Ophiocordyceps sinensis is one of the best-known traditional Chinese medicines with distribution confined to the Tibetan Plateau and its surrounding regions. Harvesting the fungus contributes greatly to the livelihood of local communities. The quality and price varies amongst different production regions, usually resulting in an intentional mix-up of its production locality during trading processes, which leads to a demand of developing a reliable way that can trace the geographical origin of this fungus. In the present study, a DNA barcoding-based method applying two universal DNA barcodes for identifying fungal and insect, respectively i.e. the nuclear ribosomal internal transcribed spacer (ITS) and the mitochondrial cytochrome oxidase I (COI), was evaluated and used for geographical origin authentication of O. sinensis. A total of 24 ITS and 78 COI haplotypes were recognised from 215 individuals collected from 75 different geographic localities (county level). Ninety-nine haplotypes were defined using the combination of ITS and COI, discriminating the 75 investigated production counties into 99 distinct regions. A “core” production region was recognised which covers areas of Nagqu and Qamdo in Xizang, Yushu and Guoluo in Qinghai, Gannan (Maqu and Xiahe) in Gansu and certain regions in Nyingch (Bomi and Zayü) and Lhasa (Damxung) in Xizang and Garzê (Sêrxü) in Sichuan Province. Haplotype analyses using the combined barcodes of ITS and COI showed an excellent performance in the geographical origin authentication of O. sinensis and the definition of “core” and “non-core” production region.

Barcoding, biodiversity, COI, ITS, Tibetan Plateau

Ophiocordyceps sinensis (Berk.) G.H. Sung, J.M. Sung, Hywel-Jones & Spatafora [≡ Cordyceps sinensis (Berk.) Sacc.] (Sung et al. 2007) is a fungus that parasitises underground dwelling larvae of moths (Lepidoptera), especially Thitarodes species (Wang and Yao 2011). The fungus has been traditionally used as a tonic for almost 2000 years (Pegler et al. 1994), commonly for the treatment of kidney and lung problems (Committee of Pharmacopeia, Chinese Ministry of Health 1964, 2005). Recent studies have shown that it possesses various pharmacological effects, including antioxidant (Dong and Yao 2008), anti-aging (Jian et al. 2018), anti-tumour (Wu et al. 2007; Ji et al. 2021), immunomodulating (Kuo et al. 1996; Kuo et al. 2007; Wu et al. 2014; Chen et al. 2022), hypoglycaemic (Zhang et al. 2006), hypotensive and vasorelaxant activities (Chiou et al. 2000).

Due to the increasing demand and the significant decrease in wild resources, the price of natural O. sinensis products has increased sharply during the past three decades. It has become the most precious fungus worldwide and the top-rated products are even sold at a price higher than gold. Collecting and trading the fungus has become one of the most important income sources for local communities in production areas (Winkler 2008; Cannon et al. 2009; Weckerle et al. 2010; Bohra et al. 2014; Shrestha and Bawa 2014). The species also plays an indispensable role in maintaining the stability of the ecosystem of the Tibetan Plateau (Li et al. 2021b). However, the recent global climate change and over-harvesting have resulted in an apparent habitat loss of the species (Yan et al. 2017; Hopping et al. 2018). The fungus has, thus, been assessed as vulnerable (UV) by the International Union for Conservation of Nature and Natural Resources (Yang 2020), the national Red List of macrofungi in China (Yao et al. 2020) and the Red List of biodiversity of Yunnan Province of China (https://www.cas.cn/yx/201705/t20170522_4602412.shtml). The fungus has also been listed as endangered under the second class of state protection by Chinese government since 1999 (State Forestry Administration and Ministry of Agriculture 1999). Different management strategies have been developed for better utilisation and conservation of the natural resource of this fungus, both in China and in other Himalayan countries (Childs and Choedup 2014; Li et al. 2021b).

Due to the host specificity and the limitations of its living conditions, O. sinensis is confined to the alpine shrub and alpine meadow areas on the Tibetan Plateau and its surrounding regions, usually found at altitudes from above 3,000 m to the snow line (Li et al. 2011). The species was reported from Qinghai, Xizang, Sichuan, Gansu and Yunnan in China and the Himalayan countries of Nepal, Bhutan and India (Li et al. 2011). The price of natural O. sinensis varies greatly due to its quality variance. Products with the highest quality produced from Nagqu, Xizang (1100–1300 individuals per kg) is selling at a price of 860,000 RMB (~ US$ 117,800) per kg at present, while low-quality products from the same area (~ 5000 individuals per kg) can be purchased at 165,000 RMB (~ US$ 22,600) per kg at local markets (information were from a middleman who sells caterpillar fungus in Lhasa, Xizang on 30 Sept 2023). The top-rated products (~ 2000 individuals per kg) were even valued up to 998,000 RMB (~ US$ 145,000) per kg in 2017 by Tongrentang Pharmaceutical Co., the most well-known pharmacy shop that sells traditional Chinese medicine (Li et al. 2021b).

Although significant genetic variance was observed within this species (Zhang et al. 2009; Quan et al. 2014a, b; Dai et al. 2020) which would probably result in intrinsic differences, the quality and price of O. sinensis is mainly graded according to their extrinsic properties, including size, colour, smell and maturity. Those properties are usually considered as highly relevant to their geographic origin. Specimens from Nagqu in Xizang, Yushu in Qinghai and certain other restricted areas, for example, Gyaca County in Shannan, Xizang and Darlag County in Guoluo, Qinghai, are generally considered to be of higher quality and are much more expensive than those from other areas. Those areas were widely recognised as “core” production regions by traders and the general consuming public. Individuals from the core region are usually larger in size compared with the non-core region and the larval host part is firmer with saturated yellow to yellowish-brown colour. The intrinsic quality of the core region is also higher, exhibiting lower levels of heavy metals, such as lead (Pb), cadmium (Cd) and arsenic (As) (Wang et al. 2014) and higher levels of pharmacologically active substances comparing with individuals from non-core regions (unpublished data). Since specimens of different production regions are hardly being distinguished (if not impossible) from their appearance, it is pretty common that unscrupulous traders would mix individuals from other areas and label the mixture as produced from a deceptive better core area such as “Nagqu” or “Yushu”. Undoubtedly, this practice significantly harmed the customers’ benefits and resulted in a loss of consumer trust. The current applied morphology-based methods to discriminate the fungus’ production locality largely rely on the personal experience accumulated over the years. It is difficult for ordinary consumers and market managers to master and may give unreliable identification, making it practically inapplicable, especially when legal disputes are involved. In these circumstances, an accurate identification method of the production locality is obviously needed to standardise the market and protect the consumers’ benefit. Then, this raises an important issue: whether the core and non-core production regions of O. sinensis could be distinguished and whether the specific geographic production localities could be traced.

Up to now, different technologies have been developed for geographical origin authentication for this fungus, including using a near-infrared reflectance (NIR) spectroscopy analysis of the methanol extracts (Wang et al. 2006), a random amplified polymorphic DNA (RAPD) fingerprinting analysis (Qian et al. 2011), a nitrogen and carbon isotope analyses by isotope ratio mass spectrometry (Li et al. 2014), a content ratio analysis of methionine to total amino acids (Shi et al. 2021; Ding et al. 2023), a non-targeted metabolomics from UPLC-QTOF-MS (Wang et al. 2024a) and the mineral element analysis (Wang et al. 2024b). These studies either used limited sampling or were only able to classify the geographical origin to the provincial level. A fine-scaled geographical origin authentication not only calls for a more sensitive method, but also need a better sampling which covers the entire production region.

DNA barcoding is a rapid, accurate and cost-effective species identification method (Hebert et al. 2003). In addition to species identification, it could also provide valuable information for analysing population-level variation in some cases (Hajibabaei et al. 2007; Rach et al. 2008; Craft et al. 2010). DNA barcodes may also contain geographical information since population variances may be driven by geographic isolation and, thus, could be used for geographical origin authentication. The nuclear ribosomal internal transcribed spacer (ITS) and the mitochondrial cytochrome oxidase I (COI) are the two universal DNA barcodes for identifying fungal (Schoch et al. 2012) and insects (Hebert et al. 2003) species, respectively. Earlier studies have shown that the intraspecific ITS variation is relatively small in O. sinensis (Kinjo and Zang 2001) or so-called “highly homologous” regardless of geographical origin (Chen et al. 2004), very limited ITS haplotypes have been recognised in these studies. However, significant genetic divergences and more ITS haplotypes have been identified within the species when larger sampling sizes were applied in a number of more recent studies (Zhang et al. 2009; Quan et al. 2014b; Dai et al. 2020). While the ITS sequences are not sufficient for geographical origin authentication to an ideal level, for instance, to the county level, other molecular markers are required to increase the discriminability. The diversity of the host insects was far greater than the fungus itself as revealed by COI and other mitochondrial DNA fragments, such as cytochrome c oxidase subunit II (COII) and cytochrome b (Cytb) (Quan et al. 2014a, b; Zhang et al. 2014; Dai et al. 2019). At least fifty-seven species in seven genera of the family Hepialidae have been recognised as potential hosts of the fungus, according to a literature review (Wang and Yao 2011). Mitochondrial DNA fragments from the host insects thus would probably give more information on geographical origin authentication than the fungal ITS.

Since DNA from both the fungus and its host insects can be obtained from single individuals, a better resolution would be achieved if the combined ITS and COI data were used rather than using a single barcode. In this study, the ITS and COI sequences were obtained from a total of 215 sample individuals collected from 75 different counties of five production provinces and haplotypes were defined and used for geographical origin authentication of this precious fungus. The results would not only provide valuable information for geographical origin authentication, but also benefit conservation of this species.

The qualities of traditional Chinese medicine from different producing regions vary due to the differences in effective medicinal components. Environmental factors such as climate, soil, biology and topography may significantly impact the growth and formation of medicinal materials, resulting in variance in quality amongst production regions. Besides, the intrinsic genetic differences amongst geographic populations may also be critical for the quality and the ‘geo-herbalism’ of medical materials. The quality of the Chinese caterpillar fungus was mainly evaluated by its present appearance, primarily the specimen’s size, colour and smell. The geographic production regions were usually supposed to be highly related to those extrinsic characteristics, causing an intentional mislabelling of the production locality during the trading process and a strong desire for the authentication of its geographical origin as a result.

Using the combination of the universal DNA barcodes for fungi (ITS) and host insects (COI), 99 combined haplotypes were recognised, which discriminated 75 investigated production counties into 99 regions. If administrative regionalisation were considered, the combined haplotypes could determine the production region to provincial level in nearly all cases with only a very few exceptions observed, i.e. Xiahe County of Gannan, Gansu shared the same haplotype F01H03 with Maqên County of Guoluo, Qinghai; Minle County of Zhangye, Gansu shared haplotype F02H01 with seven counties around Qinghai Lake (Datong, Gangca, Gonghe, Huzhu, Menyuan, Qilian, Tianjun); Zadoi County of Yushu, Qinghai shared haplotype F01H04 with the other nine Xizang counties (Baxoi, Biru, Bomi, Dêngqên, Seni, Nyainrong, Riwoqê, Sog, Zayü) (Table 1). Those exceptions are geographically close to each other or their shared haplotypes are widespread, representing the main populations of the fungus. The combined haplotypes also showed ability to discriminate production regions to the prefectural level. The confusing areas that cannot be distinguished to a prefectural level include Guoluo and Yushu in Qinghai, Prefectures of Hainan, Haixi, Haidong, Haibei and Xining around Qinghai Lake (shared the haplotype F02H01) and adjacent areas of Nagqu, Qamdo and Nyingch Prefectures (Table 1; Fig. 3).

Amongst all the 75 counties investigated in this study, 39 (52%) possessed a unique combined haplotype, which means samples from those counties could be traced; 18 (24%) counties contained both unique and shared haplotypes, those counties possibly being traced if the detected samples had the unique haplotypes, or not able to be traced if shared haplotypes were seen; the remaining 18 (24%) shared haplotypes with other counties, in other words, samples from those counties could not be accurately traced to a particular county, but to a corresponding area. It is noteworthy that all counties of Yunnan (2) and Sichuan (13) possessed unique haplotypes, the number of haplotypes ranging from 1 to 5, depending on the sample sizes (Table 1). The production region of the two provinces is characterised by its remarkable topographic of steep ridges and deep valleys of the Hengduan Mountains. This region is recognised as one of the Earth’s 34 biodiversity hotspots (Mittermeier et al. 2004) and has been reported as a diversity centre of the fungus O. sinensis and its host insects (Zhang et al. 2009; Quan et al. 2014a, b; Dai et al. 2019; Dai et al. 2020).

Up to now, a total of 32 ITS haplotypes have been identified from our own collections, amongst which eight ITS haplotypes (Li et al. 2021a) were not included in this study because we could not amplify COI sequences from the corresponding specimens. In contrast, Zhang et al. (2009) defined eight ITS haplotypes from 56 isolates across 10 populations. A different criterion has been used in their study which could affect the haplotype count. A more recent study by Dai et al. (2020) identified as many as 111 ITS haplotypes from 948 individuals across 96 sampled populations. The explosion of the ITS haplotype number was partly contributed by their extensive sampling size and also caused by the inclusion of numerous un-trimmed sequences from GenBank. However, some GenBank sequences may contain errors introduced during PCR or sequencing. Additionally, 18 haplotypes (Hap77–Hap90, Hap95, Hap96, Hap98 and Hap99) identified as ‘Clade VII’ in the study by Dai et al. (2020) were likely misidentified and may represent another distinct species O. laojunshanensis (Chen et al. 2011). Furthermore, Lu (2022) suggested that the haplotypes in ‘Clade VI’ and ‘Clade VIII’ were probably not from O. sinensis, but from other closely-related yet undescribed species. It is noteworthy that pseudogenic ITS sequences were not considered and included in this study. Those pseudogenes would contribute far more haplotype numbers than the functional ones according to our former studies (Li et al. 2013, 2020). ITS pseudogenes may provide valuable information in analysing species diversification and dispersal and also in authentication of geographic origin of this fungus.

In the present study, 215 individuals (specimens) from 75 counties were included, covering nearly half of the confirmed (113) and possible (55) distribution sites at the county level (Li et al. 2011). Additionally and more importantly, those specimens were all collected with detailed and credible locality information recorded and all samples were sequenced in both directions and assembled, with all ambiguous base pairs manually checked and corrected by checking the sequencing chromatograms. Sequences deposited in GenBank were not used in this study to ensure the data credibility, since single mutations could yield different haplotypes and the geographical origin authentication largely relies on the solid background database. Even though a comprehensive sampling programme was carried out in this study, collections from the remaining un-sampled regions, especially those non-core areas of Sichuan, Yunnan and south-eastern Xizang, will facilitate studies of genetic diversity, evolutionary dynamics and geographical origin authentication of this species. In addition, the sampling density to the core regions like Yushu in Qinghai, Nagqu and Qamdo in Xizang must also be improved for a better resolution. It should also be noted that the present study failed to include samples from the southern Himalayan countries of Nepal, Bhutan and India, the genetic diversity of those areas remaining largely unknown. A better picture of the core and non-core production delimiting and a more precise authentication of the geographic origin would be achieved if sampling sizes and coverage to those areas increased.

This study indicates that the host insects are more diverged than the fungus, with a total number of 78 COI haplotypes being recognised. Host insect’s diversity and complexity could be attributed to their unique life history and habitat isolation. The adult ghost moths cannot fly long distances and survive only 3–8 days (Yang et al. 1996). The complex topography resulting from the uplift of the Tibetan Plateau also limited the gene flow amongst populations (Wang et al. 2008). The host insects of O. sinensis showed complex vertical and regional distribution patterns on the plateau; different host species usually occupy different mountain ranges or even different sides and/or altitudes of the same mountain (Liu et al. 2005). Studying its host insects would give a better understanding of the speciation and diversification of this fungus, the co-evolution relationship between insects and their fungal associates and achieve a better resolution of defining the “core” and “non-core” production regions. Extensive insect specimens from different regions of the whole production area must be collected, identified and/or described by the taxonomists rather than just analysing sequences of one or certain more DNA fragments.

The price of O. sinensis varies greatly amongst different production regions even with the same exterior quality, while natural products from different localities are hardly being distinguished from their appearance, resulting in confusion during trading processes. It leads to the demand of a reliable way to trace the geographical origin of this fungus. The present study developed a DNA barcoding-based method which uses ITS and COI, i.e. two universal DNA barcodes for identifying fungal and insect species, respectively. As many as 24 ITS and 78 COI haplotypes were recognised from 215 individuals that were collected from 75 different geographic localities (county level) and ninety-nine combined haplotypes were defined using both ITS and COI. The combined haplotype analysis showed an excellent performance in the geographical origin authentication of the fungus, discriminating the 75 investigated production counties into 99 distinct regions. Additionally, haplotype analysis was also found capable to define the “core” and “non-core” production regions.

Y.L. is grateful to Caidanzhuoma and Gang Zhang in Nagqu Inspection and Testing Center for providing assistant in specimen collecting.