GoKawiilStudy area
We delimited our study area of the Amazon basin following the biogeographical limits proposed by the Amazon Network of Georeferenced Socio-Environmental Information (RAISG)37.
Plants used in Amazonia
To understand how climate change may affect the biocultural heritage of Amazonia, we first compiled bibliographic information on the plants recorded as utilized by Indigenous and non-Indigenous local communities (Supplementary Table 1). We gathered five different types of information: (1) regional compilations for the Amazon basin16, Northwest Amazonia13 and the Guianas38; (2) country-level compilations, which exist for Ecuador39, Peru40 and Brazil41; (3) monographs on individual Indigenous groups (for example, refs. 42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68); (4) specific plant services, such as food (for example, refs. 69,70,71) and medicine (for example, refs. 72,73,74,75,76,77,78,79); and (5) early historical accounts, going back to the sixteenth century (for example, refs. 80,81,82,83,84). We defined a ‘use report’ as the concatenation of a plant species with plant part, use category and use subcategory (for example, Clusia schultesii Maguire + entire leaf + medicine + skin and subcutaneous tissue). We classified each use report as either ‘unique’ or ‘shared’ based on the cultural groups that cited it. Use reports linked to a single language or a single specified cultural group were considered unique, whereas those cited by multiple languages or groups—including cases labelled ‘not specified, but more than one’, or ‘not specified’ occurring across multiple countries or references—were classified as shared. Together, our review represents the most comprehensive ethnobotanical assessment in Amazonia, covering 50% of its Indigenous cultures. As more field ethnobotanical studies are conducted, the number of studied cultures will continue to increase, as will the number of utilized species, because the ethnobotany of many Amazonian Indigenous groups remains superficially studied85. To address the latter, we focused our local-scale analyses on only the best-studied cultures (see below).
Harmonization of scientific names
Three steps were followed to harmonize the scientific names of utilized species and to confirm their native status in Amazonia: first, a checklist of Amazonian vascular seed plants was built (hereafter ‘Amazon checklist’) by combining the tree species of ter Steege et al.15 and the non-tree species of Cardoso et al.14 (‘tree.10 cm.DBH = no’). After removing duplicates (n = 1,123), there were 16,219 vascular seed plant species. These were reviewed using the R package rWCVP (v1.2.4)86, yielding 14,293 accepted names. Second, all non-accepted names from the rWCVP analysis (16,219 − 14,293 = 1,926; that is, illegitimate, invalid, misapplied, orthographic, synonym, unplaced and not available (NA)) were checked against Plants of the World Online (POWO; https://www.plantsoftheworldonline.org) and the online Catalogue of Vascular Plants of the Americas (VPA)87. This verification resulted in two lists, depending on the online synonymy portal that was consulted: POWO (15,638 species) or VPA (15,806 species). Third, as the Amazon checklist only includes seed plants, the utilized species list (hereafter ‘ethnobotany list’) was filtered for seed plants. This filtered list was then compared with the POWO (89% match rate) and VPA (90%) lists. All non-matching names were double checked by consulting online type specimens at JSTOR Global Plants (https://plants.jstor.org), specimen occurrences at the Global Biodiversity Information Facility (GBIF; https://www.gbif.org) and synonymy in the VPA and POWO online portals. Most of the species missed in the ethnobotany list (82 of 87 species) were added to it, except for five non-native species: Colocasia esculenta (L.) Schott, Cyperus rotundus L., Nymphoides indica (L.) Kuntze, Portulaca oleracea L. and Psidium guajava L. Of the names in the ethnobotany list not listed in the Amazon checklist, 407 were native species missed by the Amazon checklist (and added to it), 153 were synonymy variants present in both lists and 145 were non-native species (removed from the ethnobotany list). Ultimately, the Amazon seed plant checklist (Supplementary Table 2) had 15,585 or 15,749 species according to POWO and VPA, respectively, whereas the utilized seed plant list had 5,679 species. These totals were used to calculate the proportion of utilized species in the Amazon seed plant flora. Non-seed plant names were checked against World Ferns, the most up-to-date synonymic checklist of the ferns and lycophytes of the world88.
Indigenous language names
Indigenous group names and their spoken languages and geographical coordinates were verified using Ethnologue22 and Glottolog25. Classification of language endangerment followed the Ethnologue and was compared with Glottolog as the latter tends to classify more languages as endangered. The Ethnologue classifies language endangerment based on the Expanded Graded Intergenerational Disruption Scale89. The languages in our sample were classified into 11 of the 13 levels of the Expanded Graded Intergenerational Disruption Scale: national, wider communication, educational, developing, vigorous, threatened, shifting, moribund, nearly extinct, dormant and extinct. Of these levels, the first five are not endangered. The Glottolog uses the agglomerated endangerment status (derived from The Catalog of Endangered Languages, UNESCO Atlas of the World’s Languages in Danger and Ethnologue) and has six levels: not endangered, threatened, shifting, moribund, nearly extinct and extinct.
Species range models
All SDMs were implemented through the R package wallace 2 (ref. 90) (Supplementary Note 1). We obtained primary species occurrence records through automated programmatic queries to the Botanical Information and Ecology Network (BIEN) using the R packages BIEN (v1.2.8)91 and spocc (v1.2.4)92, which provides a unified interface to major biodiversity repositories including BIEN, GBIF, VertNet, Ocean Biodiversity Information System (OBIS) and others. For each species in the study list, we executed a standardized observation record retrieval routine in wallace 2, specifying BIEN as the source database. BIEN aggregates curated plant occurrence records from herbaria, plot networks and digitization initiatives worldwide, and supports taxonomically resolved queries suitable for large-scale biodiversity analyses92. Returned records were parsed and cleaned automatically by the wallace 2 internal cleaning step, which removes malformed coordinates, impossible values and duplicated entries. For transparency and reproducibility, all cleaned occurrence tables were exported as comma-separated values files and stored by species. SDMs were produced for species with more than ten unique occurrence records in the list of Amazonian useful plants. In addition, modelled plant species with more than 20% of the total modelled range in the Amazon basin were selected as the control group25. The final number of species used in this analysis was 4,509 Amazonian utilized plant species and 3,920 control group species, for a combined total of 8,429 species.
... continue reading