In this study, we examined the pollen morphology of the platycodonoid group in Campanulaceae s. str. using a scanning electronic microscope. We used pollen grains of 25 accessions representing 24 species of the Codonopsis complex (including Campanumoea, Cyclocodon, Leptocodon, and all three subgenera of Codonopsis), which is extremely controversial among authors for taxonomic treatment. Pollen morphology of all the other genera in the group observed by previous authors is taken into account in our discussion. A total of nine pollen types with two subtypes in the group were recognized and named for the first time. Molecular and morphological data imply that each pollen type corresponds to a natural group at generic level, and thus the mergence of Leptocodon with Codonopsis and the restoration of Cyclocodon as a separate genus are justifiable, and Codonopsis subg. Pseudocodonopsis, subg. Obconicicapsula, and two species of Codonopsis subg. Codonopsis (C. purpurea and C. chimiliensis) may be better classified as three independent genera separate from the core Codonopsis.
The Campanulaceae s. str. is a well-established family (Cosner et al. 1994; Gustafsson and Bremer 1995; Gustafsson et al. 1996; Lundberg and Bremer 2003). However, the subdivisions of the family and the relationships of genera were unclear before this century. Recent studies on molecular systematics of the Campanulaceae s. str. have resulted in more insight into infrafamilial phylogenetic relationships. Eddie et al. (2003) analyzed the phylogenetic relationships of the family using ITS sequences of nuclear ribosomal DNA. Borsch et al. (2009) conducted analyses of phylogenetic relationships using DNA sequences of the chloroplast gene petD. In the same year, Haberle et al. (2009) used DNA sequences of three chloroplast genes, atpB, matK, and rbcL, to analyze the phylogenetic relationships of the Campanulaceae s. str. These three studies tell us that the Campanulaceae s. str. consists of three major monophyletic groups: campanuloids, platycodonoids and wahlenbergioids.
On the other hand, the three studies mentioned above also indicate that the phylogenetic relationships within each of the three major groups are far from well established. They all clearly show that Campanula and Wahlenbergia each is a polyphyletic genus in the traditional sense. Eddie et al. (2003) and Haberle et al. (2009) have convinced us that the genus Codonopsis Wall. in the platycodonoid group is also polyphyletic.
Pollen morphology has been used as a very powerful tool to consider systematics of the Campanulaceae. Erdtman (1952) conducted a rather extensive survey on pollen of the Campanulaceae s. lat. and pointed out that Schönland's “subtribes of Campanulaceae – Campanuloideae do not seem to be homogeneous.” For example, Schönland's Platycodinae is a mixture of Platycodon with pollen having elongate apertures, and Microcodon and Musschia with porate pollen. Thulin (1975) further emphasized the significance of pollen morphology in taxonomy, stating that for the arrangement of genera in the Campanulaceae s. str., the first consideration should be pollen morphology. Morris and Lammers’ (1997) work is the most extensive on the Codonopsis complex. They observed pollen morphology of 23 accessions representing 17 species, and took into account the results of previous investigators. As a result, they recognized four pollen types, and revealed their close relationships with taxonomy.
Molecular phylogenetic results clearly show that there is a major dichotomy between the colpate/colporate pollen alliance (platycodonoid taxa) and the porate pollen alliance (wahlenbergioid and campanuloid taxa). The wahlenbergioid and campanulioid groups both have porate pollen, but the former possesses capsules dehiscent by apical valves (loculicidal), while capsules of the latter are dehiscent by lateral pores (poricidal).
In the platycodonoid group, the circumscription of Codonopsis has been controversial among authors. Moeliono and Tuyn (1960) merged Campanumoea s. lat. (including the genus Cyclocodon) with Codonopsis. This circumscription of Codonopsis was accepted by Grey-Wilson (1990) and Lammers (1992). Grey-Wilson (1990) merged Leptocodon into Codonopsis, and this action was followed by Lammers (1999). Thus, Lammers adopted the broadest concept of Codonopsis, accommodating all four genera mentioned above. However, Hong (1983) recognized Campanumoea s. lat. (including sect. Cyclocodon) and Leptocodon each as an independent genus. Hong and Pan (1998) restored Cyclocodon at generic rank based on data from the pollen and seed-coat, which are distinct from those in Codonopsis and Campanumoea sect. Campanumoea. In “Flora of China” (Hong et al. 2011: p. 505), readers can find controversial views on how to treat the group under discussion. Hong recognized four genera, but the two co-authors, Lammers and Klein, stated their belief that Campanumoea and Leptocodon should be merged into Codonopsis. Another issue involved in the taxonomic treatment of the group under study is whether Codonopsis s. str. is monophyletic or polyphyletic. According to the current designation (Shen and Hong 1983; Hong in Hong et al. 2011), Codonopsis consists of three subgenera, subg. Codonopsis, subg. Obconicicapsula D. Y. Hong (1980) and subg. Pseudocodonopsis Kom. (1908). Two molecular trees, one based on the ITS sequence of nuclear ribosomal DNA (Eddie et al. 2003), and the other based on three cpDNA genes, atpB, matK and rbcL (Haberle et al. 2009), indicate that the only species of subg. Obconicicapsula, Codonopsis dicentrifolia (C. B. Clarke) W. W. Smith, does not join the Codonopsis clade, but instead forms a clade with Cyananthus lobatus. Based on their observations, Morris and Lammers (1997) revealed that three species in subg. Pseudocodonopsis have pollen morphology distinctly different from that of most species of subg. Codonopsis. If the other species of subg. Pseudocodonopsis are the same as these three species in terms of pollen morphology, one could question the justification for keeping them in Codonopsis, considering the sharp differences in external morphology between two subgenera.
Consequently, we think that it is worthwhile to conduct more extensive observations on pollen morphology of the platycodonoid group, and to conduct a comprehensive analysis because of the diverse pollen morphology of the group and the controversial views on taxonomy among authors as mentioned above.
The results of our observations are presented in Figures 1–6 and Table 1, where previous observations are also provided to give adequate information for discussion purposes. From Table 1, the pollen of Platycodon is 5–6-colporate with medium-high sexine spinules (1.5 μm), and is thus a distinct type of pollen called the Platycodon Type. The pollen of genus Ostrowskia is very special, and is 6–7-colpate with large-verrucose sexine. It is deserving of being named an independent pollen type, the Ostrowskia Type. The genus Echinocodon D. Y. Hong possesses 4–5-colpate pollen with short colpi (colpus length/polar axis length ≅ 0.36) and basally divided sexine spinules. Such pollen cannot be assigned to any of the above-mentioned types, and thus is here designated as the Echinocodon Type. According to Dunbar (1975a) and Shrestha & Tarasevich (1992), the pollen of the genus Cyananthus is 6–9-colpate with long colpi, verrucose colpus membrane, and verrucose sexine (Table 1). It should also be designated an independent pollen type, the Cyananthus Type.
Table 1. Pollen morphology of the platycodonoid group (Campanulaceae s. str.)(the classification of Codonopsis and its allies follows “Flora of China” Vol.19. 2011)
1) colpus length/polar axis length: long (0.70–0.78); medium-long (0.54–0.57); short (0.2–0.4)
2) spinule/verrucae height: high (≥ 2 μm); short (≤ 1 μm).
We examined pollen of the type species of Codonopsis, C. viridis (Figure 2G, H), which is 7-colpate with long colpi, spinulose colpus membrane, and short exine spinules. The pollen morphology of all the species of Codonopsis subg. Codonopsis listed in Table 1, except for C. chimiliensis and C. purpurea, and shown in Figure 1 and Figure 2A–F, is similar to that of C. viridis, and distinctly different from all the others observed here. It is here designated as the Codonopsis Type, and corresponds to Morris and Lammers’ (1997) Type I.
The genus Canaria is homogeneous in pollen and has a pollen type of its own, and its generic status has never been questioned. Pollen grains of 17 species of Cyananthus (nearly all) have been observed, illustrating that the genus is homogeneous in pollen morphology. The genera Platycodon, Ostrowskia, and Echinocodon each is morphologically distinct, and each has a separate pollen type of its own. Their generic positions have also not be brought into question.
All the three species of Cyclocodon are uniform in pollen morphology and share the same pollen type: the Cyclocodon Type. Moeliono and Tuyn (1960) merged Campanumoea s. lat. (including Cyclocodon) with Codonopsis, ignoring the sharp differences in pollen between Campanumoea parviflora (=Cyclocodon parviflorus (Wall. ex A. DC.) Hook. f. & Thomson) (3-colporate) and three species of Codonopsis (7–9-colpate) as already revealed by Erdtman (1952). Not only do molecular (Borsch et al. 2009; Wang et al. unpublished) and morphological data support the restoration of Cyclocodon to generic status based on pollen and seed-coat morphology as proposed by Hong and Pan (1998), but these data also indicate that the closest relative of Cyclocodon is the genus Platycodon, not Codonopsis. Leptocodon was merged with Codonopsis by Grey-Wilson (1990), which was followed by Lammers (1999, 2001). The present work shows that Leptocodon is of the same pollen type as that of the core Codonopsis. In our molecular tree constructed using sequences of four chloroplast DNA regions and the ITS sequences of nuclear ribosomal DNA (Wang et al. unpublished), Leptocodon is nested among the species of the core Codonopsis. Therefore, palynological and molecular data both support Grey-Wilson's (1990) mergence of Leptocodon into Codonopsis. It is rather hard to classify Campanumoea Bl. s. str. in taxonomy from a palynological view. The taxon was merged with Codonopsis by Moeliono and Tuyn (1960), and this action was followed by Grey-Wilson (1990) and Lammars (1992). However, Hong (1983b, 1995, in Hong et al. 2011) kept it separate from Codonopsis. The pollen of the taxon is recognized here as a subtype of the Codonopsis Type. Morphologically, it differs from Codonopsis in having baccate fruits. In our molecular tree mentioned above, Campanumoea forms a clade sister to the core Codonopsis. Therefore, whether to retain the taxon as an independent genus or to merge it with Codonopsis are two options that can both be considered, but we prefer for it to have generic status.
Although it seems that we have clarified the taxonomy of the platycodonoid group, we have still not reached that point. The pollen of Codonopsis s. str., as circumscripted by Hong (1983b, 1995, in Hong et al. 2011), is highly heterogeneous. Codonopsis subg. Pseudocodonopsis Kom. is a distinct pollen type of its own: the Pseudocodonopsis Type. Codonopsis subg. Obconicicapsula D. Y. Hong, with only one species, also has a distinct pollen type: the Obconicicapsula Type. Even Codonopsis subg. Codonopsis itself is heterogeneous in pollen morphology. Pollen grains of two species, C. chimiliensis and C. purpurea, are uniform, but differ distinctly from those of all the other species examined in the subgenus. It is named here as the Obconicicapsula Type Purpurea subtype. If, for the convenience of discussion, we use the Codonopsis complex to cover all the taxa with controversial views regarding their taxonomic treatments, then the complex includes Codonopsis s. str., Campanumoea, Cyclocodon, Leptocodon. Table 2 shows that the pollen in the complex is polytypical, containing four types with two subtypes. The pollen of Campanumoea, Cyclocodon, and Leptocodon and their taxonomical classifications have been discussed above. From Table 2, the genus Codonopsis s. str. (Shen and Hong 1983; Hong in Hong et al. 2011) possesses three pollen types, which have been recognized by Morris and Lammers (1997), and two subtypes. The above discussion implies that a particular pollen type generally corresponds to generic status. Therefore, Codonopsis subg. Pseudocodonopsis and subg. Obconicicapsula, each possessing their own pollen type, should be separated as an independent genus. The typical subgenus, subg. Codonopsis, is still heterogeneous in pollen morphology, containing two types, the Codonopsis Type and the Obconicicapsula Type Purpurea subtype. Taking morphological and molecular data (Table 2; Wang et al. unpublished) into consideration, the two species Codonopsis purpurea and C. chimiliensis, with the Obconicicapsula Type Purpurea subtype of pollen, should be separated from Codonopsis and given generic status.
Table 2. Pollen morphology and external morphological characters in the Codonopsis complex (including Campanumoea, Cyclocodon and Leptocodon) (the classification system follows “Flora of China”, Vol.19. 2011)
We sampled 17 species of Codonopsis and all the species of Campanumoea (2), Cyclocodon (3), and Leptocodon (2). All the materials and vouchers are listed in Table 3. Pollen grains from herbarium specimens were treated according to Hong (1983a), being first soaked in warm water for 5 min, and then dehydrated in 100% alcohol for a few minutes. They were coated before being observed under a Scanning Electronic Microscope (Hitachi S-4800). Because lengths of apertures and spinules on the sexine surface are constant within species and even within pollen types (Dunbar 1975a; Morris and Lammers 1997), we used measurements from only one or two pollen grains. The terminology here used follows Dunbar (1975a).
Table 3. Materials for observations on pollen morphology of the platycodonoid group (Campanulaceae s. str.) (the classification system follows “Flora of China”, Vol.19. 2011)
C. inflata (Hook. f. & Thomson) C. B. Clarke
Mêdog, Tibet, China
Qinghai-Xizang Exped. 1588 (PE)
C. javanica Bl. ssp. javanica
Ceheng, Quizhou, China
Z. Y. Cao 533 (PE)
Codonopsis subg. Codonopsis
C. benthamii Hook. f. & Thomson
Gongshan, Yunnan, China
Gaoligong Shan Biodiversity Survey 31195 (CAS)
C. chimiliensis J. Anthony
G. Forrest 29890 (PE)
C. farreri J. Anthony
Fugong, Yunnan, China
Gaoligong Shan Biodiversity Survey 32059 (CAS)
C. meleagris Diels
Lijiang, Yunnan, China
R. C. Ching 30503 (PE)
C. purpurea Wall.
Nyalam, Tibet, China
D. Y. Hong et al. H11003 (PE)
C. subglobosa W. W. Smith
Muli, Sichuan, China
D. Y. Hong et al. H11029 (PE)
C. subscaposa Kom.
Xiangcheng, Sichuan, China
Qinghai-Xizang Exped. 3981 (PE)
C. subsimplex Hook. f. & Thomson
G. H. Cave s. n. (PE)
Yadong, Tibet, China
P. C. Tsoong 5931 (PE)
C. thalictrifolia Wall.
Nyalam, Tibet, China
D. Y. Hong & Q. Wang XZ002 (PE)
C. viridis Wall.
R. Strachey & J. E. Winterbottom No.3 (PE)
Codonopsis subg. Obconicicapsula
C. dicentrifolia (C.B. Clarke.) W. W. Smith
Nyalam, Tibet, China
D. Y. Hong et al. H11004 (PE)
Codonopsis subg. Pseudocodonopsis
C. convolvulacea Kurz
Yuanjiang, Yunnan, China
W. Q. Yin 1962 (PE)
ssp. grey-wilsonii (J. M. H. Shaw) D. Y. Hong
Nyalam, Tibet, China
D. Y. Hong et al. H11005 (PE)
ssp. vinciflora (Kom.) D. Y. Hong
Gongbo’gyamda, Tibet, China
D. Y. Hong & Q. Wang H10003 (PE)
C. graminifolia H. Lév.
Yanbian, Sichuan, China
D. Y. Hong et al. H11042 (PE)
C. hirsuta (Hand. -Mazz.) D. Y. Hong
Lijiang, Yunnan, China
D. Y. Hong et al. H10029 (PE)
C. rosulata W. W. Smith
Yanbian, Sichuan, China
D. Y. Hong et al. H11039 (PE)
C. celebicus (Bl.) D. Y. Hong
Mêdog, Tibet, China
Z. C. Ni et al. 0650 (PE)
C. lancifolius (Roxb.) Kurz
Emei Shan, Sichuan, China
D. Y. Hong et al. H11021 (PE)
C. parviflorus (Wall. ex A. DC.) Hook. f. & Thomson
J. D. Hooker s.n. (PE)
L. gracilis Hook. f. & Thomson
Muli, Sichuan, China
G. Forrest 22158 (PE)
L. hirsutus D. Y. Hong
Cona, Tibet, China
D. Y. Hong et al. H11009 (PE)
Slightly different from previous observations, we paid special attention to type of aperture (colpate or colporate), colpus length, colpus membrane, type of sexine sculpture (spinulose or verrucose), spinule length, and morphology of the spinule base. We believe that these six characters are stable, and can thus reflect the differences and relationships between the taxa being studied.
(Co-Editor: Chun-Ming Liu)
We are grateful to Mr. MA Xin-Tang, Mr. YUE Ji-Pei, Mr. WANG Huan-Chong and Dr. WANG Qiang for contributing to our field work, and to Mr. XIAO Yin-Hou for the SEM operation and Mr. LI Min and Miss GONG Xiao-Lin for assistance in preparation of the manuscript. This work was supported by the National Natural Science Foundation of China (31170175).