Mid leaf sections representative of the Taxus Sumatrana Group, drawn by R. Spjut.  From top to bottom: Upper two are T-sections of T. mairei, top: from Guangdong (China), Tsang 20694 (US), showing elevated truncate midrib on abaxial surface that has slightly larger epidermal cells than those on adaxial surface, the lower section from Yunnan, type (A), showing channeled midrib and larger spherical parenchyma cells lying against smaller epidermal cells— most conspicuous along midrib and marginal zones.  Note elliptically shaped epidermal cells.  Lower two sections: abaxial epidermis from margin to across midrib; upper most drawn from type of T. speciosa, showing marginal region of 23 smooth cells in width, a stomata band with 16 rows of stomata, and a smooth midrib, the lowest section from type of T. sumatrana, showing long rectangular cells and 12–14 stomata rows

     The Taxus Sumatrana Group is characterized by leaves having a relatively broad region of marginal cells to the stomata bands, varying from 8–36 cells across, further differentiated by the irregular shape of epidermal cells and by their discoloration upon drying, appearing  glossy and reddish when dried—as shown below.    Leaves in other species groups usually have stomata bands bordered by fewer marginal cells, or marginal cells are less differentiated from stomata bands in shape, color, and in their development of papillae.  The above figure excludes the Subgroup Chinensis to emphasize the contrast among the three major groups of Taxus in Asia. The Chinensis subgroup is depicted on a similar map for number of stomata rows per band under the Wallichiana Group, and also in the following figure for number of marginal cells, from Spjut (2007a); the vertical axis shows the number of specimens tabulated for the  number of marginal cells across the abaxial leaf margin (without papillae) indicated along the horizontal axis (see illustrated examples above).

     The species and varietal taxonomy of the Sumatrana Group is difficult due to partly overlapping character traits among taxa.   Previous taxonomists recognized only one taxon in this group, either as a variety (Cheng & Fu 1975, 1978; Li & Fu 1997), or species (Florin 1948; Handel Manzzetti 1929; Hu 1964), but the correct name was not applied.  De Laubenfels (1978) for instance regarded T. sumatrana as the name for all yew plants in Southeast Asia (East Himalayas to Sumatera), but this is antedated by Taxus wallichiana.  This broad species concept (T. wallichiana) makes it difficult to distinguish T. wallichiana from Taxus cuspidata and its allies (Taxus baccata Group).   

     Spjut (2007b, 2013, this website) distinguishes a Sumatrana Group of four species and three varieties, two varieties remain unpublished. The taxa are differentiated by leaf characters that include three aspects of leaf curvature: (1) lengthwise vertically (down-curved vs. up-curved near apex), (2) lengthwise horizontally as evident by shape such as machete-shaped (curved more on one side than the other), sickle-shaped  (curved equally along both sides) vs. straight linear to oblong for most of their length, and (3) transverse (convex vs. flat) across the adaxial leaf surface.  These characters are partially correlated with the leaf tapering from its widest point, and whether they taper symmetrically or asymmetrically to base and to apex.  Leaves shortly tapered to apex  may have parallel margins for most the the leaf length  as seen in T. mairei var. mairei, a leaf shape also referred to as oblong except for its asymmetrical tapered base and apex, the overall shape generally machete-like.  Taxus celebica and T. sumatrana leaves taper to apex starting from well below the mid region of the blade, and except for appearing arcuate near base, their overall shape is lanceolate (sword-shaped).  Leaves of T. mairei var. speciosa  often taper to apex from near the mid region, but their shape is not always elliptical since its leaves are often strongly curved along both sides, appearing sickle shaped. 

     The abaxial leaf midribs  of T. sumatrana and T. mairei are elevated with truncated margins and a central channel. The epidermal cells are usually inflated (mammillose). This is in contrast to a keeled midrib of non-inflated epidermal cells in T. kingstonii, in which its midrib may vary on a plant from an elevated round keel to nearly flush with the leaf surface.

     The species are also differentiated by the dried leaves, particularly the color of stomata bands and puckering of the blade surface to the extent that channels appear along each side of the adaxial midrib (see T. sumatrana below).

     The degree to which leaves spread from a branch is partly related to the habit of the plant and its habitat.  A yew growing on an open rocky slope most likely will have ascending to erect branches from which leaves may not spread out as far as one growing in the forest understory that will likely have more horizontal branches and divergent leaves. It is important to keep in mind that leaves in all species of Taxus generally spread to expose the abaxial surface away from direct light, but the manner in which this is done varies.

     Species are also distinguished by the frequency of leaf development along a branch—as may be measured by distance between adjacent leaves (phyllotaxy). The spiral arrangement of leaves follows a reoccurring pattern (phyllotaxy) along a branch in which leaves reappear along one line, and those along that line may spread differently from those along other lines on the same branch. Their arrangement may be further affected by whether they spread by bending of the petiole and/or by bending or twisting of the blade. These differences appear partially correlated with leaf shape.  Treating these characters independently in multivariate analyses (Möller et al 2013) may lead to biased results.

   An easy to recognize  phyllotaxy is that of Taxus mairei var. mairei.  In many plants, leaves develop at 1 mm intervals with a complete cycle at 6 mm as measured from the junction of blade with petiole.  Its leaf arrangement is also distinctive in that the leaves spread out from the stems by a sharp twist of the blade near base where they, along with their petioles, are adpressed to the stem, rather than spreading from branches strictly by the petiole. It may be noted that leaves of Taxus mairei var. mairei developing on the underside (ventral surface) of stems sharply bend, while those developing more along the dorsal surface sharply twist, a common trait to most species of Taxus.  Not only do the leaves of T. mairei var. mairei align in one plane along two sides of a branch, but they appear almost consistently equidistant and parallel to one another.

     However, phyllotaxy and leaf arrangement in related species are not as distinctive; for example, Taxus celebica from Sulawesi (type) has larger leaves slightly more separated from one another, developing at ~3 mm intervals with a completed cycle at 22 mm, but not all spread in the same direction along one side of the branch; some leaves diverge more by bending of the petioles with their blades appearing relatively straight, while other leaves are arcuate near base.  Taxus kingstonii in Taiwan has leaves developing at 4 mm intervals with a completed cycle at 14 mm, and they crisscross more frequently as seen along one side of the branch, a phyllotaxy  intermediate between that of T. celebica and T. mairei var. mairei.  This type of spread seems related to machete shape of the leaves, in contrast to more linear, symmetrically tapered leaves of T. kingstonii that spread away from the branch more by its petiole.

     Species variation in phyllotaxy has not been quantitatively assessed, because other factors such as leaf shape and size and branch thickness appear correlated, and that genetic differences within a species are also evident.  Nevertheless, phyllotaxy is largely distinctive for distinguishing most species and varieties of Taxus.

     Generally, each species within the Taxus sumatrana group is recognized by a suite of character features from  which one or more are given more taxonomic weight for identification, while the same taxonomic character states do not apply to distinguishing all species in the Taxus sumatrana Group.  For example, phyllotaxy and the lanceolate acuminate leaf shape are considered most important for Taxus celebica, while also appearing correlated with  dried leaves retaining a green color on the adaxial surface and remaining flat across both surfaces.  Taxus sumatrana has similarly shaped but slightly smaller leaves, but taxonomic weight is given to its leaves that—upon drying—become puckered across the adaxial surface and recurved along margins, while also turning dark green to blackish green in color.  Taxus mairei and its var. speciosa—treated as varieties because of their close similarity in leaf anatomy and geographical distribution—are distinguished by many features that include phyllotaxy, leaf arrangement, leaf shape, branching pattern, and seed color and seed shape; taxonomic weight is subjectively applied mostly in that order. Their morphological differences and close geographical association (Spjut 2007b) appears correlated  with molecular data (Gao et  al. 2007, haplotypes #13, #14), but their treatment as varieties by Spjut (2007b) is contrary to Möller et al (2013) who argued that they should not be distinguished unless geographically distinct.  Taxonomic weight for recognizing Taxus kingstonii is given to its rusty orange colored stomata bands in dried leaves, and to its leaves curving downwards lengthwise along the blade, sometimes more strongly near leaf tips, and to the blades not adpressed to stem at base.  This is in sharp contrast to the upturned leaves of T. mairei var. mairei where they are tapered to apex and are adpressed up against the stem at base.  In putative hybrids from northern Yunnan (T. celebica, T. mairei var. speciosa, T. kingstonii), the pale orange leaf color was given the most weight for identifying T. kingstonii.

Examples of Specimens for Species in the Taxus sumatrana Group

1. Taxus kingstonii—What to look for in leaves:
   
   1. Spreading at various angles from branchlets (stems)  more by bending of the petiole than by curving of the blade.
   2. Appearing  convex across the adaxial surface.
   3. Curved downwards lengthwise along the blade, sometimes with a slight twist, especially where they taper to apex.
   4. ± Symmetrically tapered to apex and to base.
   5. Stomata bands relatively broad, in contrast to a narrow region of marginal cells, often discolored rusty orange in herbarium specimens.
   6.  Abaxial midrib elevated into a round keel or flush with the surface as sometimes seen on a single leaf,  occasionally truncated at margins in putative hybrids with T. mairei var. speciosa or T. celebica.

      Branches: Dividing ± equally into branchlets (isodichotomous branching), the main stem not straight, zigzag line.

     Leaves:

1. Spreading closely parallel to one other, slightly overlapping along their margins—to slightly separated—along two sides of a branch in one plane, blade at base bending on ventral leaves, twisting on dorsal leaves.
2. Often shaped like a machete with the margins straight and parallel for than 1/2 of the blade.  Slightly sickle shape in other specimens but then closely parallel and not widely separated.
3. Curved upwards along the margins where tapered to apex but otherwise slightly puckered to flat.
4. Asymmetrically tapered to apex and to base.
5. Stomata bands relatively narrow in contrast to a broad region of marginal cells, often discolored a dried blood red  in herbarium specimens.
6. Midrib elevated with truncated margins and centrally channeled (see illustrations above).

     Seeds: reddish brown to purplis, Hunan. Guizhou

      Branches: Appearing straight with divergent branchlets, appearing more straight than zigzag, occasionally with terminal isodichotomous branchlets.

     Leaves:

1. Spreading along two sides of a branch in one plane, but mostly parallel only in pairs, or often crisscrossing one other, most not overlapping along their margins, many spaced > width of leaf blade.
2. Slightly elliptical to strongly sickle shaped, the margins often curved for than 1/2 of the blade.
3. Generally tapered to apex from near mid region whether straight or curved, weakly convex across adaxial surface.
4. Asymmetrically tapered to apex and to base but less so than in var. mairei.
5. Stomata bands relatively narrow in contrast to a broad region of marginal cells, in dried specimens pale green to yellowish orange, overlaps with T. celebica.
6. Midrib elevated but truncated and channeled (see illustrations above).

     Seeds: pale brown to tan (Sichuan) with a nipple. Guizhou

Leaves:

1. Spreading along two sides of a branch in one plane, but usually parallel only in pairs and also occasionally crisscrossing, many widely spaced, most not overlapping along their margins. Often spreading by petiole.
2. Slightly elliptical to strongly sickle shaped. the margins often curved for than 1/2 of the blade.
3. Generally tapered from near base to apex, mostly flat across adaxial surface.
4. Asymmetrically tapered to apex and to base, but less tapered than T. mairei var. mairei.
5. Stomata bands often very narrow in contrast to a broad region of marginal cells, in dried specimens pale green to yellowish orange.
6. Midrib slightly elevated but truncated and channeled (see illustrations above).

Leaves:

1. Spreading along two sides of a branch in one plane, but usually closely parallel only in pairs and also frequently crisscrossing one other, most not overlapping along their margins. Often the base of blade adpressed to branch, spreading by curving of the blade, and by twisting of petiole.
2. Slightly elliptical to strongly sickle shaped. the margins often curved for than 1/2 of the blade.
3. Generally tapered to apex from near the base, strongly puckered across adaxial surface.
4. Asymmetrically or symmetrically tapered to apex and to base.
5. Stomata bands slightly broader than marginal cells, in dried specimens pale green to yellowish orange.
6. Midrib elevated but truncated and channeled (see illustrations above).

     Möller et al (2013) carved out a “hidden" "new species” they recognized to occur from Yunnan to Vietnam, distinguished by the absence of papillae of the abaxial leaf midrib and by bud-scales persisting at the base of branchlets as related to their determination of one of 19 chloroplast haplotypes (Gao et al. 2007), clip of illustration of type shown below from Möller et al (2013).  The variation in the papillae on the abaxial midrib was discussed in detail  by Spjut (2003) under Taxus chinensis.  

     The lack of papillae on the abaxial midrib suggests that Hengduan Mountain species in question may belong to the Sumatrana Group; however, judging from the phyllotaxy and leaf shape in the illustration (Möller et al 2013), it would appear to belong to the Chinensis Subgroup (Spjut 2007b).  The use of the papillae character should be corroborated by other characters such as leaf epidermal cell shape in cross section and number of abaxial leaf marginal cells without papillae (Spjut 2007a,b), characters that Möller et al (2013) excluded from their study, in which they employed principal component analysis (PCA) for 27 morphological characters based on Möller et al (2007), except for modification to how leaves are inserted. 

     A major flaw with morphometric analyses is that biological species distributions are not linear; rather, they are geometric (Willis 1922).  One might expect character features to cluster for the common species, a biased result in which data for large numbers of individuals sampled randomly may appear convincing when visualized on graphs; however, there is likely to a black hole in the data in considering the "hollow curve" distribution (See also: Spjut, 2010; Review of plants collected for antitumor screening, Fig. 1, p. 18). http://www.worldbotanical.com/images/ARS%20NCI%20Active%20Plants/Review%20of%20plants%20collectod%20for%20antitumor%20screening-Spjut-2010.pdf. The rare species may appear as anomalies.  The results are further biased when one holds the view that species must be geographically separated.

     The persistence of bud-scales can be variable in T. chinensis in that they can persist longer at the base of some branches more than others as evident in this link to a specimen from Hubei.   Among the species in the Sumatrana Group,  T. kingstonii is most similar to T. chinensis. Spjut (2007b) noted that he reidentified his annotations for some specimens of T. chinensis—that were initially identified from examination under a dissecting scope—to T. kingstonii when leaf sections were later prepared and examined under a compound microscope.  Thus, from his point of view the distinction of Möller et al (2013) new species is questionable.  Taxus scutata Spjut ineditus is also similar to the illustration but would seem to differ by  having larger (conspicuous persistent) bud scales and a papillose abaxial midrib (Spjut 2007b, 2013; also this website under the Chinensis Subgroup).

     Further, it may noted that Möller et al (2013) limited their study to mainland China with reference to their earlier studies on Himalayan yew plants (Möller et al 2007).  They excluded specimens from Taiwan, while they also recognize T. mairei to occur outside China such as in northeastern India, where specimens studied by Spjut (2007a,b) from that region were identified  as belonging to either T. celebica, T. kingstonii, or T. sumatrana, but not to T. mairei, which Spjut (2007) considers endemic to southern China, corresponding to Gao et al. (2007) distribution of Taxus haplotyptes #13, #14.  It should be noted that T. celebica and T. sumatrana are earlier names for T. mairei that Möller et al (2013) failed to account for in their paper, possibly because the types come from Indonesia and not China.

     The other species of the Sumatrana  Group appear to occur less frequently in mainland China (Spjut 2007a,b), thus, such rare occurrences may not stand out in PCA, or possibly occur as rare haplotypes in Gao et al. (2007). Additionally, Möller et al (2013) suggest that the disjunct occurrences of T. kingstonii have discordant character features that belong to T. mairei, whereas Spjut (2007b) proposed that the variation he saw in T. kingstonii may be due to hybridization with T. wallichiana in northeastern India, with T. chinensis and T. mairei in Shaanzi, Gansu and Sichuan, and with T. celebica in Yunnan. 

     Both Spjut (2007a) and Möller et al (2013) recognize hybridization between T. wallichiana and T. chinensis on Mount Emei based on different leaf characters and different analytical methods, while Möller et al (2013) did not reference Spjut's (2007a) study.   But their philosophical views on species differ. Möller et al (2013) reportedly look for a geographical fit to their data, drawing on geographical defined their preconceived species (e.g., T. mairei occurring from India to China while excluding T. sumatrana), whereas Spjut (2007a,b) defined species based on reoccurring morphological character traits  regardless of their geographical continuity or discontinuity (Spjut 2000, 2007, 2013). Nonetheless, Spjut's (2007) character features show phytogeographical relationships for defining species groups, species subgroups and the species, which all are supported by molecular data from other studies (J. Li 2001; Gao et al. 2007; Hao et al. 2008a,b; Shah et al. 2008), and included morphological characters that were not employed by Möller et al (2013). And while  Möller et al (2013) may see Spjut's species concept of Taxus as narrow, many disjunct taxa recognized by Spjut (2007b, 2013) in reality reflect a broad species concept by their wide geographical distribution; he (Spjut 2007b) had also stated that his objective was to recognize the fewest species and varieties that could be consistently defined based on key subjectively correlated character features as noted by Möller et al (2013).

Key to Species and Varieties in the Taxus sumatrana Group

References: see Introduction to Taxus, Overview of the genus Taxus, and Spjut 2007a, 2007b.

Spjut, R. W. 2007a. A phytogeographical analysis of Taxus (Taxaceae) based on leaf anatomical characters.  J. Bot. Res. Inst. Texas 1(1): 291–332

Spjut, R. W. 2007b. Taxonomy and nomenclature of Taxus.  J. Bot. Res. Inst. Texas 1(1): 203–289.

Willis, J. C. 1922. Age and area. Cambridge University Press.