Ancient plastid genomes solve the tree species mystery of the imperial wood “Nanmu” in the Forbidden City, the largest existing wooden palace complex in the world

Abstract

Exploring the life styles and production methods of past generations using plant resources can help us to improve our understanding of human civilization. Nanmu, known for its high wood quality, was exclusively used for imperial palace construction in the 15th–19th centuries in China, yet its species has been a subject of long‐standing debate.Here, we revisit this unresolved problem, using morphology and ancient DNA (aDNA) to analyze 21 centuries‐old Nanmu specimens sampled from representative palaces of the Forbidden City.Cytochemical staining demonstrated that endogenous aDNA sporadically occurs in the wood ray parenchyma cells of Nanmu specimens. High‐quality plastid genomes were retrieved from archeological woods for the first time via an aDNA capture method, with 90%–100% coverage (137,663–152,805 bp) and sequence depths of 27.05‐ to 1409.94‐fold. Utilizing these ancient genomes, our results demonstrate that Phoebe zhennan and Phoebe hui are most likely the main species of Nanmu in the Forbidden City.This finding diverges from the prevailing view that Nanmu encompasses woods from the whole genus Phoebe and even its close relative Machilus. It also shows that stringent criteria were used when selecting construction materials for the Forbidden City. By combining morphological traits with aDNA analyses, we provide a new solution for identifying the species of timber used for ancient architecture, and we increase our understanding of the way in which forest resources were recognized and utilized by our ancestors despite the lack of a plant taxonomic framework in ancient times.

INTRODUCTION

Interactions among biogeography, forests and human activities become increasingly prominent as human societies develop, especially with respect to human demands for livelihoods and wood production. Wood derived from forest resources runs like a vein throughout human civilization (Glausiusz, 2020). Since ancient times, it has been integral to human activities, from possibly the earliest evidence for woodworking, 1.5-million-year-old hunting weapons (Glausiusz, 2020) to charcoal used in early copper and iron smelting, and production tools, transportation, dwellings, and art objects (Aranguren et al., 2018; Conard et al., 2020). The desire to better understand cultural heritage and preserve it for future generations has long been a major driver in the exploration of wooden relics in the natural and social sciences, and the present study should be seen in this light.

Nanmu is a timber of outstanding reputation due to its high quality, brilliant color, remarkable durability, and medicinal value (Cheng et al., 1992). It was mainly used for columns in palace construction and to make high-value furniture that symbolized the power and status of the aristocracy (Chen et al., 2020). In the 15th until the 19th centuries, Nanmu timber was supplied exclusively to the imperial court with the name “Imperial Wood” (Ding et al., 2019). The largest existing wooden palace complex in the world, the Forbidden City, was originally built during the early Ming Dynasty (1406–1420 AD). However, damaged by fire, the Forbidden City underwent multiple restorations during the Ming and Qing Dynasties (1368–1912 AD). It has been recorded that large-diameter logs of Nanmu were commonly harvested in southern China, particularly in southwestern China (Figure 1), for the construction and maintenance of the Forbidden City (Lan, 1994). However, the species attribution of Nanmu has remained a subject of long-standing debate.

FIGURE 1. The distribution of Nanmu harvested in Ming and Qing Dynasties. The historical distribution record of Nanmu harvested in Ming and Qing Dynasties was mainly derived from Lan (1994), and the distribution of four Phoebe species pointed in this study was from Flora of China. All ranges are approximations aimed at conveying each species' general distribution, rather than detailed, distribution.

The relevant historical records in The Book of Songs and Shan Hai Ching and the unearthed Nanmu cultural relics suggest that Nanmu only refers to the single species Phoebe zhennan (Lin, 1988). However, in The Atlas of Chinese Woods (Cheng et al., 1992), timber from both genera Phoebe Nees and Machilus Nees is called Nanmu, with P. zhennan from Sichuan province as the most important species. Phoebe was listed under the category of Nanmu in the China National Standard Names of Chinese main woods (Cheng et al., 1997), which is in line with Wei (1980). In addition, views on the concept of Nanmu also differ across different geographic regions. In southwestern areas of China, such as Sichuan and Yunnan, P. zhennan, P. hui, and M. nanmu are usually called Nanmu; by contrast. P. bournei is considered to be Nanmu in southern regions like Zhejiang, Fujian, Guangdong, and Guangxi (Li & Wei, 1988). The question of concern, then, is which tree species are represented in the Nanmu structural components of the Forbidden City?

The most frequently used wood identification method relies on anatomical features, either macroscopic or microscopic, and generally allows wood identification at the genus level, or to species groups below genus level (Gasson, 2011; Wheeler & Baas, 1998), but not down to individual species. Molecular identification methods have become feasible tools for breaking through these limitations (Dormontt et al., 2015; Gasson et al., 2021; Jiao et al., 2020), although wood materials present particular challenges in comparison to other tissues from plants. In general, even in standing trees, only a small amount of endogenous DNA is present in the living parenchyma cells in the sapwood (Deguilloux et al., 2002; Jiao et al., 2012), and heartwood extractives often hinder the extraction of endogenous DNA (Lu et al., 2020; Rachmayanti et al., 2009).

Ancient DNA (aDNA) from archeological wood has been investigated since the 1990s, but to a much more limited extent, with only a handful of studies (Dumolin-lapègue et al., 1999; Jiao et al., 2015; Liepelt et al., 2006; Wagner et al., 2018), and their methods generally relied on a methodological framework including the amplification of short PCR fragments or metabarcodes (Lendvay et al., 2018). Under the combined effects of time, environmental factors such as humidity, temperature, and pH value, and biological factors such as fungi and bacteria, the DNA of archeological wood is usually severely degraded, highly fragmented, and contaminated with exogenous DNA (Jiao et al., 2015; Wagner et al., 2018). This makes it difficult to harvest the target DNA of ancient remains through the conventional method of PCR amplification. One possible alternative to PCR amplification is target enrichment via DNA hybridization capture, a well-tested method for obtaining DNA data from bones of ancient humans (Haak et al., 2015; Liu et al., 2021; Wang et al., 2019; Yang et al., 2020), ancient animals (Wen et al., 2022; Zhang, Sun, et al., 2020), and environmental sediments (Massilani et al., 2022; Schulte et al., 2021; Slon et al., 2017; Zhang, Xia, et al., 2020). However, to our knowledge, no studies on aDNA capture from wood remains have reported using this method so far (Orlando et al., 2021; Wagner et al., 2018). Here, we took centuries-old wood specimens from the structural components of the Forbidden City and used an aDNA hybridization capture method to obtain their plastid genomes for the botanical identification of imperial Nanmu.

MATERIALS AND METHODS Samples collections for aDNA analyses

Thin slices of small Nanmu specimens were carefully sampled from dominant structural components of nine representative timber palaces of the Forbidden City, including square columns, beams, and pillars, to avoid visible damage to the structural components of this well-conserved cultural heritage. We randomly selected 21 Nanmu specimens (Figure 2a; Table 1) to perform morphological anatomy identification (Method S1) and aDNA analyses.

FIGURE 2. Morphological characteristics of Nanmu specimens of structural components of the Forbidden City. (a) Twenty-one Nanmu specimens used in this study. (b–d) Microscopic features of Nanmu specimens of structural components of the Forbidden City (take sample GGN03 as an example). Note: V-vessels; R- ray parenchyma cells; C-crystals; * indicates oil cells.

TABLE 1 Detailed information of Nanmu specimens taken from structural components from the Forbidden City in this study

No.	Sample ID	Site	Type of timber component	Completion timea	Main reconstruction timea	Plastid DNA genome
						N rate (%)	Raw reads	Clean reads(Minlength > 25)	Mapping reads(mapping quality > 25)	Mapping rate
1	BHD02	Hall of Preserving Harmony	Square column	1420 AD	1561 AD	6.75	6,068,066	3,131,509	1,620,542	26.71%
2	CNG02	Palace of Compassion and Tranquility	Purlin	1536 AD	1653 AD, 1689 AD, 1751 AD, 1769 AD	5.01	3,740,806	2,157,060	622,937	16.65%
3	CNM01	Gate of Compassion and Tranquility	Pillar	1536 AD	1653 AD, 1689 AD, 1751 AD, 1769 AD	0.84	4,308,040	2,426,204	945,401	21.95%
4	CNM02	Gate of Compassion and Tranquility	Pillar	1536 AD	1653 AD, 1689 AD, 1751 AD, 1769 AD	10.16	4,092,850	2,427,522	427,253	10.44%
5	GGN03	Forbidden City	—	1420 AD	—	0.15	2,187,522	1,515,312	352,951	16.13%
6	GGNXD	Nanxun Hall	—	1420 AD	—	0.49	3,997,360	2,246,291	398,540	9.97%
7	QQG02	Palace of Heavenly Purity	Pillar	1420 AD	1440 AD, 1602 AD, 1644 AD, 1680 AD, 1798 AD	0.22	10,775,152	6,891,932	1,799,444	16.70%
8	THD01	Hall of Supreme Harmony	Beam	1420 AD	1440 AD, 1561 AD, 1627 AD, 1695 AD	1.48	3,405,796	1,779,160	344,704	10.12%
9	THD02	Hall of Supreme Harmony	Square column	1421 AD	1440 AD, 1561 AD, 1627 AD, 1695 AD	2.38	4,564,986	2,352,139	366,552	8.03%
10	THD03	Hall of Supreme Harmony	Square column	1422 AD	1440 AD, 1561 AD, 1627 AD, 1695 AD	0.46	4,600,570	2,510,229	1,023,040	22.24%
11	THD05	Hall of Supreme Harmony	Square column	1423 AD	1440 AD, 1561 AD, 1627 AD, 1695 AD	0.02	5,531,710	2,833,620	2,313,923	41.83%
12	TRG01	Belvedere of Embodying Benevolence	Pillar	1420 AD	1783 AD	0.00	9,869,190	4,956,883	4,821,328	48.85%
13	TRG02	Belvedere of Embodying Benevolence	Pillar	1420 AD	1783 AD	0.12	4,138,756	2,165,059	713,039	17.23%
14	TRG03	Belvedere of Embodying Benevolence	Square column	1420 AD	1783 AD	0.02	8,327,908	4,185,871	3,985,864	47.86%
15	TRG04	Belvedere of Embodying Benevolence	Pillar	1420 AD	1783 AD	0.02	8,693,014	4,363,403	4,143,352	47.66%
16	WM01	Meridian Gate	Pillar	1,420 AD	1647 AD, 1801 AD	1.68	4,508,742	2,449,146	561,946	12.46%
17	WM05	Meridian Gate	Square column	1420 AD	1647 AD, 1801 AD	0.14	6,881,588	3,572,525	901,753	13.10%
18	YHD01	Hall of Exuberance	Beam	1420 AD	1771 AD	0.13	4,712,840	2,835,793	622,857	13.22%
19	YHD02	Hall of Exuberance	Square column	1420 AD	1771 AD	0.35	4,043,542	2,449,792	434,119	10.74%
20	ZHD05	Hall of Central Harmony	Pillar	1420 AD	1561 AD	3.65	5,385,720	2,722,972	524,087	9.73%
21	ZHD18	Hall of Central Harmony	—	1420 AD	1561 AD	9.91	4,545,606	2,481,894	456,741	10.05%

Note: “—” indicates no information available.

^aFrom The Palace Museum (www.dpm.org.cn/Home.html).

Radiocarbon dating

Four Nanmu specimens were collected for AMS (Accelerator Mass Spectroscopy) radiocarbon dating at the Guangzhou Institute of Geochemistry, Chinese Academy of Sciences, and two of them were also used for DNA analyses (Table 2). All chronological data were calibrated using OxCal v4.4.4 (Bronk Ramsey, 2021), applying the latest consensus calibration curve, IntCal20 (Reimer, 2020). The AMS ¹⁴C dates determined ranged from 335 to 770 years BP (95.4% probability) (Table 2).

TABLE 2 Calibrated radiocarbon dates of Nanmu specimens taken from structural components from the Forbidden City

Sample ID	Laboratory no.	Dating method	Collecting site	Age 14C, y ΒΡ	Calibrated age, cal y BP
					1σ (68.3%)	2σ (95.4%)
CNM02a	21110214OCSP	AMS	Gate of Compassion and Tranquility	335 ± 15	443 ± 20	433 ± 47
GGN03a	21110220OCSP	AMS	Forbidden City	620 ± 20	706 ± 10	655 ± 31
WM03	21110218OCSP	AMS	Meridian Gate	770 ± 20	752 ± 10	768 ± 26
ZHD03	21110219OCSP	AMS	Hall of Central Harmony	650 ± 20	647 ± 11	647 ± 16

Abbreviation: AMS, Accelerator Mass Spectroscopy.

^aSample included in aDNA analysis.

Residual DNA staining in the wood cells

Confocal Laser Scanning Microscopy (CLSM, TCS SPE, Leica) with Propidium Iodide (PI) staining was used to observe the position and presence of residual DNA in the cells of the Nanmu specimens. Further details are given in Method S2.

DNA extraction, hybridization capture, and sequencing

DNA was extracted from each specimen in a dedicated wood DNA identification laboratory at the Chinese Academy of Forestry, using the method of Lu et al. (2020) and Rohland et al. (2018) with modification (Method S3). Blank DNA extraction and PCR amplification were conducted to monitor for contamination. No modern Nanmu DNA was present in the same building.

RNA probes (baits) from modern chloroplast DNA (cpDNA) were obtained from iGeneTech Bioscience (design.igenetech.com, China) for liquid capture. DNA repair was performed on each DNA extract using the NEBNext FFPE DNA Repair Mix (New England BioLabs Inc, Ipswich, MA, USA) before DNA library construction. The enrichment was done following the protocol detailed in Method S4. After the DNA library was constructed, it was sequenced on a Novaseq 6000 sequencer (Illumina Inc., San Diego, CA, USA) with 150 base paired end reads (Method S4).

Ancient data mapping, alignment, filtering, and assembly

Raw sequencing reads were processed by the BAM Pipeline with the PALEOMIX v.1.3.6 software (Schubert et al., 2014). Illumina adapter sequences were trimmed from the sequencing data, short length reads below 25 bp were filtered, and overlapping pairs were merged with AdapterRemoval v.2.2.0 (Schubert et al., 2016). The cleaned reads were aligned against the reference chloroplast genome of P. zhennan (MH033832), with the bwa-backtrack algorithm within the Burrows-Wheeler Aligner (Li et al., 2009). Seeding was disabled as per the recommendations from Schubert et al. (2012), and alignments with a mapping quality below 25 unmapped and PCR duplicates were discarded using the Picard tools (http://broadinstitute.github.io/picard/). The DNA damage patterns, including DNA fragmentation and nucleotide misincorporation, were characterized using mapDamage 2.0 (Jónsson et al., 2013). The transition sites (C/T and G/A) were excluded in ancient samples to minimize the false-positive results.

Phylogenetic analysis and taxonomic assignment

MrBayes 3.2 (Ronquist et al., 2012) was used to analyze the dataset based on Bayesian Inference (BI) methodology for phylogenetic tree reconstruction. The best-fit model GTR + I + G was set by running the MrBayes program for 1,000,000 generations, with sampling every 1,000 generations. Upon completion of the runs, the first 25% of generations were discarded as burn-in. Meanwhile, the program Median-joining (Bandelt et al., 1999) was implemented using the default parameters in PopART v.1.7 (Leigh & Bryant, 2015) to infer the haplotypes networks.

RESULTS Wood anatomy

The microscopic features of Nanmu specimens from the Forbidden City are shown in Figure 2. According to the list of microscopic features for hardwood identification prepared by the International Association of Wood Anatomists (IAWA Committee, 1989), the wood anatomical characteristics are diffuse-porous with mainly solitary vessels in radial or diagonal multiples (usually two to three cells), sometimes in small clusters (Figure 2b); alternating intervessel pits; simple or scalariform vessel perforation plates; abundant septate fibers (Figure 2c); scanty paratracheal and vasicentric axial parenchyma (Figure 2b); rounded, oval, and gash-like vessel-ray pit outlines similar to intervessel pittings (Figure 2d); uniseriate rays are rare, with multiseriate rays of two to three cells wide, composed of procumbent cells with one to two marginal rows of upright and square cells (Figure 2c,d) more common; oil and mucilage cells are commonly associated with axial and ray parenchyma (Figure 2b,c); and prismatic crystals present in ray cells (Figure 2d).

Figures S1 and S2 show the anatomical ranges of wood in indigenous tree species of Phoebe and Machilus in China, both completely overlapping with the Nanmu specimens studied. This is typical of these closely related genera in the Lauraceae.

Ancient DNA preservation, capture and authenticity

Propidium Iodide (PI) staining was used to reveal the preservation of endogenous DNA in the Nanmu specimens. DNA was only sporadically present in wood ray parenchyma cells (Figure 3b,d,f), and no DNA signal was found in other types of cells such as wood fibers, vessel elements, and intercellular regions (Figure 3a,c,e).

FIGURE 3. DNA staining in Nanmu specimens of structural components of the Forbidden City (take sample GGN03 as an example). (a, b) Bright field mode in laser confocal microscope. (c, d) Dark field mode. (e, f) Overlay field mode. F-fiber cell; R-ray parenchyma cell; V-vessel. Note: The blue boxes indicate the position of residual DNA in the wood cells; the white arrows indicate the residual DNA.

Twenty-one high-quality plastid sequences of the Nanmu specimens randomly sampled from nine representative palaces of the Forbidden City were obtained using the aDNA capture method. The enriched libraries were sequenced, yielding a total of 114 million raw pair-end reads for all the specimens, of which 23.94% reads mapped to the reference chloroplast genome (MH033832) (Table 1). We recovered 137,663–152,805 bp (90.09%–100% coverage) of the plastid genomes, with sequence depths of 27.05- to 1409.94-fold (mean 234.15-fold) (Figure 4a; Table 1).

FIGURE 4. Ancient DNA capture, authenticity and species identification. (a) The fragment length and sequence depth of aDNA. Note: bp-base pair. (b) Terminal deamination patterns of aDNA after repair treatment. Note: IQR-interquartile range. (c) The Bayesian tree built from plastid genomes depicting the species affiliation of 21 Nanmu specimens within the range of genus Phoebe. Note: P-Phoebe; the red text indicates 21 Nanmu specimens; the letters “A,” “B,” and “C” indicate three monophyletic groups within the genus of Phoebe. (d) Median-joining network illustrating the relationship between 21 Nanmu specimens and six closely related species of Phoebe. Note: The number of short lines on these segments indicates the number of substitutions separating one haplotype from its neighbor; the gray shapes indicate six haplogroups A to F.

To demonstrate the authenticity of aDNA without modern DNA contamination, we examined the empirical sequence features of these ancient DNA samples, which included short lengths of sequenced DNA fragments in ancient samples and abundant damage patterns at the 5′ and 3′ ends of sequenced DNA fragments (reflected as deamination-derived C-to-T and G-to-A nucleotide mismatches) (Wu et al., 2019). Fragment distributions for mapped reads generally ranged between average fragment sizes of 68–99 bp (Figures 4a and S3). Terminal deamination follows patterns after repair treatment, with 5′ C-to-T rates between 0.3% and 2.4%, and 3′ G-to-A rates between 1.5% and 5.7% (Figures 4b and S3). These characteristics suggest that the sequence data originated from typical templates of aDNA.

Resolving the species affiliation of Nanmu using aDNA

The dataset for this study consists of 61 complete chloroplast genomes from GenBank (seven of them generated in this study, i.e., OM022239-OM022244 and MH033832, Figure S4 and Table S1), including 17 Phoebe species and 22 Machilus species, covering the full range of plastid DNA genome data from these two genera published to date. The Bayesian phylogenetic analyses showed that the two genera Phoebe and Machilus clustered as two well-supported monophyletic groups I and II, respectively. All the Nanmu specimens from the Forbidden City were clustered into group I and identified as Phoebe spp. (Figure S5).

Based on the above results, we further identified the Nanmu specimens within the range of genus Phoebe. The genus Phoebe has 29 indigenous species in China (Table 3). Due to tree size and geographical distribution, only 18 of these species qualify as candidates for Nanmu (Table 3). We lacked DNA data on three species of Phoebe and carried out phylogenetic analyses using the chloroplast genome of the remaining 15 species, representing the most dominant species (83.3% of relative coverage) in China, to explore the species identity of Nanmu specimens studied. Our phylogenetic analyses resolved three monophyletic groups A, B, and C, while the six species, i.e., P. bournei, P. chekiangensis, P. hui, P. neuranthoides, P. sheareri, and P. zhennan showed close relationships and fell into group A (Figure 4c). Within this group, the 19 Nanmu specimens were placed together with the chloroplast genome sequences of P. zhennan and P. hui with a posterior probability of 0.75, while the other two specimens THD05 from the Hall of Supreme Harmony and TRG03 from the Belvedere of Embodying Benevolence were classified as P. bournei or P. chekiangensis, and P. bournei, respectively (Figure 4c).

TABLE 3 The list of Phoebe species indigenous in China

No.	Species	Notea
1	Phoebe angustifolia	Small tree/shrub size
2	Phoebe bourneib	Data used
3	Phoebe brachythyrsa	Small tree/shrub size
4	Phoebe cavalerieib	Data used
5	Phoebe chekiangensisb	Data used
6	Phoebe faberib	Data used
7	Phoebe formosana	Not distributed in Nanmu collecting regions
8	Phoebe glaucophyllab	Data used
9	Phoebe hainanensis	Not distributed in Nanmu collecting regions
10	Phoebe huib	Data used
11	Phoebe hunanensis	Small tree/shrub size
12	Phoebe hungmoensisb	No complete cpDNA data
13	Phoebe kwangsiensis	Rare/scarce distribution
14	Phoebe lanceolatab	Data used
15	Phoebe legendreib	No complete cpDNA data
16	Phoebe lichuanensis	Rare/scarce distribution
17	Phoebe macrocarpab	Data used
18	Phoebe megacalyx	Rare/scarce distribution
19	Phoebe motuonan	Not distributed in Nanmu collecting regions
20	Phoebe neuranthab	Data used
21	Phoebe neuranthoidesb	Data used
22	Phoebe nigrifolia	Small tree/shrub size
23	Phoebe puwenensisb	Data used
24	Phoebe rufescensb	No complete cpDNA data
25	Phoebe shearerib	Data used
26	Phoebe tavoyanab	Data used
27	Phoebe yaiensis	Not distributed in Nanmu collecting regions
28	Phoebe yunnanensisb	Data used
29	Phoebe zhennanb	Data used

^aFrom Flora of China; Phoebe faberi (Synonyms: Phoebe omeiensis).

^bCandidated Phoebe species for Nanmu.

In order to further investigate the phylogenetic relationship of the haplotypes, we also constructed a median-joining (MJ) network with 39 sequences in the group A of Figure 4c. It showed 135 variable sites with 27 haplotypes, and a Haplotype diversity (Hd) of 0.9676. The haplotypes of our study fell into six haplogroups, labeled haplogroup A to F (Figure 4d). Overall, the results of the MJ network analysis are consistent with the Bayesian phylogenetic tree. Eighteen specimens were clustered into haplogroup A together with the chloroplast genome sequences of P. zhennan and P. hui, except sample TRG04 from the Belvedere of Embodying Benevolence alone with haplogroup B, sample TRG03 with haplogroup C including P. bournei and sample THD05 with haplogroup D including two species P. bournei and P. chekiangensis (Figure 4d).

DISCUSSION

Obtaining centuries-old wood DNA using conventional DNA extraction and PCR techniques is extremely difficult (Deguilloux et al., 2002; Lendvay et al., 2018; Wagner et al., 2018). It is challenging even in the xylem of standing trees, especially in their heartwood tissues, generally composed of dead cells; their organelles and nuclei degrade with programmed cell death (PCD), and residual DNA molecules eventually adsorb on the cell walls. Moreover, extractives including tannins, essential oils, gums, and pigments existing in wood tissues could greatly hinder the efficient extraction and PCR amplification of wood DNA. This is especially true in Phoebe and Machilus wood, which, like many other Lauraceae, contain enlarged oil cells (Figure 2b,c) (IAWA Committee, 1989).

The endogenous DNA of ancient wood specimens from the Forbidden City has inevitably been subject to more serious degradation than DNA in the heartwood of standing trees due to centuries of temporal change and climatic fluctuation, as well as through the intervention of external factors such as insects and microbial attack. Hence, a major concern was whether it would be possible to retrieve DNA from these ancient wooden structures, and whether this DNA could still be further analyzed and identified. It was therefore necessary to first clarify the state of the wood's DNA preservation and to pinpoint the precise physical location of the DNA. The staining results showed that the DNA signal is located solely in the ray parenchyma cells of wood tissues (Figure 3). The pore-rich structure (open cell lumina, perforations, and pits) of wood makes it highly susceptible to exogenous DNA contamination. In this way, the positive DNA stainings could raise the question of whether they may come from exogenous contamination rather than endogenous DNA. Here, we conclude that the stained DNA is endogenous. First of all, it has been proved that modern wood DNA is commonly found in ray parenchyma cells (Abe et al., 2011; Nakaba et al., 2006). Furthermore, if the fluorescence was the result of contamination by bacteria or fungi, the staining would not be localized in the ray parenchyma cells, but more likely in the vessel elements with their large volumes, and even in the wood fibers which compose the largest proportion of hardwood. The relatively enclosed space formed by wood ray cells with lignified cell walls might also provide a potentially favorable environment for the preservation of aDNA. In general, clarification of the physical location of wood aDNA facilitates the development of more efficient sampling strategies to maximize DNA recovery from archeological wood materials. The aDNA extraction and purification protocol used in this study effectively addresses the problem of the low purity of DNA extracted from archeological wood and provides a good source of DNA for subsequent library construction.

The aDNA capture technique retrieved 21 plastid genomes of Nanmu specimens from structural components of the Forbidden City, including pillars, beams, and square columns. This provided direct molecular evidence for identifying the tree species affiliation of Nanmu. Our results show that all of the Nanmu specimens belong to the genus Phoebe, with a high proportion of P. zhennan or P. hui from southwestern China, as well as a smaller proportion of P. bournei or P. chekiangensis from south China (Figure 4c). To explore the question of the species attribution of Nanmu, it is necessary to clarify the criteria used for collecting logs for the royal department of timber harvesting in the Ming and Qing Dynasties. Historical documents like the Compendium of Materia Medica recorded (Wang, 1993) in detail the geographical distribution of Nanmu and its tree morphological characteristics such as leaves, flowers, stems and timbers, which are consistent with the tree taxonomical features of the four Phoebe species (Table 4) indicated by the molecular identification results of this study. During the Ming and Qing Dynasties, the most concentrated areas where Nanmu was harvested were reported to be in southwestern China, that is, Sichuan, Guizhou, Hubei, and Yunnan provinces (Lan, 1994), which coincide with the main distribution areas of P. zhennan and P. hui (Figure 1), the two species identified by aDNA analysis. The Bayesian results show that P. zhennan and P. hui are clustered as one clade, indicating that they are very closely related (Figure 4c). However, the haplotype results suggest that there is a same sense mutation between P. zhennan (C) and P. hui (T) at Position 132,432, belonging to the coding region ycf1 (Figure 4d). In addition to the molecular evidence, there are also minor differences in leaf and fruit morphology (Li et al., 2008), but according to The Flora Sichuanica (Chao & Kung, 1981), the two species are sympatric and their wood properties and uses are generally similar. Considering the similarity of their spatial distribution and wood properties, as well as the enormous human and financial costs involved in harvesting timber, we can speculate that the craftsmen would have collected both species simultaneously. Additionally, during the Qing Dynasty (1644–1912 AD), the areas from which Nanmu resources were harvested extended further into southern China (Lan, 1994). With the continuous extension of the collection area and the increasing scarcity of Nanmu resources, it is quite possible that the two other species identified in this study, that is, P. bournei and P. chekiangensis, which are similar in tree morphology and wood properties to P. zhennan and P. hui (Table 4), were also utilized for the construction of the Forbidden City. In summary, we speculate that P. zhennan and P. hui were the main species of this imperial wood used in the Forbidden City. This differs from the current prevailing view that the concept of Nanmu refers to the whole genus Phoebe and even includes its close relative Machilus. We also speculate that the criteria for selecting Nanmu materials used in the construction of the Forbidden City were more stringent than is suggested by the broader mainstream concept of the range of species included in Nanmu in folklore.

TABLE 4 Comparison of morphological characteristics between the historical records of Nanmu and the tree species identified in this study

Morphological characteristics	Historical recordsa	Phoebe zhennanb	Phoebe huib	Phoebe bourneib	Phoebe chekiangensisb
Geographical distribution	Distributed in the southern China, especially in the Southwest China (mountains of Sichuan and Guizhou provinces)	Distributed in Southwest China (western Hubei, northwestern Guizhou and Sichuan)	Distributed mainly in Southwest China (South Shaanxi, Sichuan and Northeast Yunnan)	Distributed in the southern China (Jiangxi, Fujian, South Zhejiang, Guangdong, north and Northeast Guangxi, Hunan, Hubei, southeast and Northeast Guizhou)	Distributed in the southern China (northwest and Northeast Zhejiang, North Fujian and East Jiangxi)
Leaf	Leaves as big as a cattle's ears with the end acuminate	Leaves blade elliptic, rarely lanceolate or oblanceolate, 7–11(−13) × 2.5–4 cm, leathery, apex acuminate, apex straight or falcate, base cuneate, terminal obtuse or acute	Leaf blade elliptic, elliptic-oblanceolate, or elliptic-lanceolate, 5–8(−10) × 1.5–3 cm, leathery, base narrowly cuneate, apex acuminate or caudate-acuminate	Leaf blade lanceolate or oblanceolate, 7–13(−15) × 2–3(−4) cm, leathery, conspicuously foveolate abaxially, base attenuate or cuneate, apex acuminate or long acuminate	Leaf blade obovate-elliptic or obovate-lanceolate, rarely lanceolate, 7–17 × 3–7 cm, usually 8–13 × 3.5–5 cm, leathery, base cuneate or subrounded, apex acute-acuminate or long acuminate
Flower	Yellow	Yellow	Yellow	Yellow	Yellow
Stem	Large trees, to more than 10 Zhang (Chinese unit) in height (approx. 30 m); trunk straight	Large trees, 30 m in height; trunk straight	Large trees, to 25 m in height; trunk straight, to 60 cm DBH	Large trees, 15–20 m in height; trunk straight	Large trees, to 20 m in height; trunk straight, to 50 cm DBH
Timber	Fragrant smelling; fine-grained	Fragrant smelling; straight grain and fine structure	Fragrant smelling; fine-grained	Fragrant smelling; straight grain and fine structure	Fragrant smelling; straight grain and fine structure

Abbreviation: DBH, diameter at breast height.

^aFrom Wang (1993).

^bFrom Flora of China.

Our results beg the questions of how and why these selected Phoebe species, growing thousands of kilometers away from Beijing, were logged and transported by the royal department of timber harvesting in the Ming and Qing Dynasties. The initial construction of the Forbidden City (1406 to 1420 AD) is more than 600 years old, although several renovation projects were carried out later (Table 1). Throughout these periods, a framework of plant taxonomy and wood science was absent in China (Ma et al., 2020), but the craftsmen had a clear concept of the materials to select for the construction of the palaces. In addition to the precise identification and selection of specific tree species, their expertise is also evident from their understanding of the wood's properties. The choice of Nanmu for load-bearing components of the Forbidden City, such as columns, beams, and pillars, might suggest that craftsmen knew Nanmu had favorable mechanical load-bearing strength, straight grain, natural durability, and resistance to insects.

Since most of the wood-harvesting areas for Nanmu species are in the mountains and forests, where the terrain is very steep, and since the Nanmu trunks were of large diameter, they would have been very difficult to transport. Consequently, it could have been common for the craftsmen to use river transport from southern China to Beijing during the rainy season (Lan, 1994). This is in contrast to the principle often reported in the literature that wooden cultural relics, including ancient wooden buildings, were generally made of local tree species due to the limitations of transportation (Dong, Lu, et al., 2017; Dong, Zhou, et al., 2017). The fact that Nanmu species, used for construction for the Forbidden City located in northern China, grew in southwest and southern China and therefore could only be collected and transported at great expense, further highlights the significance of this imperial wood selected specially for the construction of the palaces. In conclusion, it can be said that our ancestors' knowledge of tree species and wood properties, transportation, and utilization imply a highly sophisticated civilization.

Our work fills a gap in the genetic map of Nanmu wooden relics and provides a new perspective for the long-standing dispute, which previously rested solely on historical documents and lacked direct evidence. By combining morphological traits with aDNA analyses, we offer a new solution for the tree species affiliation of Nanmu relics (Figure 5) and also open a window for the accurate identification and restoration of wooden cultural relics. Future aDNA studies, combined with multidisciplinary methods on wooden components from a greater range of archeological sites, will help to solve questions related to the species used and further our understanding of the past interactions among biogeography, forests, and human activities.

FIGURE 5. A flow chart of the method used to identify the species of Nanmu specimens taken from structural components of the Forbidden City, based on morphological and aDNA analyses

ACKNOWLEDGEMENTS

This study was supported financially by the Fundamental Research Funds of Chinese Academy of Forestry (Grant CAFYBB2021QB003) and the National High-level Talents Special Support Program of China (Grant W02020331). We would like to express our gratitude to the Department of Historic Architecture of the Palace Museum and Mr. Hua Li, Mr. Yunshi Chen, and Mrs. Xiuying Liu of the Research Institute of Wood Industry, Chinese Academy of Forestry, and Ms Xiaoling Liao and Ms Wei Luo of Sichuan Agricultural University for their help on sample collection, and Professor Jinxing Lin of Beijing Forestry University for his academic suggestions.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

AUTHOR CONTRIBUTIONS

Y. Y. and L. J designed the research; Y. C., J. W. and X. J. provided the specimens; L. J., Y. L., Y. G., C. X., and Y. Z. performed the research; L. J., Y. L., Y. G., M. Z., Z. W., J. G., T. H., L. M., W. G., Y. C., J. W., S. Z., P. B., X. J., and Y. Y. analyzed the data; L. J., Y. L., P. B. and Y. Y. wrote the paper.

DATA AVAILABILITY STATEMENT

Sequencing data are available in the NCBI Sequence Read Archive under BioProject ID PRJNA866908. Please contact the corresponding author for other data.