Introduction
The Himalayan region is a globally important biodiversity hotspot (Myers et al., Reference Myers, Mittermeier, Mittermeier, da Fonseca and Kent2000), but its ecology is fragile (Valdiya, Reference Valdiya1984, Reference Valdiya2016). The area is prone to natural disasters (Pathak, Reference Pathak2016) such as earthquakes (Dal Zilio et al., Reference Dal Zilio, Hetényi, Hubbard and Bollinger2021) and subsequent landslides that can lead to habitat and vegetation clearance. In addition, human population growth and anthropogenic activities are now affecting not only Himalayan ecology and vegetation but also the geomorphology (BBC News, Reference News2023). The melting of Himalayan glaciers (Xu et al., Reference Xu, Grumbine, Shrestha, Eriksson, Yang, Wang and Wilkes2009) and bursting of glacial lakes (Veh et al., Reference Veh, Korup and Walz2020) have put further pressure on the region's ecosystems, increasing the need to study the species and ecology of this area.
The genus Buddleja (Scrophulariaceae) encompasses c. 90 species and has a wide distribution across the tropical, subtropical and warm-temperate regions of Africa, Asia and the Americas (Li & Leeuwenberg, Reference Li, Leeuwenberg, Wu and Raven1996; Chau, Reference Chau2017; Norman, Reference Norman2000). The Sino-Himalayan region in Southeast Asia is a centre of diversity for the genus and harbours 25 of the 27 Asian Buddleja species (excluding only Buddleja curviflora and Buddleja japonica; Wu et al., Reference Wu, Sun, Zhou, Li and Peng2010). Many Buddleja species are notable for their aesthetic appeal (Chen et al., Reference Chen, Gong, Ge, Dunn and Sun2012, Reference Chen, Gong, Ge, Dunn and Sun2014; Zhang, Reference Zhang2020), medicinal properties (Backhouse et al., Reference Backhouse, Rosales, Apablaza, Goïty, Erazo and Negrete2008; Khan et al., Reference Khan, Ullah and Zhang2019; Yang et al., Reference Yang, Luo, Guo, Liu and Li2023b), cultural significance (Namsa et al., Reference Namsa, Mandal, Tangjang and Mandal2011; Li et al., Reference Li, Zhang, Guo, Yang and Wang2020) and ecological role (Gong et al., Reference Gong, Chen, Vereecken, Dunn, Ma and Sun2015; Verbeke et al., Reference Verbeke, Boeraeve, Carpentier, Jacquemyn and Pozo2023).
During several years of fieldwork it became evident to us that some Himalayan Buddleja species, notably Buddleja colvilei, Buddleja sessilifolia, Buddleja delavayi and Buddleja yunnanensis, are facing threats from anthropogenic disturbance and environmental pressures. These species share distinctive features such as a limited distribution range, a preference for high-altitude habitats and a susceptibility to human interference, all of which make them particularly vulnerable to changes in their environment. The four species have not yet been assessed for the IUCN Red List of Threatened Species (IUCN, 2023a). To date, Buddleja bhutanica is the only Asian species of Buddleja that has been assessed, and is categorized as Vulnerable (Bhutan Endemic Flowering Plants Workshop, Reference .2017). Buddleja colvilei, B. delavayi and B. yunnanensis are included on the Threatened Species List of China's Higher Plants (Qin et al., Reference Qin, Yang, Dong, He, Jia and Zhao2017). Buddleja sessilifolia and B. delavayi are included on the List of Yunnan Protected Plant Species with Extremely Small Populations (Sun, Reference Sun2021) because of their reduced populations, restricted habitats, the presence of severe human disturbance and the heightened risk of extinction (Ren et al., Reference Ren, Zhang, Lu, Liu, Guo and Wang2012; Ma et al., Reference Ma, Chen, Edward Grumbine, Dao, Sun and Guo2013). Further evaluation is required to assess these four species for inclusion on the IUCN Red List.
Buddleja colvilei Hook.f. & Thomson is a shrub or small tree characterized by its attractive foliage and flowers, and was referred to as ‘the handsomest of all Himalayan shrubs’ by Hooker in 1849 (Stuart, Reference Stuart2006). It is endemic to the eastern Himalayas of Nepal, India (Sikkim), Bhutan and China (Tibet) at 1,600–4,200 m elevation.
Buddleja sessilifolia B.S. Sun ex S.Y. Pao is a perennial shrub with sessile leaves that usually grows in dense clusters beneath Himalayan hemlock Tsuga dumosa forests or at the edges of humid valleys. The species was described in 1983 as being conspecific with B. colvilei, a plant distributed throughout the Gaoligong Mountains at 2,600–3,200 m (Ge et al., Reference Ge, Cai, Bi, Chen and Sun2018). In 2018, B. sessilifolia was delineated as a distinct species (Ge et al., Reference Ge, Cai, Bi, Chen and Sun2018), suggesting that the disparity in the morphological and molecular characters of these two species was caused by differences in ploidy levels or speciation processes during the uplift of the Himalayas (Yang et al., Reference Yang, Ge, Guo, Olmstead and Sun2023a).
Buddleja delavayi Gagnepain is a perennial shrub characterized by two distinct colour polymorphisms that exhibit purple or white flowers (Chen et al., Reference Chen, Gong, Ge, Niu, Zhang, Dunn and Sun2015). This species is distributed in sparse montane forests and moist broad-leaved forests on hillsides in Dali, Yunnan, at 2,000–3,000 m.
Buddleja yunnanensis Gagnepain, which is a narrowly endemic species, is limited to Pu'er and Xishuangbanna, Yunnan, China. It is distributed on the edges of shrub forests on mountains or hillsides at 1,000–2,500 m (Li & Leeuwenberg, Reference Li, Leeuwenberg, Wu and Raven1996).
Data-driven analyses can illuminate the interplay of variables that affect a species’ conservation status, but a lack of reliable data, obsolete information and inconsistencies can limit the accuracy and reliability of conservation assessments (Zizka et al., Reference Zizka, Silvestro, Vitt and Knight2021). Automated assessments have the potential to mitigate these issues by creating standardized evaluations or by prioritizing species for reassessment (Cazalis et al., Reference Cazalis, Di Marco, Butchart, Akçakaya, González-Suárez and Meyer2022; de Caetano et al., Reference de Caetano, Chapple, Grenyer, Raz, Rosenblatt and Tingley2022). Two approaches that have been used for automated conservation assessment are: (1) the criteria-explicit approach, which uses automatic calculation of the IUCN Red List parameters from distribution data, using tools such as GeoCAT and rCAT (Bachman et al., Reference Bachman, Moat, Hill, de la Torre and Scott2011) or packages such as red (Cardoso, Reference Cardoso2017), ConR (Dauby et al., Reference Dauby, Stévart, Droissart, Cosiaux, Deblauwe and Simo-Droissart2017) and Rapid Least Concern (Bachman et al., Reference Bachman, Walker, Barrios, Copeland and Moat2020) in R (R Core Team, 2023), and (2) the category-predictive approach, which employs machine learning such as random forests (Pelletier et al., Reference Pelletier, Carstens, Tank, Sullivan and Espíndola2018) or neural networks (Zizka et al., Reference Zizka, Andermann and Silvestro2022) and generalized linear models (Böhm et al., Reference Böhm, Williams, Bramhall, McMillan, Davidson and Garcia2016). Here we assess the conservation status of the four Buddleja species (Plate 1) as a case study, comparing traditional assessments based on field studies with automated assessments.
(a) Buddleja colvilei, (b) Buddleja sessilifolia, (c) Buddleja delavayi, (d) Buddleja yunnanensis, (e) mature B. colvilei cut down in Ilam, Nepal, (f) loss of habitat of B. sessilifolia caused by debris flow and mudslides in the Gaoligong Mountains, Yunnan, China, (g) B. sessilifolia plantlets in Kunming Botanical Garden, and (h) living collection of B. delavayi in Kunming Botanical Garden. Photos: Fengmao Yang (a,c); Jia Ge (b,d–h).
Methods
Our study area encompasses the trans-Himalayan region, spanning the eastern Himalayas to south-western China (Fig. 1), including eastern Nepal and western China (Yunnan and Tibet). We first collected species occurrence records from the Flora of China (Li & Leeuwenberg, Reference Li, Leeuwenberg, Wu and Raven1996), Herbarium KUN at the Kunming Institute of Botany, Chinese Academy of Sciences, the Chinese Virtual Herbarium (2023), GBIF (2023) and iPlant (2023). Using the distribution information from these sources we selected 12 locations that together represent almost all known sites of the four species. We conducted field studies over 13 years (2010–2022) to obtain comprehensive data on the four species, following the recommendations for field investigations of Plant Species with Extremely Small Populations (Yang & Sun, Reference Yang and Sun2017; Yang et al., Reference Yang, Cai, Liu, Chen, Gratzfeld and Sun2020). Our field study did not include the populations of B. colvilei in Bhutan and India, mainly because of geographical barriers and feasibility constraints. Additionally, we excluded populations of Buddleja heliophila (considered a synonym for B. delavayi) because recent research by Ge et al. (unpubl. data, 2023) has revealed distinct molecular and morphological differences between them.
Distribution of the four Buddleja species across the Himalayan region and south-western China.
During our field surveys of the species and their habitats, we recorded habitat type, geographical coordinates (using a GPS), elevation, extent and impact of any human disturbance, and identified any other potential abiotic or biotic threats, using the IUCN Red List Threats Classification Scheme (IUCN, 2023b). We recorded the number of mature individuals in each location, and resurveyed each location 1–7 years after the first survey (Table 1). The assessment was conducted according to IUCN B and C criteria (IUCN, 2012, 2023b), considering area of occupancy, habitat status, number of mature individuals, and major threats.
Occurrence records of four Buddleja species endemic to the Himalayas, with latitude and longitude, collection years, location descriptors, data sources (Global Biodiversity Information Facility (GBIF), Chinese Virtual Herbarium (CVH) or observation from field surveys), and status (with the number of plants recorded in the respective collection years).
1 Numbers indicate the number of mature plants recorded in the latest survey/first survey.
2 Intial survey year.
3 Year of resurvey.
4 Recorded as Buddleja heliophila.
From each population we collected healthy, mature seeds of 10 individuals with 1–3 infructescences, and recorded collection time and locality. We dried and cleaned the seeds, and deposited seeds of all four species in the National Wild Plant Germplasm Resource Centre at the Kunming Institute of Botany, China. We initiated ex situ conservation in Kunming Botanical Garden through tissue culture, cuttings and cultivation.
We implemented automated conservation assessment using the packages red and ConR in R 4.2.2 (R Core Team, 2023) and GeoCAT (GeoCAT, 2023), which use only occurrence records. We retrieved data from GBIF (2023) using the rgbif package in R (Chamberlain & Boettiger, Reference Chamberlain and Boettiger2017; Chamberlain et al., Reference Chamberlain, Oldoni, Barve, Desmet, Geffert and Mcglinn2024), and cross-checked these data with data from the Chinese Virtual Herbarium (2023). We filtered and cleaned data prior to automatic assessment. We excluded records with no location information or with a location outside of the species’ native ranges according to Plants of the World Online (2023). We removed duplicates and records with misinformation based on knowledge from our field surveys. We crosschecked the remaining records with the Chinese Virtual Herbarium (2023) and Herbarium KUN, and added any missing data where required. We combined the resulting data with the occurrence records from our field surveys.
We estimated the extent of occurrence (the smallest polygon in which no internal angle exceeds 180° and that contains all occurrences) using the convex hull method, and we estimated the area of occupancy (the area within the extent of occurrence that is occupied by a taxon) using 2 × 2 km grid cells (IUCN, 2012). Using ConR we also estimated the number of locations and subpopulations using a sliding grid approach and a circular buffer method, respectively (Rivers et al., Reference Rivers, Bachman, Meagher, Nic Lughadha and Brummitt2010). Using red, ConR and GeoCAT we determined the Red List category for each species based on validated IUCN B1 and/or B2 criteria (IUCN, 2012, 2023b).
Results
In our field surveys, we recorded 68 healthy, mature individuals of B. colvilei (2018), 29 of B. delavayi (2017), > 532 of B. sessilifolia (2015, 2018, 2022) and 18 of B. yunnanensis (2022). After filtering and cleaning the data from a combination of field surveys and data from GBIF and the Chinese Virtual Herbarium, we had 14 unique occurrence records of B. colvilei, eight of B. sessilifolia, six of B. delavayi (including B. heliophila) and seven of B. yunnanensis to use for automated assessment (Table 1). We found that all four Buddleja species had restricted distributions, with relatively small extents of occurrence and areas of occupancy, few occurrence locations and small populations (Table 2). All three automated assessments evaluated B. delavayi and B. yunnanensis as Endangered and B. colvilei and B. sessilifolia as Vulnerable, whereas our assessment based on our field studies is that both B. delavayi and B. yunnanensis should be categorized as Critically Endangered, B. sessilifolia as Endangered and B. colvilei as Vulnerable. The areas of occupancy calculated by the automated assessments were notably different from that of the field assessments, particularly for B. sessilifolia and B. delavayi.
Conservation status of four Buddleja species, as assessed using GeoCAT (Bachman et al., Reference Bachman, Moat, Hill, de la Torre and Scott2011), red (Cardoso, Reference Cardoso2017) and ConR (Dauby et al., Reference Dauby, Stévart, Droissart, Cosiaux, Deblauwe and Simo-Droissart2017), on the Threatened Species List of China's Higher Plants (Qin et al., Reference Qin, Yang, Dong, He, Jia and Zhao2017), and based on our field assessment, with the calculated extent of occurrence (EOO) for each automated assessment and area of occupancy (AOO) for the automated and field assessments. Threat classification is based on IUCN (2023b).
1 Results from ConR only.
2 VU, Vulnerable; EN, Endangered; CR, Critically Endangered; NE, Not Evaluated.
Following the Red List criteria (IUCN, 2012), unique occurrences refer to cleaned and filtered records, subpopulations indicate distinct groups with limited genetic exchange, and locations are areas where a single threat can affect all individuals. The number of locations differs between the automated and traditional assessment methods (Table 2) because the former is based on unique occurrence records whereas the latter is based on direct field investigations.
Buddleja colvilei is widespread, but this species is at risk of extinction because of logging and harvesting of wood (threat classification 5.3.1; IUCN, 2023b) and road construction (4.1). Of 57 B. colvilei plants originally recorded in Goruwale, Ilam, Nepal, we observed that 15 were cut and did not regenerate (Plate 1e). Only 26 were observed in Yadong, Tibet, China, with a 64% decline compared to the number of plants recorded in the original survey.
Although the combined records of B. delavayi demonstrate a large extent of occurrence, during our fieldwork in 2017 we found the species to be facing significant threats. It has a limited area of occupancy of 0.14 km2, with only 29 individuals. Its habitat, which lies alongside a village roadside in Jianchuan County, Dali, appears to be at significant risk from housing and urban areas (threat 1.1) and road construction (4.1). The small number of mature individuals also poses a further severe risk of extinction. Our findings suggest that ex situ conservation is necessary to safeguard this species, and we have already implemented this: after 11 years of cultivation following their initial introduction in 2012, two individuals are thriving in Kunming Botanical Garden (Plate 1h).
The distribution of B. sessilifolia on Gaoligong Mountain reflects the fragmentation and isolated nature of its habitat. We recorded > 768 B. sessilifolia in total during our initial field surveys in 2010, 2015 and 2021. On returning to the sites in 2015, 2018 and 2022, respectively, although three populations remained stable, four had decreased, by 145 (91% decline), 23 (17%), 22 (24%) and 18 (90%) individuals. In addition, the population of 28 individuals in Fugong county had disappeared as a result of road construction (threat 4.1) and debris flow (10.3; Plate 1f). Despite our attempts, ex situ conservation of this species has so far failed because of the difficulty of establishing seedlings (Plate 1g).
Our field assessment of B. yunnanensis revealed that its distribution spans various habitats. However, these habitats are threatened by the development of housing and urban areas (threat 1.1) and shifting agriculture (2.1.1), posing a significant risk of extinction to this species from habitat loss and fragmentation. In Zhenyuan and Ninger counties, Yunnan, we found a small population of 18 individual plants with an area of occupancy of 14.86 km2. This population lies outside any protected areas, in a region potentially prone to threats that would significantly impact survival. Despite our ex situ conservation efforts, the single individual that we had in cultivation died after 5 years.
Both the automated and field study assessments suggest B. colvilei should be categorized as Vulnerable. The automated assessments suggest classification under criteria B1a+B2a because of an extent of occurrence of < 20,000 km2 (B1), area of occupancy of < 2,000 km2 (B2) and documentation in < 10 localities (a). The field assessment further suggests classification under criteria B2ab(iii)c(iv); C2a(i)b, with observed declines (B2b) in area, extent and quality of habitat (iii), extreme fluctuations (c), the presence of < 1,000 mature individuals in locations (iv) and continuing decline (C2a) in the number of mature individuals within each subpopulation (i) along with extreme fluctuations in their numbers (b).
From our field studies, we propose that both B. yunnanensis and B. delavayi are categorized as Critically Endangered based on criteria B2ab(iii)c(iv) because their area of occupancy is < 10 km2 (B2), with a severely fragmented habitat (a), continuous decline observed (b) in area, extent and quality of habitat (iii), and significant fluctuations (c) in the number of locations (iv), with only 29 mature individuals of B. delavayi and 18 of B. yunnanensis recorded.
Also from our field studies, we propose that B. sessilifolia is categorized as Endangered based on criteria B2ab(ii)c(iv) because of its area of occupancy < 500 km2 (B2), fragmented distribution occurring at no more than five locations (Gongshan, Yunnan and Kachin, Myanmar) (a), continuous decline (b) in its area of occupancy (ii), and significant fluctuations (c) in the number of mature individuals (iv) due to multiple threats (4.1 and 10.3).
Discussion
Of the 36 globally significant biodiversity hotspots, three (Eastern Himalayas, Indo-Burma, and the Mountains of Southwest China) are in the Himalayan region (CEPF, 2023) and have a high conservation priority. Circa 45.1% of the world's angiosperm species are potentially under threat (Bachman et al., Reference Bachman, Brown, Leao, Lughadha and Walker2024). Therefore, conservation assessments are critical for the identification of threatened species and for the development of conservation plans, but different assessment methods can provide varying degrees of accuracy and reliability. Automated assessments are based solely on georeferenced data, whereas field studies incorporate additional parameters such as habitat status, number of mature individuals and threats. However, both methods highlighted the precarious status of these four species of Himalayan Buddleja.
The Threatened Species List of China's Higher Plants (Qin et al., Reference Qin, Yang, Dong, He, Jia and Zhao2017) is an expert-driven assessment that covers a large number of species. Although it can be informative, the categorizations may not consider all nuances of the ecology and behaviour of a species and can also be subject to bias. Three of the four Buddleja species we assessed have been categorized on the Threatened Species List, but two of these categorizations, for B. delavayi and B. yunnanensis, are markedly different from our field assessment (Vulnerable as opposed to Critically Endangered). This could be attributed to the limited availability of long-term investigations and the reliance on checklist data from various databases.
Conversely, automated assessment tools are fast, efficient and cost-effective, can provide preliminary conservation assessments for large numbers of species, and can minimize bias and incorporate a wide range of data. The utilization of three tools (GeoCAT, red and ConR), each with unique features, can help to support field investigations, and in our study the estimated extents of occurrence and areas of occupancy (Table 2) were consistent across all tools.
ConR seems to be the most useful of the three tools because it computes the number of geographical locations in which a species is present. This could be a key determinant of the conservation status of a species because it may indicate the level of habitat fragmentation, which may in turn highlight threats facing a particular species. For example, the latest data on B. colvilei revealed a high abundance in Bhutan (Table 1) that we could not ascertain through our field studies, highlighting the importance of field surveys in this region. At the same time, it is important to exercise caution when using open-source databases for conservation assessments as intensified anthropogenic stresses and climate change may cause a decline in species occurrences, affecting the results of assessments. In cross-checking records with herbarium specimens, we found instances of mislabelled specimens. Making automated assessments with mislabelled records could lead to incorrect conclusions, as evidenced by the evaluation of B. delavayi as Endangered rather than Critically Endangered.
Automated assessments may not be as reliable as other methods, however, as they do not consider the number of individuals in a population or specific threats. Field surveys are generally considered the most accurate method for assessing extinction risk (Nic Lughadha et al., Reference Nic Lughadha, Walker, Canteiro, Chadburn, Davis and Hargreaves2018) as they provide detailed information on species’ distribution, population sizes and threats. Anthropogenic activities such as road construction related to the national or local prioritization of development works and population growth must be considered when prioritizing conservation intervention. Our field surveys of B. sessilifolia revealed it is facing a significant risk because its habitat is being fragmented by roads and damaged by mudslides. Similarly, our surveys indicated concern regarding the status of B. yunnanensis, with its limited distribution and a decline in the quality of its habitat resulting in it being catalogued as a Plant Species with an Extremely Small Population. Our field surveys revealed significant threats to these species, highlighting the conservation challenges they face, including habitat degradation, fragmentation and loss, and the negative impacts of anthropogenic activities. However, field surveys can be costly in terms of time and resources, making it difficult to survey large numbers of species.
Although automated assessments are recommended for a preliminary evaluation of conservation status (Zizka et al., Reference Zizka, Silvestro, Vitt and Knight2021), our comparison of methods revealed their limited effectiveness for our focal species. Automated assessments are particularly suitable for species with a wide range of distribution records (Palacio et al., Reference Palacio, Negret, Velásquez-Tibatá and Jacobson2021) and for which there are constraints on field surveys, as in the case of B. colvilei in Bhutan and India. We acknowledge, however, that the paucity of accurate distribution data and taxonomic ambiguity could limit the effectiveness of automated assessment (Mackay-Smith & Roberts, Reference Mackay-Smith and Roberts2019).
Our findings revealed significant issues with digitized information, including identification of errors (e.g. the treatment of B. delavayi and B. heliophila as the same species) and lack of up to date information (e.g. the Fugong population of B. sessilifolia recorded in 2010 that had disappeared by 2015), which may have adversely affected the outcomes of our automated assessments (AuBuchon-Elder et al., Reference AuBuchon-Elder, Minx, Bookout and Kellogg2023). We recommend increasing the digitalization of information for occurrence records, implementing cleaning and filtering processes and establishing user-friendly platforms for data feedback and integration into data repositories such as the Global Biodiversity Information Facility. Only with field exploration and the collection of first-hand data can we fill gaps in the knowledge of species distributions (Nic Lughadha et al., Reference Nic Lughadha, Walker, Canteiro, Chadburn, Davis and Hargreaves2018).
In our case study, field surveys provided the most accurate information for evaluating conservation status, particularly for range-restricted species, because they provided direct and informative data, which is essential when other data are limited or unavailable. Our study therefore highlights the importance of making assessments using data from field surveys when information on a species is limited or unavailable.
Because the Himalayan region is home to many endemic species, a multifaceted approach combining multiple sources of accurate information can contribute to a more comprehensive and reliable species conservation plan than any one approach alone. We recommend first conducting a preliminary assessment using automated assessment methods and then carrying out field surveys for validation and detailed analysis.
Our study also highlights the need for further research on the ecology and conservation biology of threatened species. This could include studies of conservation genetics, responses to different environmental stressors, and interactions with other species, such as pollination biology. (Chen et al., Reference Chen, Gong, Ge, Dunn and Sun2012; Ollerton, Reference Ollerton2017; Coates et al., Reference Coates, Byrne and Moritz2018; Nonić & Šijačić-Nikolić, Reference Nonić, Šijačić-Nikolić, Leal Filho, Azul, Brandli, Özuyar and Wall2019). In addition, scientifically sound conservation efforts are needed to protect such species and their habitats, including the identification of threats, habitat restoration, in situ or ex situ conservation (or both) and public education and outreach (Ma et al., Reference Ma, Chen, Edward Grumbine, Dao, Sun and Guo2013; Volis, Reference Volis2016; Chen & Sun, Reference Chen and Sun2018). Many plant species face serious threats in the Himalayan region, and targeted and comprehensive conservation measures are required. Overall, our study emphasizes the need for a more integrated conservation approach to address the constraints currently hampering species conservation in this region.
Author contributions
Study design: JG, GC; fieldwork: all authors; data analysis: BG, JG; writing: BG, JG.
Acknowledgements
We thank Weibang Sun for supporting our work; Zhilin Dao, Jing Yang, Lei Cai, Lidan Tao, Yinchun Li, Wei Wei and Yunmeng Li (Kunming Institute of Botany) and Pramod Aryal (Central Department of Biotechnology, Tribhuvan University) for assistance with the field investigations; Luo Guifen Luo, Hua Huang, Heli Mao, Hanrun Li, Wang Xi (Kunming Institute of Botany) and Shen Wang for cultivating and caring for the plants; Fengmao Yang (Kunming Institute of Botany) for providing photographs; the anonymous reviewers and editor for their comments and suggestions; and Kunming Botanical Garden, the National Wild Plant Germplasm Resource Center (Kunming Institute of Botany) and the Gongshan Administrative Sub-Bureau of the Gaoligongshan National Nature Reserve for supporting our work. This work was supported by the National Natural Science Foundation of China (grant numbers 32071653, 30970192, 31770240 and 31400478), the Second Tibetan Plateau Scientific Expedition and Research Program (grant number 2019QZKK0502) and Yunnan Fundamental Research Projects (grant number 202001AT070097).
Conflicts of interest
None.
Ethical standards
Investigation and collection of plant materials were with the permission and under the supervision of the relevant local authorities. This research abided by the Oryx guidelines on ethical standards.
Data availability
The authors confirm that the data supporting the findings of this study are openly available via the Global Biodiversity Information Facility at doi.org/10.15468/dl.nkxucp.
Open access
