated (S. Richards and F. Kraus pers. obs.). Most material documenting candidate species has been collected in the last 20 years, and the vast majority is from Papua New Guinea (Supplementary Fig. 1). There is some suggestion of a slowing in the rate of candidate species discovery in the last decade (Supplementary Fig. 2); however, several of the most active field workers in this region have ceased survey work in recent years, which likely accounts for much of this decline. The pervasiveness of complexes of morphologically and/or acoustically cryptic taxa is poorly understood; survey work continues to reveal novelties, and large areas of the region remain unsurveyed or undersampled. In particular, comparisons of area-to-diversity ratios between the better-known eastern portion of New Guinea (Papua New Guinea) with the poorly surveyed western (Indonesian) portion of the island further suggest that, even with candidate species included, diversity in the latter region may be underestimated by as much as 50% (Supplementary Methods and Results, Supplementary Table 2). These trends and patterns all indicate that ~ 700 species is a very conservative minimum estimate of total diversity and support analyses in other taxa showing Melanesia remains a hotspot of unrecognised diversity39,40.Endemism and distributional patternsThe Melanesian frog fauna is highly endemic (97.2%), with tiny proportions of species shared with Australia (2.4%) or with islands farther west in Indonesia (0.6%), indicating that Australia and Melanesia are discrete centres of frog diversification, despite periodic connection via land bridges through the late Tertiary41. The vast majority of Melanesian frog species (471) occur on New Guinea and nearby land-bridge islands (Raja Ampats, Japen and the Milne Bay islands). In comparison, the frog fauna of the much smaller region of Maluku is depauperate (16 species, of which nine are endemic) but also almost certainly underestimated (e.g., there are no Microhylidae recorded from Buru). Most taxa from Maluku are congeneric (and several conspecific) with lineages centred on New Guinea, supporting the biogeographic clustering of Maluku’s amphibians with the main island of New Guinea. In contrast, the frog fauna of East Melanesia is more diverse and highly endemic and dominated by an ecologically diverse radiation of a different family (Ceratobatrachidae) with only four (all pelodryadid treefrogs) out of 56 species shared with nearby New Guinea. East Melanesia and New Guinea appear to be discrete and long-isolated centres of diversification, as expected from their independent geological histories42.Melanesia spans five countries, and this has possibly to some degree masked the exceptional species diversity of the overall region. Papua New Guinea has the highest number of species (398) and endemic species (318). This likely reflects some combination of its slightly larger area (when islands to the north are included), more diverse geological origins, and greater inventory work than seen in neighbouring regions of Indonesia7. Papua, West Papua and Maluku (Indonesia) have many fewer documented species (197), of which a majority (134) is endemic. The boundary between Papua and Papua New Guinea is visible in species-richness maps (Fig. 3a), with lower diversity to the west, indicating that the distribution and diversity of frogs in Indonesia remain less documented in science. The frog faunas of the Solomon Islands (21 species) and Fiji (two species) are more depauperate but include a significant endemic or near-endemic component, whereas the geographically intervening islands of Vanuatu support no native frogs.Fig. 3: Frog species richness in Melanesia based on IUCN distributional maps for all species described by 2019.a All species; b Ceratobatrachidae; c Microhylidae; d Pelodryadidae. Areas of highest estimated diversity correspond to mountain ranges in central and northern New Guinea. The boundaries between Maluku, New Guinea and East Melanesia are indicated. Fiji has only two frog species and is geographically distant from other areas of Melanesia inhabited by frogs and is not visble on this map.Full size imageBased on distribution maps generated for all species recognised by 31 August 2019, the highest regional alpha diversity of frogs occurs along the Central Cordillera of New Guinea (especially in Papua New Guinea) and around the higher mountain ranges along the north coast of Papua New Guinea (Fig. 3a). These centres of diversity correlate with extensive areas of hill and montane forest and broadly correspond with elevational species-richness patterns for mammals and birds in Melanesia43 and for many other taxa elsewhere in the tropics44,45. Large areas of montane forest with lower species richness along the northern versant of the Central Cordillera in Papua New Guinea and in mountain ranges across Papua certainly reflect inadequate sampling. The ceratobatrachid-dominated frog fauna of East Melanesia is richest in Bougainville (Fig. 3d), with attenuating richness towards the west and especially to the east. The two most speciose families both show alpha diversity peaks in mountainous areas of central New Guinea (Fig. 3c–d). In contrast, microhylids are largely absent from the seasonally dry woodlands of the Trans-Fly region in southern New Guinea and exhibit high diversity in northern New Guinea, whereas pelodryadids are much more speciose in the lowlands of southern New Guinea than northern New Guinea. These broad trends may have both ecological (sensitivity of direct-developing microhylids to dry conditions) and historical (Australia as a centre of origin for savanna-adapted Pelodrydidae) underpinnings.The historical and contemporary factors underpinning high frog species diversity in New Guinea remain largely unstudied, especially when compared to other species-rich insular amphibian faunas such as Madagascar46 or the Greater Antilles47. When compared to some areas of the Neotropics, alpha and beta diversities of frogs in lowland forests in the basins of the Sepik and Ramu rivers in New Guinea are unremarkable48. However, the Milne Bay Region has exceptionally high levels of endemism21, so species turnover will be higher in this area. Extent-of-occurrence estimates derived from IUCN maps indicate that direct-developing microhylids have smaller mean and median range sizes than all other families of frogs in Melanesia (Supplementary Table 3). Microhylidae also dominate anuran species diversity in Milne Bay21 and many mountain areas where standing water is very limited49. These data suggest that, as with some areas in the Neotropics50, high beta diversity in lineages with direct development is a key factor underpinning amphibian megadiversity in Melanesia. To address these questions further, synthetic analyses are required to better quantify the extent to which regional megadiversity in Melanesia reflects high community diversity versus species turnover, how elevation and insularity moderates these two parameters, and to what extent emergent patterns may differ from diverse frog communities in other regions such as the Neotropics.The conservation status of Melanesian FrogsThe frog fauna of Melanesia is currently less threatened but more Data Deficient than other comparable insular regions (Fig. 4a). The vast majority of Melanesian frogs are categorised as Least Concern (68%) or Data Deficient (24%). Thirty-one species (6%, or 8% if Data Deficient taxa are excluded) are threatened (Critically Endangered, Endangered, Vulnerable) (Supplementary Table 4), and eight species are considered Near Threatened. No species are assessed as Extinct or Extinct in the Wild. Since the first Global Amphibian Assessment in 2004, the number of Melanesian frog species has grown by 44%, and nearly 60% of the 31 Melanesian frog species now considered threatened were described after 2004 (Fig. 1a). Only one change in status between 2004 and 2019 was considered genuine (Cophixalus sphagnicola), due to the emerging threat of a newly opened mine. All other status changes (for 116 taxa) reflect better information on distribution or changed assessment protocols (Supplementary Table 5). Applying stricter criteria for use of the Data Deficient category in the 2019 IUCN assessment reduced the number of Data Deficient species when compared to 2004 (125 versus 197), but Melanesia still has a higher percentage of Data Deficient taxa than other species-rich tropical insular faunas (Fig. 4a).Fig. 4: The conservation status of Melanesian frogs.a Comparison of number of species in each IUCN threat category across Melanesia, other diverse insular regions, and the nearby continent of Australia. Melanesia has a proportionally low number of threatened taxa but high number of Data Deficient taxa (EX Extinct, CR Critically Endangered, EN Endangered, VU Vulnerable, NT Near Threatened, DD Data Deficient, LC Least Concern, NE Not Evaluated); b Slopes around Mt Simpson, Milne Bay Province, a hotspot of threatened frog diversity due to forest loss through conversion to anthropogenic grasslands; c Choerophryne sanguinopicta from Mt Simpson (Critically Endangered); d Oreophryne ezra from Rossel Island (Critically Endangered) and; e Cornufer citrinospilus from New Britain (Vulnerable). Photographs F. Kraus (b–d), S. Richards (e).Full size imageAll Critically Endangered and Endangered—and most of the Vulnerable—species were listed because of their small extent of occurrence and on-going decline in habitat area and/or quality (criteria B1ab(iii)) (Supplementary Table 6). The key threatening processes were typically forest disturbance or loss due to conversion to plantations or gardens, repeated burning, or mining (Fig. 4b–c). Only two insular species with very localised montane distributions were considered threatened by climatic disturbance and/or climate change alone (Cornufer citrinospilus and Oreophryne ezra) (Fig. 4d–e). No species were currently declining from pathogens, and in particular Batrachochytrium dendrobatidis (Bd), which remains undetected in Melaneisa51. However, the introduction and establishment of Bd has been identified as a severe threat for well over one hundred taxa52, especially for montane pelodryadid treefrogs, a group that has been devastated by this disease in parts of Australia.Although much of New Guinea has historically been considered a ‘wilderness area’ with comparatively little human impact53, the distributions of threatened taxa also highlight areas of conservation concern wherein range-restricted (often single-island endemic) taxa overlap with extensive and increasing anthropogenic impacts (Fig. 5a–b). Nearly half the species identified as threatened (13) are restricted to a recently delineated dramatic centre of herpetofaunal endemism in the Milne Bay Region at the eastern tip of Papua New Guinea21. Three clusters of small-range endemics in this region (all documented in the last two decades) present immediate conservation issues. The first is Mount Simpson, where six microhylids (four named, two awaiting description) with highly restricted ranges are threatened by habitat loss, especially repeated burning and associated conversion of forest to grassland (Fig. 4b). The second is Woodlark Island, where the status of seven endemic microhylids (six named, one undescribed) is likely to worsen rapidly if current, approved proposals to convert large areas of primary forest to oil-palm plantation and/or gold mines proceed21. Finally, Misima Island is home to four endemic microhylids (two considered threatened) with ranges that overlap areas disturbed by mining and forest loss21. Other regions with multiple overlapping threatened taxa are the Adelbert Mountains in Morobe Province (two species), New Britain (two lowland species and one highland species), and Greater Bukida in the Solomon Islands (three lowland species). These clusters of narrow-range taxa highlight important—and in most cases largely overlooked—conservation priorities for Melanesian frogs (and likely other taxa as well21,54). The high percentage of Data Deficient species and low level of survey effort in many areas (especially Papua and West Papua Provinces, Indonesia) also raise the possibility that other threatened hotspots remain overlooked. One area of particular concern may be the island of Biak in Indonesia, which has lost much of its primary vegetation but is home to at least three endemic frogs (one Data Deficient, two Least Concern).Fig. 5: The distribution of threatened frogs in Melanesia.a The estimated distribution of all 31 Melanesian frog species considered Critically Endangered, Endangered or Vulnerable at the end of 2019. Distributional areas are not colour coded by the number of threatened taxa. b Close up of the Milne Bay endemism hotspot. Distributional areas are colour coded by number of taxa, with darker tones indicating more taxa. In both a and b upland areas or islands where the distributions of two or more threatened species overlap are labelled and the number of threatened taxa are indicated in parentheses. Background maps uses the Shuttle Radar Topography Mission (SRTM) 30-meter digital elevation model, accessed from USGS Earth Explorer (https://earthexplorer.usgs.gov/).Full size imageUnderstanding and conserving a megadiverse biotaThe Melanesian flora and frog fauna are both now shown to be megadiverse and highly endemic, yet both also remain poorly known with large areas under-surveyed. An updated comprehensive assessment of threats and taxonomic trends across the frog fauna presented here further highlights that the biota of Melanesia remains relatively intact and less threatened when compared to other biodiverse insular regions. However, a large proportion of the fauna remains Data Deficient or undescribed, and key hotspots of endemism have been overlooked and are increasingly threatened. In both plants and anurans much scientific knowledge of Melanesia’s biota has also been contributed by a relatively small number of productive, but later-career researchers based outside of Melanesia7.Further documenting and conserving the exceptional diversity of Melanesia presents a suite of challenges and opportunities. Recommendations to enable improved documentation of plant megadiversity in Melanesia7 centre around training, capacity-building and support for taxonomy in Melanesia and globally, improving access to specimen collections and diagnostic resources, and ongoing support for survey and collecting within Melanesia. These recommendations apply equally to amphibians. However, addressing these challenges is tempered by the limited career opportunities available to ecologists and taxonomists (both in developed, but especially in developing countries), the variable quality of scientific infrastructure that exists across the region, and the high cost of doing fieldwork in remote areas with limited logistical infrastructure. In the context of these challenges, we hereby focus on suggesting some short-term key priorities and opportunities to build capacity for understanding and conserving frog biodiversity in Melanesia.First, over the last twenty years opportunities to employ Melanesian nationals in survey, monitoring and outreach work have been (and will continue to be) generated predominantly by NGOs, universities and large-scale extractive projects, for example through recent work in the gas fields of the Papua New Guinea Highlands49. While there are diverse perspectives on extractive industries, monitoring and survey work associated with large development projects are a key source of funds to provide training to enable Melanesians to undertake biodiversity work within the region. A key driver of this is strong environmental legislation required by some governments and major lending agencies, in particular the International Finance Corporation under Performance Standard 655. These requirements need to be maintained, enforced and, where possible, exceeded.To further support fieldwork by national scientists there is a need for more readily accessible identification resources for Melanesian researchers, land-owners and managers. An up-to-date comprehensive identification guide to the frog fauna of the whole region would assist and promote taxonomic, ecological and conservation research. However, for many Melanesians, small, regionally focused guides are more usable. These have already been produced for several areas (Supplementary References), providing a model that can be updated and transferred to other regions. Mobile phones are widely used throughout Melanesia, so app- and online-based identification resources may become increasingly accessible. Smartphone-friendly citizen science platforms like iNaturalist56 or even Facebook groups57 also provide potentially powerful resources through which locally collected data can be captured, vetted and disseminated, although their use is currently limited in Melanesia due to patchy internet coverage in many areas. Working with and supporting people from Melanesia to explore and increase the use of these resources could help to ensure longer-term preservation and accessibility of species records and associated data.The latest IUCN assessment for Melanesian frogs also highlights how taxonomic and conservation knowledge is accumulating rapidly. The key geographic areas of threat identified in our study were largely invisible to assessments made less than two decades ago (in 2004) both because the relevant taxonomic work had not been done, and because the situation in Melanesia is changing rapidly. To keep track of these rapid changes it is critical for workers in the region to work together to synthesise and collate new taxonomic, distributional and conservation data. Indeed, since the 2019 IUCN assessment over 20 additional species of Melanesian frogs have been described, and their conservation status should be assessed as a matter of urgency. Preliminary conservation assessments against IUCN criteria are increasingly being included in descriptions, and this trend should be supported and encouraged. More Melanesian nationals need to be involved in conservation assessment processes. Updated comprehensive conservation assessments of other vertebrate groups will also identify complementarity of conservation priorities among taxa in the Melanesian region.Patterns of distribution and threat suggest some geographic priority areas for documenting the diversity of amphibians (and potentially other low-vagility taxa) in Melanesia. First, work in eastern New Guinea has allowed the delineation of geographically localised clusters of threatened taxa that have until now gone unnoticed, perhaps in part because of the designation of much of Melanesia as a sparsely populated and comparatively undisturbed ‘wilderness’ area21. Most threatened frog taxa in these regions are associated with small islands or isolated ‘sky island’ mountains. The degree to which other taxa show endemism in these areas is poorly known. The biotas of potentially comparable islands in Indonesia such as the Raja Ampat Islands, Geelvink Bay and southern Maluku, also remain poorly known, suggesting additional priority areas for survey, taxonomic investigation and conservation assessment. Second, mid-elevation areas show highest alpha diversity, but large areas of this habitat, especially along the northern slopes of the Central Cordillera, remain poorly surveyed. The frog pathogen Bd has devastated montane communities of two Australian frog families that also occur in montane New Guinea (Myobatrachidae and Pelodryadidae)52. In the unfortunate event that Bd colonised New Guinea a wave of rapid declines and extinctions would likely follow52, so a strong baseline of information on montane species diversity, distributions and population status is critical for detecting these impacts. More

Potts, S. G. et al. (eds.). IPBES: The Assessment Report of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services on Pollinators, Pollination and Food Production (Secretariat of the Intergovernmental Science-Policy Platform on Biodiversity and Ecosystem Services, Bonn, Germany) (2016).Potts, S. G. et al. Global pollinator declines: Trends, impacts and drivers. Trends Ecol. Evol. 25, 345–353 (2010).PubMed 

Google Scholar 
Goulson, D., Nicholls, E., Botías, C. & Rotheray, E. L. Bee declines driven by combined stress from parasites, pesticides, and lack of flowers. Science 347, 1255957 (2015).PubMed 

Google Scholar 
Majewska, A. A. & Altizer, S. Planting gardens to support insect pollinators. Conserv. Biol. 34, 15–25 (2020).PubMed 

Google Scholar 
Image, M. et al. Does agri-environment scheme participation in England increase pollinator populations and crop pollination services?. Agric. Ecosyst. Environ. 325, 107755 (2022).
Google Scholar 
Vaissière, B., Freitas, B. M. & Gemmill-Herren, B. Protocol to Detect and Assess Pollination Deficits in Crops: A Handbook for Its Use (FAO, 2011).
Google Scholar 
Archer, C. R., Pirk, C. W. W., Carvalheiro, L. G. & Nicolson, S. W. Economic and ecological implications of geographic bias in pollinator ecology in the light of pollinator declines. Oikos 123, 401–407 (2014).
Google Scholar 
M’Gonigle, L. K., Ponisio, L. C., Cutler, K. & Kremen, C. Habitat restoration promotes pollinator persistence and colonization in intensively managed agriculture. Ecol. Appl. 25, 1557–1565 (2015).PubMed 

Google Scholar 
Garbuzov, M. & Ratnieks, F. L. W. Listmania: The strengths and weaknesses of lists of garden plants to help pollinators. Bioscience 64, 1019–1026 (2014).
Google Scholar 
Garbuzov, M. & Ratnieks, F. L. W. Quantifying variation among garden plants in attractiveness to bees and other flower-visiting insects. Funct. Ecol. 28, 364–374 (2014).
Google Scholar 
Garbuzov, M., Alton, K. & Ratnieks, F. L. W. Most ornamental plants on sale in garden centres are unattractive to flower-visiting insects. PeerJ 5, e3066 (2017).PubMed 
PubMed Central 

Google Scholar 
Nichols, R. N., Goulson, D. & Holland, J. M. The best wildflowers for wild bees. J. Insect Conserv. 23, 819–830 (2019).
Google Scholar 
Harmon-Threatt, A. Influence of nesting characteristics on health of wild bee communities. Annu. Rev. Entomol. 65, 39–56 (2020).CAS 
PubMed 

Google Scholar 
Requier, F. & Leonhardt, S. D. Beyond flowers: Including non-floral resources in bee conservation schemes. J. Insect Conserv. 24, 5–16 (2020).
Google Scholar 
Sinu, P. A. & Bronstein, J. L. Foraging preferences of leafcutter bees in three contrasting geographical zones. Divers. Distrib. 24, 621–628 (2018).
Google Scholar 
Cecala, J. M. & Rankin, E. E. Pollinators and plant nurseries: How irrigation and pesticide treatment of native ornamental plants impact solitary bees. Proc. R. Soc. B Biol. Sci. 288, 20211287 (2021).
CAS 

Google Scholar 
Gonzalez, V. H., Gustafson, G. T. & Engel, M. S. Morphological phylogeny of Megachilini and the evolution of leaf-cutter behavior in bees (Hymenoptera: Megachilidae). J. Melittology 85, 1–123 (2019).
Google Scholar 
Kambli̇, S. S. et al. M. S. Aiswarya, K. Manoj, S. Varma, G. Asha, T. P. Rajesh, P. A. Sinu, Leaf foraging sources of leafcutter bees in a tropical environment: Implications for conservation. Apidologie 48, 473–482 (2017).Ascher, J. S. & Pickering, J. Discover Life Bee Species Guide and World Checklist (Hymenoptera: Apoidea: Anthophila) (2019).McCabe, L. M., Aslan, S. E. & Cobb, N. S. Decreased bee emergence along an elevation gradient: implications for climate change revealed by a transplant experiment. Ecology 103, e03598 (2021).PubMed 

Google Scholar 
Pitts-Singer, T. L. & Cane, J. H. The Alfalfa leafcutting bee, Megachile rotundata: The worlds most intensively managed solitary bee. Annu. Rev. Entomol. 56, 221–237 (2011).CAS 
PubMed 

Google Scholar 
MacIvor, J. S. & Packer, L. “Bee hotels” as tools for native pollinator conservation: A premature verdict?. PLoS ONE 10, e0122126 (2015).PubMed 
PubMed Central 

Google Scholar 
Maclvor, J. S. DNA barcoding to identify leaf preference of leafcutting bees. R. Soc. Open Sci. 3, 150623 (2016).ADS 

Google Scholar 
Wissemann, V. & Ritz, C. M. The genus Rosa (Rosoideae, Rosaceae) revisited: Molecular analysis of nrITS-1 and atpB-rbcL intergenic spacer (IGS) versus conventional taxonomy. Botanical J. Linn. Soc. 147, 275–290 (2005).
Google Scholar 
Wang, G. Study on the history of Chinese roses from ancient works and images. Acta Hort. 751, 347–356 (2007).
Google Scholar 
Nybom, H. & Werlemark, G. Realizing the potential of health-promoting rosehips from dogroses (Rosa sect. Caninae). Curr. Bioact. Compd. 13, 3–17 (2016).
Google Scholar 
Chang, Y. Z., Chen, H. M. & Qi, R. S. Ornamental pest—studies on leafcutting bees Megachile subtranquilla Yasumatsu. Acta Agriculturae Universitatis Pekinensis 15, 208–213 (1989).
Google Scholar 
Stroom, K., Fetzer, J. & Krischik, V. Insect Pests of Roses. 1–12 (Minnesota Extension Service, University of Minnesota, 1997).Knox, G. W., Paret, M. & Mizell, R. F. III. Pests of roses in Florida (2008).Hayward, A. et al. The leafcutter bee, Megachile rotundata, is more sensitive to N-cyanoamidine neonicotinoid and butenolide insecticides than other managed bees. Nat. Ecol. Evol. 3, 1521–1524 (2019).PubMed 

Google Scholar 
Fox, J. et al. Package ‘car’, Vol. 16, (R Foundation for Statistical Computing, 2012).K. Barton, Package Multi-Model Inference (MuMIn). https://cran.r-project.org/web/packages/MuMIn/MuMIn.pdf (2013).Hartig, F. & Hartig M. F. Package ‘DHARMa’:R package (2017).R Core Team. R: A Language and Environment for Statistical Computing https://www.R-project.org/ (R Foundation for Statistical Computing, 2021).Boff, S., Raizer, J. & Lupi, D. Environmental display can buffer the effect of pesticides on solitary bees. Insects. 11, 1–15 (2020).
Google Scholar 
Cameron, S. A. et al. Patterns of widespread decline in North American bumble bees. Proc. Natl. Acad. Sci. USA. 108, 662–667 (2011).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Cameron, S. A. & Sadd, B. M. Global trends in bumble bee health. Annu. Rev. Entomol. 65, 209–232 (2020).CAS 
PubMed 

Google Scholar 
Kopit, A. M. & Pitts-Singer, T. L. Routes of pesticide exposure in solitary, cavity-nesting bees. Environ. Entomol. 47, 499–510 (2018).CAS 

Google Scholar 
Pitts-Singer, T. L. & Barbour, J. D. Effects of residual novaluron on reproduction in alfalfa leafcutting bees, Megachile rotundata F. (Megachilidae). Pest Manag. Sci. 73, 153–159 (2017).CAS 
PubMed 

Google Scholar 
McKinney, M. L. Urbanization, biodiversity, and conservation. Bioscience 52, 883–890 (2002).
Google Scholar 
Baldock, K. C. R. et al. A systems approach reveals urban pollinator hotspots and conservation opportunities. Nat. Ecol. Evol. 3, 363–373 (2019).PubMed 
PubMed Central 

Google Scholar 
Choate, B. A., Hickman, P. L. & Moretti, E. A. Wild bee species abundance and richness across an urban–rural gradient. J. Insect Conserv. 22, 391–403 (2018).
Google Scholar 
Theodorou, P. et al. Pollination services enhanced with urbanization despite increasing pollinator parasitism. Proc. R. Soc. B Biol. Sci. 283, 20160561 (2016).
Google Scholar 
Theodorou, P. et al. Urban areas as hotspots for bees and pollination but not a panacea for all insects. Nat. Commun. 11, 1–13 (2020).
Google Scholar 
Rocha-Filho, L. C., Martins, A. C. & Marchi, P. Notes on a nest of Megachile (Moureapis) apicipennis Schrottky (Megachilidae) constructed in an abandoned gallery of Xylocopa frontalis (Olivier) (Apidae). Sociobiology 64, 442–450 (2017).
Google Scholar 
Sheffield, C. S. Unusual nesting behavior in Megachile (Eutricharaea) rotundata (Hymenoptera: Megachilidae). J. Melittol. 69, 1–6 (2017).
Google Scholar 
Krischik, V., Rogers, M., Gupta, G. & Varshney, A. Soil-applied imidacloprid translocates to ornamental flowers and reduces survival of adult Coleomegilla maculata, Harmonia axyridis, and Hippodamia convergens lady beetles, and larval Danaus plexippus and Vanessa cardui butterflies. PLoS ONE 10, e0119133 (2015).PubMed 
PubMed Central 

Google Scholar 
Senapathi, D. et al. The impact of over 80 years of land cover changes on bee and wasp pollinator communities in England. Proc. R. Soc. B Biol. Sci. 282, 20150294 (2015).
Google Scholar 
Potts, S. G. et al. Role of nesting resources in organising diverse bee communities in a Mediterranean landscape. Ecol. Entomol. 30, 78–85 (2005).
Google Scholar 
Acar, C., Acar, H. & Eroǧlu, E. Evaluation of ornamental plant resources to urban biodiversity and cultural changing: A case study of residential landscapes in Trabzon city (Turkey). Build. Environ. 42, 218–229 (2007).
Google Scholar 
Wang, H. F., Qureshi, S., Knapp, S., Friedman, C. R. & Hubacek, K. A basic assessment of residential plant diversity and its ecosystem services and disservices in Beijing, China. Appl. Geogr. 64, 121–131 (2015).
Google Scholar 
Pergl, J. et al. Dark side of the fence: ornamental plants as a source of wildgrowing flora in the Czech Republic. Preslia 88, 163–184 (2016).
Google Scholar 
Avolio, M. et al. Urban plant diversity in Los Angeles, California: Species and functional type turnover in cultivated landscapes. Plants People Planet. 2, 144–156 (2020).
Google Scholar 
Orr, M. C. et al. Global patterns and drivers of bee distribution. Curr. Biol. 31, 451–458 (2021).CAS 
PubMed 

Google Scholar 
Sinu, P. A., Kuriakose, G. & Shivanna, K. R. Is the bumblebee (Bombus haemorrhoidalis) the only pollinator of large cardamom in central Himalayas, India?. Apidologie 42, 690–695 (2012).
Google Scholar 
Veereshkumar, V. V. & Gupta, A. Parasitisation of leaf-cutter bees (Megachilidae: Apoidea) by Melittobia. Entomon 40, 105–112 (2015).
Google Scholar 
Cecala, J. M. & Wilson Rankin, E. E. Petals and leaves: Quantifying the use of nest building materials by the world’s most valuable solitary bee. Ecology 103, e03584 (2021).PubMed 

Google Scholar 
Soh, E. J. Y., Soh, Z. W. W., Ascher, J. S. & Tan, H. T. W. Diversity of plants with leaves cut by bees of the genus Megachile in Singapore. Nat. Singap. 12, 63–74 (2019).
Google Scholar 
MacIvor, J. S. & Moore, A. E. Bees collect polyurethane and polyethylene plastics as novel nest materials. Ecosphere 4, 155 (2013).
Google Scholar 
Allasino, M. L., Marrero, H. J., Dorado, J. & Torretta, J. P. Scientific note: First global report of a bee nest built only with plastic. Apidologie 50, 230–233 (2019).
Google Scholar  More

Sites overview and characteristicsThis study focused on three regions located in subantarctic, arctic, and subarctic latitudes. The respective latitudinal and longitudinal ranges covered in this study were: 54.95 to 52.08 °S, and 72.03 to 67.34 °W in Patagonia; 67.44 to 67.54 °N, and 86.59 to 86.71 °E in Siberia; 63.21 to 68.63 °N, and −150.79 to −145.98 °W in Alaska (Figs. 1 and 2). The exact coordinates for each sample were included in the submitted dataset. The field campaigns were conducted in 2016, during the summer for each respective region: January-February in Chilean Patagonia, June-July in Alaska and July-August in Siberia.Fig. 1Location of the three areas included in this study (panel a). The permafrost state and the number of sites and samples per region is indicated for each area. General views of 5 sites are provided as examples (b–f). Panel B provides a large view of the ecosystem surrounding the wetland ALP2 (Alaska, exact location indicated by the white circle). Lake PCL1 (panel c) is representative of the lakes on Navarino island (Chilean Patagonia). The glacial lake SIL2 is shown in panel d. At site SIP5, the hollow at first plan is surrounded by palsa (hummock, second plan), characterized by dark organic matter and lichen vegetation (panel e). The PPP3 peatland shown in panel f is dominated by Sphagnum magellanicum, like most peatlands in the area.Full size imageFig. 2Maps of sampling sites in Patagonia, Alaska and Siberia, indicating the ecosystem type (lake, wetland, soil). The tables show the complete- (in white) and the partial- (in grey) characterization sites. The exact coordinates of each sample are provided in the data record (See data records section).Full size imageFor every site included in the present study, a set of nine qualitative environmental and/or ecological site-scale descriptors was selected and adapted from ENVO Environment Ontology40, which included for example permafrost state, biome, environmental feature and vegetation type (Table 1, Fig. 3). Permafrost state was obtained from the NSIDC permafrost map41. The biome, large-scale descriptor based on climate and vegetation criteria, was derived from Olson et al.42. Temperate forest, boreal forest, and tundra biomes were included. The environmental features that were representative for the three regions were considered: lakes, wetlands, broadleaf/coniferous/mixed forest soils, grassland, tundra, and palsa. All the metadata was included in the submitted dataset. Table 2 summarizes the main types of sampled ecosystems and their main characteristics in the three regions, while Supplementary Table S1 provides the details of each sampling site.Table 1 Overview of the dataset contained in Mimarks sheet.Full size tableFig. 3Description of the qualitative environmental/ecological descriptors used to describe every sample, derived from ENVO Environment Ontology40.Full size imageTable 2 Main types of sampled ecosystems in the three studied regions.Full size tableIn Alaska, the studied area ranged from the Alaska Range and Fairbanks area (interior, continental climate, 63–65°N, discontinuous permafrost) up to Toolik Field Station (North Slope, arctic climate, 66–69°N, continuous permafrost; Fig. 2). The physiochemistry and CH4 emissions of lakes ALL1 (Killarney lake), ALL2 (Otto lake), ALL3 (Nutella lake), and ALL4 (Goldstream lake) were previously characterized35. A number of heterogeneous soil and wetland samples were collected around the studied Alaskan lakes and/or from monitored sites, as detailed in Supplementary Table S1. In the Alaska Range and Fairbanks area, soils were mostly covered by mixed or taiga forests, alpine tundra, and bogs or fens wetlands. In the norther Brooks Ranges mountain system, the landscape was piedmont hills with a predominant soil of porous organic peat underlain by silt and glacial till, all in a permafrost state, characterized mainly by Sphagnum and Eriophorum vegetation, as well as dwarf shrubs.In Siberia, the studied area was located in the discontinuous permafrost region surrounding Igarka, on the eastern bank of the Yenisei River (Fig. 2). This region was mainly covered by forest, dominated by larch (Larix Siberica), birch (Betula Pendula), and Siberian pine (Pinus Siberica), and palsa landscapes (frozen peat mounts), the latter being dominated by moss, lichens, Labrador tea and dwarf birch. In degraded areas, thermokarst bogs were dominated by Sphagnum spp. and Eriophorum spp. Land cover was an indicator of permafrost status, since forested areas reflected a deep permafrost table ( >2 m) associated with Pleistocene permafrost, while palsa-dominated landscapes were indicative of the presence of near-surface ( More

Most records of N-tiger cats were from savanna environments, and it was not surprising that this vegetative formation has a key influence on the N-tiger cat range in the Amazon. The bulk of the L. t. tigrinus distribution lies in the savannas, dry forests and shrublands of the Cerrado and Caatinga biomes. These are also the areas with the vast majority of records for this lowland subspecies (Supplemental Fig. S5). Hence, L. t. tigrinus is more associated with savannas and savanna-like environments than with rainforests. In fact, more than 80% of the records in the Amazon were within 100 km of a savanna patch. Colonization of the northern savanna formations of the Amazon by the N-tiger cat likely occurred during the forest-savanna shifts of the glacial period18, and the cat currently shows a patchy distribution. Strong evidence of established biogeographic corridor connections between the savannas of the Cerrado and those of the Amazon exists, suggesting northward expansion of the former during glacial periods, perhaps predating the Last Glacial Maximum19,20,21. Further corroborating this evidence, tiger cat ‘gene flow’ niche modelling showed prior connectivity between the Guiana population and that of Central Brazil and no connectivity with the Andean population22. Additionally, Guianan tiger cat skin patterns are found in savanna and transitional savanna/Amazon areas and in the semiarid shrub-woodland of Brazil and are very distinct from the patterns of the tiger cats from the Andes of northwestern South America and Central America (Supplementary Information Fig. S6).The bioclimatic variables in the best model also supported the cat’s preference for savanna areas. The best model indicated a positive effect of precipitation in the driest month on the probability of the presence of the N-tiger cat, likely indicating the Aw/As climates of tropical savannas23. These climates are marked by seasonal variation in rainfall, with a pronounced dry season. Higher rainfall during the dry season favors the growth of vegetation, which results in some tree cover within the savannas. Thus, our results agree with previous research suggesting that tiger cats avoid open savanna formations24. Similarly, the species had a significant negative response to net primary productivity. This also supports the species’ avoidance of dense lowland rainforests, which are the most productive habitats. In the Amazon biome, the least productive areas are found in more open landscapes25.The N-tiger cat’s range considered from an ecoregion perspective12 could biogeographically explain its distribution in the Amazon. All records but 2 fell within Guiana savannas, Guiana highland forest, Guiana rainforest, part of the Uatumã-Trombetas rainforest bordering the Guianas or all of it connecting to Gurupá and Monte Alegre varzea forests, as well as Marajó varzeas, the interfluve Tocantins-Araguaia/Maranhão, and the southern block of the interfluve Xingu/Tocantins-Araguaia. There were two records from the Negro-Branco moist forest, which also includes savanna-like “campinarana” formations. The range also reaches the transitional babaçu palm forests of Maranhão and the Mato Grosso seasonal forests (Supplementary Information Fig. S7, Table S3). The N-tiger cat’s range in the Amazon was determined by combining records with species distribution modeling, also matching the ecoregion perspective.Outside the Guiana Shield and likely the savanna patches of the region of the Upper Negro River, in other parts of the Amazon, the N-tiger cat seems to be restricted to the forests of the eastern Amazon, along the arc of deforestation and to transitional areas with savanna formations. The presence and absence points at camera-trapping sites could explain the N-tiger cat’s range in the Amazon and define its distribution range in the biome. Absence points, for instance, were usually located in dense rainforest habitats throughout the Amazon biome.The species may occasionally occupy rainforests, such as those of the Guianas, where it tends to be very rare. At a site in central Suriname, after an enormous trapping effort of  > 20,000 trap days in four years by cat specialists, over an area  > 1100 km2, no records of the N-tiger cat were found (Supplementary Information Table S2), although its presence is expected in that area26. This finding attests to the inherent rarity of this felid in its limited range within the Amazon. However, could its association with the arc of deforestation be related to the replacement of forest by bushy savanna-like vegetation that succeeds abandoned pastures? The other currently recognized subspecies, L. t. pardinoides (the Andean tiger cat) and L. t. oncilla (the oncilla), and the recently split southern tiger cat L. guttulus are all associated with forested areas. Conversely, L. t. tigrinus has higher abundance and is mostly found in the nonforested habitats of the Cerrado and Caatinga domains of Brazil and only rarely in rainforests. Thus, L. t. tigrinus may be an open-habitat (sub)species. However, within savannas, N-tiger cats are restricted to denser savanna formations, with open savannas deemed unsuitable24. In the semiarid Caatinga, the N-tiger cat also prefers denser formations27,28.One of the most interesting findings was the clear relationship between the ranges of the dominant mesopredator and subordinate species. The ranges of ocelots and N-tiger cats in the Amazon were diametrically opposite (Fig. 1), a finding never recorded for felids. The reported ocelot densities and relative abundance indexes (RAIs) in the Amazon range from 0.29 to 0.95 ind/km2 and 0.07–13.2 ind/100 trap-days, respectively7,29. Thus, the expected ocelot density found using modeling that allows for N-tiger cat presence is very low (Fig. 2A). In the Rupununi, the ocelot:N-tiger cat RAI ratio was roughly 10:1, with a very low RAI and expected density for N-tiger cats (see Supplementary Material). The only other relative abundance estimate of tiger cats presented for the Amazon30 was not confirmed as an estimate of tiger cats following inspection of the original records by the authors but as an estimate of margays or ocelots. This antagonistic relationship between ocelots and all other small cat species in their area of sympatry is quite impressive. It is density-dependent, as it seems to take effect only above an ocelot density threshold of 0.12 ind./km231. The influence can range from patterns of density, distribution, and occupancy to spatial and temporal use. Conversely, such an impact was not detected when either the small cats or ocelots were compared to the larger cats31,32,33,34,35.In view of the Red List assessments and applying the limited estimates presented, the expected total population size for N-tiger cats in the Amazon would be approximately 150 and 1622 individuals, considering their AOO or EOO, respectively. Applying the IUCN’s formula for mature individuals8, these numbers would be 45 and 487 individuals for the AOO and EOO, respectively.The ocelot’s preference for very dense rainforests may explain the low probability of N-tiger cat occurrence within the Amazon biome. Notably, most tiger cat records from rainforests and all those from premontane forests came from the Guiana Shield, a region where tropical grasslands and savannas dot more forested landscapes. The Guiana Highlands and Pantepui ecoregions, which make up a considerable portion of the shield, tend to have low ocelot densities (below 0.30 ind/km2), although they do contain some rainforest. Ocelot densities reach some of their lowest values in the Guianan savanna ecoregion (mean ocelot density of 0.029 in the savanna formations), where the N-tiger cat probability of occurrence was highest. At the Karanambu site in the Rupununi, all ocelot records came from either gallery forests or forest patches embedded in the savanna. Although the data did not allow us to test further hypotheses, it is likely that spatial partitioning occurs in the Guiana Shield, with N-tiger cats favoring habitats that are more open. Conversely, areas farther west in the Amazon biome, other than the predicted area, do not have any major savanna patches and are covered mostly by lowland tropical rainforest formations, where ocelots can potentially reach densities in excess of 0.7 ind/km2. Of all Amazonian records of N-tiger cats, only one came from west of the 68th meridian: a preserved specimen from Puerto Leguizamo on the Putumayo River in Colombia. The specimen was identified as L. t. pardinoides by its collector, so it most likely represents an individual that came down from the foothills of the Andes. Alternatively, it could have been caught in the Andean foothills but labeled generally as from Puerto Leguizamo, as museum records do not always present precise locations, like most of those from our dataset; thus, they could represent a broader region, not a single collection location.The records of L. t. tigrinus in the Monte-Alegre Várzea ecoregion and Tapajós-Xingu Moist Forest ecoregion (which shares a border with the Amazon River) are actually from the small savanna patches of Terra Santa and Alter do Chão, respectively, which are imbedded within the forests of these ecoregions. Similarly, the Negro-Branco Moist Forest ecoregion includes open-canopy white sand forests with savanna-like vegetation, known as ‘campinaranas’36.Although our model predicted a high probability of N-tiger cat presence in the Marajó Várzea ecoregion, the records from the island came from savanna patches and not from flooded forests and mangroves. Hence, we did not include such large areas in the AOO for the subspecies. It is likely that the highly predicted probability of presence there is an artifact of low predicted ocelot density. Nevertheless, the environment there is not suitable for either cat. Our ocelot density model was highly significant and explained almost 50% of the variation in ocelot density. The remaining variation was related to either other variables that could not be measured via satellite imagery (such as prey availability) or the sampling design of the different studies. Nonetheless, ocelot densities predicted from our model across the Amazon were within the expected range for the species29.Why are N-tiger cats absent in camera-trapping studies in Amazonian forests throughout the biome? The most straightforward answer seems to be because they simply are not there (central and western Amazon) or, where present, their numbers are extremely low (Guianas and eastern Amazon). The lack of surveys cannot be cited as a potential reason for their apparent absence because the studies that did not detect the species were conducted throughout the Amazon biome, in all nine Amazonian countries. Some of the areas have been surveyed for several years—or decades in some cases—and have failed to record a single individual (Supplementary Information Table S2). Typically, N-tiger cats appear, even prominently, on cameras in other biomes, such as in the savannas of the Cerrado and semiarid scrub of the Caatinga domain in Brazil, including sites where ocelots are present24,27,37. Clouded tiger cats (L. t. pardinoides) have also been frequently recorded on cameras in the Andes, higher than 1500 m above sea level34,38, but not in lowland Amazonian forests. This finding indicates that the N-tiger cat is not camera-shy. In northern Brazilian savannas, its density can reach 0.25 ind/km2 24. Coincidentally, this highest density estimate of the N-tiger cat is the same as the lowest ocelot density estimate for Amazonian forests24,29.Tiger cats and margays show high similarity, making misidentifications relatively common39. However, the evaluation of  > 3000 camera trap images of small-medium felids in the Amazon revealed that only one mildly resembled a tiger cat, a finding that supports the species being absent there and does not represent a case of mistaken identity with margays or even ocelots7.The Amazonian range of L. tigrinus is very limited, and populations are expected to be very small. With the upcoming split of L. t. tigrinus and L. t. pardinoides into two different species40, this situation would have serious implications for the conservation of the former. Thus, L. t. tigrinus conservation lies outside the “Amazonian safe haven” of most other carnivore species found there7. The Brazilian drylands Cerrado and Caatinga represent such places for L. t. tigrinus populations. Unfortunately, these biomes have had  > 50% of their cover completely removed41. Very importantly, besides being extremely rare in the Amazonian savannas, this rather limited vegetative formation is also considered highly threatened and of conservation priority42. Therefore, the tiger cat could become an emblematic flagship species representing the uniqueness of this vegetative formation in dire need of protection.In short, the picture that emerges is that although the N-tiger cat uses both rainforests and deciduous forests in the Amazon, it seems to be mostly associated with savanna formations and that its distribution in the Amazon is highly influenced by the ocelot, the dominant mesopredator. The N-tiger cat’s inherent rarity, expected population size, and restricted range in the Amazon suggest that this biome does not in fact represent a safe haven for the global conservation of this small felid. In addition to shedding light on and refining the N-tiger cat distribution in the Amazon, this paper highlights the importance of including biological variables, such as the potential impacts of competitors and predators on species presence and distribution, in SDMs. More

Davies EK, Peters AD, Keightley PD (1999) High frequency of cryptic deleterious mutations in Caenorhabditis elegans. Science 285:1748–1751Article 
CAS 
PubMed 

Google Scholar 
Eyre-Walker A, Keightley PD (2007) The distribution of fitness effects of new mutations. Nat Rev Genet 8:610–618Article 
CAS 
PubMed 

Google Scholar 
Eyre-Walker A, Keightley PD (2013) A comparison of models to infer the distribution of fitness effects of new mutations. Genetics 193:1197–1208Article 

Google Scholar 
Fry JD, Keightley PD, Heinsohn SL, Nuzhdi SV (1999) New estimates of the rates and effects of mildly deleterious mutation in Drosophila melanogaster. Proc Natl Acad Sci 96:574–579Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
García-Dorado A (2007) Shortcut predictions for fitness properties at the mutation-selection-drift balance and for its buildup after size reduction under different management strategies. Genetics 176:983–997Article 
PubMed 
PubMed Central 

Google Scholar 
García-Dorado A (2012) Understanding and predicting the fitness decline of shrunk populations: inbreeding, purging, mutation, and standard selection. Genetics 190:1461–1476Article 
PubMed 
PubMed Central 

Google Scholar 
García-Dorado A (2015) On the consequences of ignoring purging on genetic recommendations for minimum viable population rules. Heredity 115:185–187Article 
PubMed 
PubMed Central 

Google Scholar 
García-Dorado A, Caballero A (2021) Neutral genetic diversity as a useful tool for conservation biology. Conserv Genet 22:541–545Article 

Google Scholar 
Garner BA, Hoban S, Luikart G (2020) IUCN Red List and the value of integrating genetics. Conserv Genet 21:795–801Article 

Google Scholar 
Hedrick PW, García-Dorado A (2016) Understanding inbreeding depression, purging, and genetic rescue. Trends Ecol Evol 31:940–952Article 
PubMed 

Google Scholar 
Kardos M, Armstrong EE, Fitzpatrick SW, Hauser S, Hedrick PW, Miller J et al. (2021) The crucial role of genome-wide genetic variation in conservation. Proc Natl Acad Sci USA 118:e2104642118Khan A, Patel A, Shukla H, Viswanathan A, van der Valk T, Borthakur U, … & Ramakrishnan U (2021) Genomic evidence for inbreeding depression and purging of deleterious genetic variation in Indian tigers. Proc. Natl. Acad. Sci. 118Kimura M, Maruyama T, Crow JF (1963) The mutation load in small populations. Genetics 48:1303–1312Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Kimura M (1980) Average time until fixation of a mutant allele in a finite population under continued mutation pressure: Studies by analytical, numerical, and pseudo-sampling methods. Proc Natl Acad Sci 77:522–526Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Morin PA, Archer FI, Avila CD, Balacco JR, Bukhman YV, Chow, W, … & Jarvis ED (2021) Reference genome and demographic history of the most endangered marine mammal, the vaquita. Mol Ecol Resour 21:1008–1020Mukai T (1964) The genetic structure of natural populations of Drosophila melanogaster. I. Spontaneous mutation rate of polygenes controlling viability. Genetics 50:1–19Article 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Nietlisbach P, Muff S, Reid JM, Whitlock MC, Keller LF (2019) Nonequivalent lethal equivalents: Models and inbreeding metrics for unbiased estimation of inbreeding load. Evol Applic 12:266–279Article 

Google Scholar 
O’Grady JJ, Brook BW, Reed DH, Ballou JD, Tonkyn DW, Frankham R (2006) Realistic levels of inbreeding depression strongly affect extinction risk in wild populations. Biol Conserv 133:42–51Article 

Google Scholar 
Pérez-Pereira N, Caballero A, García-Dorado A (2021) Reviewing the consequences of genetic purging on the success of rescue programs. Conserv Gen 23:1–17Article 

Google Scholar 
Pérez-Pereira N, Wang J, Quesada H, Caballero A (2022). Prediction of the minimum effective size of a population viable in the long term. Biodivers Conserv https://doi.org/10.1007/s10531-022-02456-zRobinson JA, Kyriazis CC, Nigenda-Morales SF, Beichman AC, Rojas-Bracho L, Robertson KM et al. (2022) The critically endangered vaquita is not doomed to extinction by inbreeding depression. Science 376:635–639Article 
CAS 
PubMed 

Google Scholar 
Teixeira JC, Huber CD (2021) The inflated significance of neutral genetic diversity in conservation genetics. Proc Natl Acad Sci USA 118:e2015096118Wade EE, Kyriazis C, Cavassim MIA, Lohmueller KE (2022) Quantifying the fraction of new mutations that are recessive lethal. bioRxiv 1–24, https://www.biorxiv.org/content/10.1101/2022.04.22.489225v1 More

DataWe tested our models on ten publicly available datasets. In Fig. 4 we show examples of images from each of the datasets. When applicable, the training and test splits were kept the same as in the original dataset. For example, the ZooScan, Kaggle, EILAT, and RSMAS datasets lack a specific training and test set; in these cases, benchmarks come from k-fold cross-validation51,52, and we followed the exact same procedures in order to allow for a fair comparison.Figure 4Examples of images from each of the datasets.(a) RSMAS (b) EILAT (c) ZooLake (d) WHOI (e) Kaggle (f) ZooScan (g) NA-Birds (h) Stanford dogs (i) SriLankan Beetles (j) Florida Wildtrap.Full size imageRSMAS This is a small coral dataset of 766 RGB image patches with a size of (256times 256) pixels each53. The patches were cropped out of bigger images obtained by the University of Miami’s Rosenstiel School of Marine and Atmospheric Sciences. These images were captured using various cameras in various locations. The data is separated into 14 unbalanced groups and whose labels correspond to the names of the coral species in Latin. The current SOTA for the classification of this dataset is by52. They use the ensemble of best performing 11 CNN models. The best models were chosen based on sequential forward feature selection (SFFS) approach. Since an independent test is not available, they make use of 5-fold cross-validation for benchmarking the performances.EILAT This is a coral dataset of 1123 64-pixel RGB image patches53 that were created from larger images that were taken from coral reefs near Eilat in the Red sea. The image dataset is partitioned into eight classes, with an unequal distribution of data. The names of the classes correspond to the shorter version of the scientific names of the coral species. The current SOTA52 for the classification of this dataset uses the ensemble of best performing 11 CNN models similar to RSMAS dataset and 5-fold cross-validation for benchmarking the performances.ZooLake This dataset consists of 17943 images of lake plankton from 35 classes, acquired using a Dual-magnification Scripps Plankton Camera (DSPC) in Lake Greifensee (Switzerland) between 2018 and 2020 14,54. The images are colored, with a black background and an uneven class distribution. The current SOTA22 on this dataset is based on a stacking ensemble of 6 CNN models on an independent test set.WHOI This dataset 55 contains images of marine plankton acquired by Image FlowCytobot56, from Woods Hole Harbor water. The sampling was done between late fall and early spring in 2004 and 2005. It contains 6600 greyscale images of different sizes, from 22 manually categorized plankton classes with an equal number of samples for each class. The majority of the classes belonging to phytoplankton at genus level. This dataset was later extended to include 3.4M images and 103 classes. The WHOI subset that we use was previously used for benchmarking plankton classification models51,52. The current SOTA22 on this dataset is based on average ensemble of 6 CNN models on an independent test set.Kaggle-plankton The original Kaggle-plankton dataset consists of plankton images that were acquired by In-situ Ichthyoplankton Imaging System (ISIIS) technology from May to June 2014 in the Straits of Florida. The dataset was published on Kaggle (https://www.kaggle.com/c/datasciencebowl) with images originating from the Hatfield Marine Science Center at Oregon State University. A subset of the original Kaggle-plankton dataset was published by51 to benchmark the plankton classification tasks. This subset comprises of 14,374 greyscale images from 38 classes, and the distribution among classes is not uniform, but each class has at least 100 samples. The current SOTA22 uses average ensemble of 6 CNN models and benchmarks the performance using 5-fold cross-validation.ZooScan The ZooScan dataset consists of 3771 greyscale plankton images acquired using the Zooscan technology from the Bay of Villefranche-sur-mer57. This dataset was used for benchmarking the classification models in previous plankton recognition papers51,52. The dataset consists of 20 classes with a variable number of samples for each class ranging from 28 to 427. The current SOTA22 uses average ensemble of 6 CNN models and benchmarks the performance using 2-fold cross-validation.NA-Birds NA-Birds58 is a collection of 48,000 captioned pictures of North America’s 400 most often seen bird species. For each species, there are over 100 images accessible, with distinct annotations for males, females, and juveniles, totaling 555 visual categories. The current SOTA59 called TransFG modifies the pure ViT model by adding contrastive feature learning and part selection module that replaces the original input sequence to the transformer layer with tokens corresponding to informative regions such that the distance of representations between confusing subcategories can be enlarged. They make use of an independent test set for benchmarking the model performances.Stanford Dogs The Stanford Dogs dataset comprises 20,580 color images of 120 different dog breeds from all around the globe, separated into 12,000 training images and 8,580 testing images60. The current SOTA59 makes use of modified ViT model called TransFG as explained above in NA-Birds dataset. They make use of an independent test set for benchmarking the model performances.Sri Lankan Beetles The arboreal tiger beetle data61 consists of 380 images that were taken between August 2017 and September 2020 from 22 places in Sri Lanka, including all climatic zones and provinces, as well as 14 districts. Tricondyla (3 species), Derocrania (5 species), and Neocollyris (1 species) were among the nine species discovered, with six of them being endemic . The current SOTA61 makes use of CNN-based SqueezeNet architecture and was trained using pre-trained weights of ImageNet. The benchmarking of the model performances was done on an independent test set.Florida Wild Traps The wildlife camera trap62 classification dataset comprises 104,495 images with visually similar species, varied lighting conditions, skewed class distribution, and samples of endangered species, such as Florida panthers. These were collected from two locations in Southwestern Florida. These images are categorized in to 22 classes. The current SOTA62 makes use of CNN-based ResNet-50 architecture and the performance of the model was benchmarked on an independent test set.ModelsVision transformers (ViTs)31 are an adaptation to computer vision of the Transformers, which were originally developed for natural language processing30. Their distinguishing feature is that, instead of exploiting translational symmetry, as CNNs do, they have an attention mechanism which identifies the most relevant part of an image. ViTs have recently outperformed CNNs in image classification tasks where vast amounts of training data and processing resources are available30,63. However, for the vast majority of use cases and consumers, where data and/or computational resources are limiting, ViTs are essentially untrainable, even when the network architecture is defined and no architectural optimization is required. To settle this issue, Data-efficient Image Transformers (DeiTs) were proposed32. These are transformer models that are designed to be trained with much less data and with far less computing resources32. In DeiTs, the transformer architecture has been modified to allow native distillation64, in which a student neural network learns from the results of a teacher model. Here, a CNN is used as the teacher model, and the pure vision transformer is used as the student network. All the DeiT models we report on here are DeiT-Base models32. The ViTs are ViT-B16, ViT-B32, and ViT-L32 models31.ImplementationTo train our models, we used transfer learning65: we took a model that was already pre-trained on the ImageNet43 dataset, changed the last layers depending on the number of classes, and then fine-tuned the whole network with a very low learning rate. All the models were trained with two Nvidia GTX 2080Ti GPUs.DeiTs We used DeiT-Base32 architecture, using the Python package TIMM66, which includes many of the well-known deep learning architectures, along with their pre-trained weights computed from the ImageNet dataset43. We resized the input images to 224 x 224 pixels and then, to prevent the model from overfitting at the pixel level and help it generalize better, we employed typical image augmentations during training such as horizontal and vertical flips, rotations up to 180 degrees, small zoom up’s to 20%, a small Gaussian blur, and shearing up to 10%. To handle class imbalance, we used class reweighting, which reweights errors on each example by how present that class is in the dataset67. We used sklearn utilities68 to calculate the class weights which we employed during the training phase.The training phase started with a default pytorch69 initial conditions (Kaiming uniform initializer), an AdamW optimizer with cosine annealing70, with a base learning rate of (10^{-4}), and a weight decay value of 0.03, batch size of 32 and was supervised using cross-entropy loss. We trained with early stopping, interrupting training if the validation F1-score did not improve for 5 epochs. The learning rate was then dropped by a factor of 10. We iterated until the learning rate reached its final value of (10^{-6}). This procedure amounted to around 100 epochs in total, independent of the dataset. The training time varied depending on the size of the datasets. It ranged between 20min (SriLankan Beetles) to 9h (Florida Wildtrap). We used the same procedure for all the datasets: no extra time was needed for hyperparameter tuning.ViTs We implemented the ViT-B16, ViT-B32 and ViT-L32 models using the Python package vit-keras (https://github.com/faustomorales/vit-keras), which includes pre-trained weights computed from the ImageNet43 dataset and the Tensorflow library71.First, we resized input images to 128 × 128 and employed typical image augmentations during training such as horizontal and vertical flips, rotations up to 180 degrees, small zooms up to 20%, small Gaussian blur, and shearing up to 10%. To handle class imbalance, we calculated the class weights and use them during the training phase.Using transfer learning, we imported the pre-trained model and froze all of the layers to train the model. We removed the last layer, and in its place we added a dense layer with (n_c) outputs (being (n_c) the number of classes), was preceded and followed by a dropout layer. We used the Keras-tuner72 with Bayesian optimization search73 to determine the best set of hyperparameters, which included the dropout rate, learning-rate, and dense layer parameters (10 trials and 100 epochs). After that, the model with the best hyperparameters was trained with a default tensorflow71 initial condition (Glorot uniform initializer) for 150 epochs using early stopping, which involved halting the training if the validation loss did not decrease after 50 epochs and retaining the model parameters that had the lowest validation loss.CNNs CNNs included DenseNet38, MobileNet39, EfficientNet-B240, EfficientNet-B540, EfficientNet-B640, and EfficientNet-B740 architectures. We followed the training procedure described in Ref.22, and carried out the training in tensorflow.Ensemble learningWe adopted average ensembling, which takes the confidence vectors of different learners, and produces a prediction based on the average among the confidence vectors. With this procedure, all the individual models contribute equally to the final prediction, irrespective of their validation performance. Ensembling usually results in superior overall classification metrics and model robustness74,75.Given a set of n models, with prediction vectors (vec c_i~(i=1,ldots ,n)), these are typically aggregated through an arithmetic average. The components of the ensembled confidence vector (vec c_{AA}), related to each class (alpha ) are then$$begin{aligned} c_{AA,alpha } = frac{1}{n}sum _{i=1}^n c_{i,alpha },. end{aligned}$$
(2)
Another option is to use a geometric average,$$begin{aligned} c_{GA,alpha } = root n of {prod _{i=1}^n c_{i,alpha }},. end{aligned}$$
(3)
We can normalize the vector (vec c_g), but this is not relevant, since we are interested in its largest component, (displaystyle max _alpha (c_{GA,alpha })), and normalization affects all the components in the same way. As a matter of fact, also the nth root does not change the relative magnitude of the components, so instead of (vec c_{GA}) we can use a product rule: (displaystyle max _alpha (c_{GA,alpha })=max _alpha (c_{PROD,alpha })), with (displaystyle c_{PROD,alpha } = prod _{i=1}^n c_{i,alpha }).While these two kinds of averaging are equivalent in the case of two models and two classes, they are generally different in any other case33. For example, it can easily be seen that the geometric average penalizes more strongly the classes for which at least one learner has a very low confidence value, a property that was termed veto mechanism36 (note that, while in Ref.36 the term veto is used when the confidence value is exactly zero, here we use this term in a slightly looser way). More

Buxó, R. & Piqué, R. Arqueobotánica: Los Usos de las Plantas en la Península Ibérica. (Grupo Planeta GBS, 2008).Miller, N. F., Spengler, R. N. & Frachetti, M. Millet cultivation across Eurasia: Origins, spread, and the influence of seasonal climate. Holocene 26, 1566–1575 (2016).ADS 

Google Scholar 
James, T. K., Rahman, A., McGill, C. R. & Trivedi, P. D. Biology and survival of broomcorn millet (Panicum miliaceum) seed. N. Z. Plant Prot. 64, 142–148 (2011).
Google Scholar 
Kirleis, W., Dal Corso, M. & Filipović, D. Millet and What Else?: The Wider Context of the Adoption of Millet Cultivation in Europe. vol. 14 (Sidestone Press, 2022).Sherratt, A. Water, soil and seasonality in early cereal cultivation. World Archaeol. 11, 313–330 (1980).
Google Scholar 
Rachie, K. O. The Millets: Importance, Utilization and Outlook. 74 (International Crops Research Institute for the Semi-Arid Tropics, 1975).Moreno-Larrazabal, A., Teira-Brión, A., Sopelana-Salcedo, I., Arranz-Otaegui, A. & Zapata, L. Ethnobotany of millet cultivation in the north of the Iberian Peninsula. Veg. Hist. Archaeobot. 24, 541–554 (2015).
Google Scholar 
Liu, L. et al. The origins of specialized pottery and diverse alcohol fermentation techniques in Early Neolithic China. Proc. Natl. Acad. Sci. USA 116, 12767–12774 (2019).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Tereso, J. P. et al. Agriculture in NW Iberia during the Bronze Age: A review of archaeobotanical data. J. Archaeol. Sci. Rep. 10, 44–58 (2016).
Google Scholar 
Liu, C., Kong, Z. & Lang, S. D. A discussion on agricultural and botanical remains and the human ecology of Dadiwan site, in Chinese. Zhongyuan Wenwu 4, 25–29 (2004).
Google Scholar 
Zhao, Z. New archaeobotanic data for the study of the origins of agriculture in China. Curr. Anthropol. 52, S295–S306 (2011).
Google Scholar 
Crawford, G. W., Xuexiang, C., Fengshi, L. & Jianhua, W. A Preliminary analysis on plant remains of the Yuezhuang site in Changqing District, Jinan City, Shandong Province. Jianghan Archaeol. 2, 107–113 (2013).
Google Scholar 
Frachetti, M. D. Multiregional emergence of mobile pastoralism and nonuniform institutional complexity across Eurasia. Curr. Anthropol. 53, 2–38. https://doi.org/10.1086/663692 (2012).Article 

Google Scholar 
Ventresca Miller, A. R. & Makarewicz, C. A. Intensification in pastoralist cereal use coincides with the expansion of trans-regional networks in the Eurasian Steppe. Sci. Rep. 9, 8363 (2019).ADS 
PubMed 
PubMed Central 

Google Scholar 
Jones, M. et al. Food globalisation in prehistory: The agrarian foundations of an interconnected continent. J. Br. Acad. 4, 73–87 (2016).
Google Scholar 
Spengler, R. et al. Early agriculture and crop transmission among Bronze Age mobile pastoralists of Central Eurasia. Proc. Biol. Sci. 281, 20133382 (2014).PubMed 
PubMed Central 

Google Scholar 
Hermes, T. R. et al. Early integration of pastoralism and millet cultivation in Bronze Age Eurasia. Proc. Biol. Sci. 286, 20191273 (2019).CAS 
PubMed 
PubMed Central 

Google Scholar 
Frachetti, M. D., Spengler, R. N., Fritz, G. J. & Maryashev, A. N. Earliest direct evidence for broomcorn millet and wheat in the central Eurasian steppe region. Antiquity 84, 993–1010 (2010).
Google Scholar 
Motuzaite-Matuzeviciute, G., Richard, A. S., Hunt, H. V., Liu, X. & Jones, M. K. The early chronology of broomcorn millet (Panicum Miliaceum) in Europe. Antiquity 338, 1073–1085 (2013).
Google Scholar 
Filipović, D. et al. New AMS 14C dates track the arrival and spread of broomcorn millet cultivation and agricultural change in prehistoric Europe. Sci. Rep. 10, 13698 (2020).ADS 
PubMed 
PubMed Central 

Google Scholar 
Hunt, H. V. et al. Millets across Eurasia: Chronology and context of early records of the genera Panicum and Setaria from archaeological sites in the Old World. Veg. Hist. Archaeobot. 17, 5–18 (2008).PubMed 
PubMed Central 

Google Scholar 
Brudenell, M., Fosberry, R., Phillips, T. & Kwiatkowska, M. Early cultivation of broomcorn millet in southern Britain: Evidence from the Late Bronze Age settlement site of Old Catton, Norfolk. Antiquity 2022, 1–6 (2022).
Google Scholar 
Weber, S. A. & Fuller, D. Q. Millets and their role in early agriculture. Pragdhara 18, 69–90 (2007).
Google Scholar 
Shelton, C. P. & White, C. E. The hand-pump flotation system: A new method for archaeobotanical recovery. J. Field Archaeol. 35, 316–326 (2010).
Google Scholar 
Barboff, M. Le millet au Portugal. In Millet– Hirse–Millet. Actes du Congres d’Aizenay (ed. Hörandner, E.) 113–122 (Grazer Beitra¨ge zur 731 europa¨ischen Ethnologie, 1995).Reddy, S. N. If the Threshing Floor Could Talk: Integration of Agriculture and Pastoralism during the Late Harappan in Gujarat, India. J. Anthropol. Archaeol. 16, 162–187 (1997).
Google Scholar 
Dayakar Rao, B. et al. Nutritional and health benefits of millets. In ICAR_Indian Institute of Millets Research (IIMR), Rajendranagar, Hyderabad 112 (2017).Mariotti-Lippi, M., Pisaneschi, L., Sarti, L., Lari, M. & Moggi-Cecchi, J. Insights into the Copper-Bronze Age diet in Central Italy: Plant microremains in dental calculus from Grotta dello Scoglietto (Southern Tuscany, Italy). J. Archaeol. Sci. Rep. 15, 30–39 (2017).
Google Scholar 
Lu, H. et al. Phytoliths analysis for the discrimination of Foxtail millet (Setaria italica) and Common millet (Panicum miliaceum). PLoS ONE 4, e4448 (2009).ADS 
PubMed 
PubMed Central 

Google Scholar 
Lucarini, G., Radini, A., Barton, H. & Barker, G. The exploitation of wild plants in Neolithic North Africa. Use-wear and residue analysis on ground stone tools from the Farafra Oasis, Egypt. Quat. Int. 410, 77–92 (2016).
Google Scholar 
Madella, M., Lancelotti, C. & García-Granero, J. J. Millet microremains—an alternative approach to understand cultivation and use of critical crops in Prehistory. Archaeol. Anthropol. Sci. 8, 17–28 (2016).
Google Scholar 
Yang, X. et al. From the modern to the archaeological: Starch grains from millets and their wild relatives in China. J. Archaeol. Sci. 39, 247–254 (2012).
Google Scholar 
Lightfoot, E., Liu, X. & Jones, M. K. Why move starchy cereals? A review of the isotopic evidence for prehistoric millet consumption across Eurasia. World Archaeol. 45, 574–623 (2013).
Google Scholar 
Armendariz, A. In Las cuevas sepulcrales del País Vasco. (Tesis Doctoral Inédita, Universidad del País Vasco-Euskal Herriko Unibertsitatea, 1992).Vazquez-Varela, J. M. El cultivo del mijo, (Panicum miliaceum, L.), en la cultura castreña del noroeste de la peninsula iberica. Cuad. Estud. Gallegos 1, 65–73 (1994).
Google Scholar 
Patterson, N. et al. Large-scale migration into Britain during the Middle to Late Bronze Age. Nature 601, 588–594 (2021).ADS 
PubMed 
PubMed Central 

Google Scholar 
Olalde, I. et al. The genomic history of the Iberian Peninsula over the past 8000 years. Science 363, 1230–1234 (2019).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Arias, P. & Armendariz, A. Aproximación a la Edad del Bronce en la región cantábrica. A Idade do Bronce en Galicia: novas perspectivas. Cadernos Semin. Sargadelos 77, 47–80 (1998).
Google Scholar 
DeNiro, M. J. Postmortem preservation and alteration of in vivo bone collagen isotope ratios in relation to palaeodietary reconstruction. Nature 317, 806–809 (1985).ADS 
CAS 

Google Scholar 
Ambrose, S. H. Preparation and characterization of bone and tooth collagen for isotopic analysis. J. Archaeol. Sci. 17, 431–451 (1990).
Google Scholar 
van Klinken, G. J. Bone collagen quality indicators for palaeodietary and radiocarbon measurements. J. Archaeol. Sci. 26, 687–695 (1999).
Google Scholar 
Nehlich, O. & Richards, M. P. Establishing collagen quality criteria for sulphur isotope analysis of archaeological bone collagen. Archaeol. Anthropol. Sci. 1, 59–75 (2009).
Google Scholar 
Cristiani, E., Radini, A., Edinborough, M. & Borić, D. Dental calculus reveals Mesolithic foragers in the Balkans consumed domesticated plant foods. Proc. Natl. Acad. Sci. USA 113, 10298–10303 (2016).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Henry, A. G. & Piperno, D. R. Using plant microfossils from dental calculus to recover human diet: A case study from Tell al-Raqā’I, Syria. J. Archaeol. Sci. 35, 1943–1950 (2008).
Google Scholar 
Minagawa, M. & Wada, E. Stepwise enrichment of 15N along food chains: Further evidence and the relation between δ15N and animal age. Geochim. Cosmochim. Acta 48, 1135–1140 (1984).ADS 
CAS 

Google Scholar 
Hedges, R. E. M. & Reynard, L. M. Nitrogen isotopes and the trophic level of humans in archaeology. J. Archaeol. Sci. 34, 1240–1251 (2007).
Google Scholar 
López-Costas, O., Müldner, G. & Martínez-Cortizas, A. Diet and lifestyle in Bronze Age Northwest Spain: The collective burial of Cova do Santo. J. Archaeol. Sci. 55, 209–218 (2015).
Google Scholar 
Jones, J. R. et al. Investigating prehistoric diet and lifeways of early farmers in central northern Spain (3000–1500 CAL BC) using stable isotope techniques. Archaeol. Anthropol. Sci. 11, 3979–3994 (2019).
Google Scholar 
DeNiro, M. J. & Epstein, S. Influence of diet on the distribution of carbon isotopes in animals. Geochim. Cosmochim. Acta 42, 495–506 (1978).ADS 
CAS 

Google Scholar 
O’Leary, M. H. Carbon isotope fractionation in plants. Phytochemistry 20, 553–567 (1981).
Google Scholar 
Chisholm, B. S., Nelson, D. E. & Schwarcz, H. P. Stable-carbon isotope ratios as a measure of marine versus terrestrial protein in ancient diets. Science 216, 1131–1132 (1982).ADS 
CAS 
PubMed 

Google Scholar 
de Blas Cortina, M. Á. De la caverna al lugar fortificado: Una mirada a la edad del bronce en el territorio Astur-Cántabro. Quad. Prehist. Arqueol. Castelló 29, 105–134 (2011).
Google Scholar 
Nehlich, O. The application of sulphur isotope analyses in archaeological research: A review. Earth-Sci. Rev. 142, 1–17 (2015).ADS 
CAS 

Google Scholar 
Richards, M. P., Fuller, B. T. & Hedges, R. E. M. Sulphur isotopic variation in ancient bone collagen from Europe: Implications for human palaeodiet, residence mobility, and modern pollutant studies. Earth Planet. Sci. Lett. 191, 185–190 (2001).ADS 
CAS 

Google Scholar 
González-Rabanal, B. et al. Diet, mobility and death of Late Neolithic and Chalcolithic groups of the Cantabrian Region (northern Spain). A multidisciplinary approach towards studying the Los Avellanos I and II burial caves. J. Archaeol. Sci. Rep. 34, 1–13 (2020).
Google Scholar 
McGovern, P. E. et al. Fermented beverages of pre- and proto-historic China. Proc. Natl. Acad. Sci. USA 101, 17593–17598 (2004).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Fernández-Crespo, T., Ordoño, J., Bogaard, A., Llanos, A. & Schulting, R. A snapshot of subsistence in Iron Age Iberia: The case of La Hoya village. J. Archaeol. Sci. Rep. 28, 1–10 (2019).
Google Scholar 
Hedges, R. E. M. On bone collagen?apatite-carbonate isotopic relationships. Int. J. Osteoarchaeol. 13, 66–79 (2003).
Google Scholar 
Arias, P. Determinaciones de isótopos estables en restos humanos de la región Cantábrica. Aportación al estudio de la dieta de las poblaciones del Mesolítico y el Neolítico. Munibe Antropol.-Arkeol. 57, 359–374 (2005).
Google Scholar 
Palencia-Madrid, L. et al. Ancient mitochondrial lineages support the prehistoric maternal root of Basques in Northern Iberian Peninsula. Eur. J. Hum. Genet. 25, 631–636 (2017).CAS 
PubMed 
PubMed Central 

Google Scholar 
Fernández-Crespo, T., Mujika, J. A. & Ordoño, J. Aproximación al patrón alimentario de los inhumados en la cista de la Edad del Bronce de Ondarre (Aralar, Guipúzcoa) a través del análisis de isótopos estables de carbono y nitrógeno sobre colágeno óseo. Trab. Prehist. 73, 325–334 (2016).
Google Scholar 
Higuero Pliego, A. In Análisis Isotópico de Carbono y Nitrógeno en Secuencias de Dentina y de Estroncio en Esmalte Procedente de Restos Humanos Prehistóricos de la Cueva de Los Canes (Cabrales, Asturias). (Tesis Doctoral Inédita, Universidad de Cantabria, 2020).Teira-Brión, A. Traditional millet cultivation in the Iberian Peninsula: Ethnoarchaeological reflections through the lens of social relations and economic concerns. In (eds. Kirleis, W. et al.) Millet and What Else?: The Wider Context of the Adoption of Millet Cultivation in Europe vol. 14 263–278 (Sidestone Press, 2022).Pechenkina, E. A., Ambrose, S. H., Xiaolin, M. & Benfer, R. A. Reconstructing northern Chinese Neolithic subsistence practices by isotopic analysis. J. Archaeol. Sci. 32, 1176–1189 (2005).
Google Scholar 
Hu, Y. et al. Palaeodietary study of Sanxingcun Site, Jintan, Jiangsu. Chin. Sci. Bull. 52, 660–664 (2007).
Google Scholar 
Motuzaite-Matuzeviciute, G., Ananyevskaya, E., Sakalauskaite, J., Soltobaev, O. & Tabaldiev, K. The integration of millet into the diet of Central Asian populations in the third millennium BC. Antiquity 96, 560–574 (2022).
Google Scholar 
Herrscher, E. et al. The origins of millet cultivation in the Caucasus: Archaeological and archaeometric approaches. Préhistoires Méditerr. 2018, 6 (2018).
Google Scholar 
Tafuri, M. A., Craig, O. E. & Canci, A. Stable isotope evidence for the consumption of millet and other plants in Bronze Age Italy. Am. J. Phys. Anthropol. 139, 146–153 (2009).PubMed 

Google Scholar 
Varalli, A., Moggi-Cecchi, J., Moroni, A. & Goude, G. Dietary variability during bronze age in central Italy: First results. Int. J. Osteoarchaeol. 26, 431–446 (2016).
Google Scholar 
Goude, G., Rey, L., Toulemonde, F., Cervel, M. & Rottier, S. Dietary changes and millet consumption in northern France at the end of Prehistory: Evidence from archaeobotanical and stable isotope data. Environ. Archaeol. 22, 268–282 (2017).
Google Scholar 
Fernández-Crespo, T., Ordoño, J., Llanos, A. & Schulting, R. J. Make a desert and call it peace: Massacre at the Iberian Iron Age village of La Hoya. Antiquity 94, 1245–1262 (2020).
Google Scholar 
Lu, H. et al. Earliest domestication of common millet (Panicum miliaceum) in East Asia extended to 10,000 years ago. Proc. Natl. Acad. Sci. USA 106, 7367–7372 (2009).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Li, M. et al. Starch grains from dental calculus reveal ancient plant foodstuffs at Chenqimogou site, Gansu Province. Sci. China Earth Sci. 53, 694–699 (2010).ADS 
CAS 

Google Scholar 
Zhang, J., Lu, H., Wu, N., Yang, X. & Diao, X. Phytolith analysis for differentiating between foxtail millet (Setaria italica) and Green Foxtail (Setaria viridis). PLoS ONE 6, e19726 (2011).ADS 
CAS 
PubMed 
PubMed Central 

Google Scholar 
Ge, Y. et al. Phytolith analysis for the identification of barnyard millet (Echinochloa sp.) and its implications. Archaeol. Anthropol. Sci. 10, 61–73 (2018).
Google Scholar 
Tao, D., Zhang, G., Zhou, Y. & Zhao, H. Investigating wheat consumption based on multiple evidences: Stable isotope analysis on human bone and starch grain analysis on dental calculus of humans from the Laodaojing cemetery, Central Plains, China. Int. J. Osteoarchaeol. 30, 594–606 (2020).
Google Scholar 
Bucchi, A., Burguet-Coca, A., Expósito, I., Aceituno-Bocanegra, F. J. & Lozano, M. Comparisons between methods for analyzing dental calculus samples from El Mirador cave (Sierra de Atapuerca, Spain). Archaeol. Anthropol. Sci. 11, 6305–6314 (2019).
Google Scholar 
Cristiani, E. et al. Wild cereal grain consumption among Early Holocene foragers of the Balkans predates the arrival of agriculture. Elife 10, 1–37 (2021).
Google Scholar 
Bocanegra, F. J. A. & Sáez, J. A. L. Caracterización morfológica de almidones de los géneros Triticum y Hordeum en la Península Ibérica. Trabprehist 69, 332–348 (2012).
Google Scholar 
Hardy, K., Buckley, S. & Copeland, L. Pleistocene dental calculus: Recovering information on Paleolithic food items, medicines, paleoenvironment and microbes. Evol. Anthropol. 27, 234–246 (2018).PubMed 

Google Scholar 
López-Dóriga, I. In The use of plants during the Mesolithic and the Neolithic in the Atlantic coast of the Iberian peninsula. (Tesis Doctoral Inédita, Universidad de Cantabria, 2016).Nava, A. et al. Multipronged dental analyses reveal dietary differences in last foragers and first farmers at Grotta Continenza, central Italy (15,500–7000 BP). Sci. Rep. 11, 1–14 (2021).
Google Scholar 
Pyankov, V. I., Ziegler, H., Akhani, H., Deigele, C. & Lüttge, U. European plants with C4 photosynthesis: Geographical and taxonomic distribution and relations to climate parameters. Bot. J. Linn. Soc. 163, 283–304 (2010).
Google Scholar 
Zapata, L. In La explotación de los recursos vegetales y el origen de la agricultura en el País Vasco. (Tesis Doctora Inédita, Universidad del País Vasco, 2002).Figueiral, I., de-Jesus-Sanches, M. & Cardoso, J. L. Crasto de Palheiros (Murça, NE Portugal, 3rd – 1st millennium BC): From archaeological remains to ordinary life. Estudos Quat. 17, 13–28 (2017).
Google Scholar 
Bettencourt, A. M. S. O povoado da Idade do Bronze da Sola, Braga, norte de Portugal. Cadernos Arqueol. 9, 29–44 (2000).
Google Scholar 
Jesus, A., Tereso, J. P. & Gaspar, R. Interpretative trajectories towards the understanding of negative features using Terraço das Laranjeiras Bronze Age site as a case study. J. Archaeol. Sci. Rep. 30, 1–14 (2020).
Google Scholar 
Alonso-Martínez, N. Registro arqueobotánico de Cataluña occidental durante el II y I milenio a.n.e.. Complutum 11, 221–238 (2000).
Google Scholar 
Tarongi-Chavarri, M. Análisis comparativo de los estudios carpológicos de la Depresión del Ebro durante el III y I milenio a. C. Un estado de la cuestión. Rev. d’arqueologia Ponent 27, 41–59 (2017).
Google Scholar 
Stika, H.-P. & Heiss, A. G. Plant cultivation in the Bronze Age. In The Oxford Handbook of the European Bronze Age (eds. Fokkens, H. & Harding, A.) 348–369 (2013).González-y-Fernández-Valles, J. M. Temas de toponimia asturiana. Archivum 21, 121–140 (1971).
Google Scholar 
de Carvallo, L. A. In Antiguedades y Cosas Memorables del Principado de Asturias. (Julian de Paredes, 1695).MacKinnon, A. T., Passalacqua, N. V. & Bartelink, E. J. Exploring diet and status in the Medieval and Modern periods of Asturias, Spain, using stable isotopes from bone collagen. Archaeol. Anthropol. Sci. 11, 3837–3855. https://doi.org/10.1007/s12520-019-00819-2 (2019).Article 

Google Scholar 
Renfrew, J. M. Palaeoethnobotany: The Prehistoric Food Plants of the Near East and Europe (Columbia University Press, 1973).
Google Scholar 
Bronk Ramsey, C. Bayesian Analysis of radiocarbon dates. Radiocarbon 51, 337–360 (2009).
Google Scholar 
Reimer, P. J. et al. The IntCal20 Northern hemisphere radiocarbon age calibration curve (0–55 cal kBP). Radiocarbon 62, 725–757 (2020).CAS 

Google Scholar 
Richards, M. P. & Hedges, R. E. M. Stable isotope evidence for similarities in the types of marine foods used by late mesolithic humans at sites along the Atlantic Coast of Europe. J. Archaeol. Sci. 26, 717–722 (1999).
Google Scholar 
Sabin, S. & James, A. In Dental Calculus Field-Sampling Protocol (Sabin version) v2 (protocols.io.bqecmtaw). (2020). https://doi.org/10.17504/protocols.io.bqecmtaw.Cristiani, E. et al. Dental calculus and isotopes provide direct evidence of fish and plant consumption in Mesolithic Mediterranean. Sci. Rep. 8, 8147 (2018).ADS 
PubMed 
PubMed Central 

Google Scholar 
Fiorin, E. et al. Combining dental calculus with isotope analysis in the Alps: New evidence from the Roman and medieval cemeteries of Lamon. Italy. Quat. Int. https://doi.org/10.1016/j.quaint.2021.11.022 (2021).Article 

Google Scholar  More