Characters
Some characters could not be used yet appear to have potential to be informative.
Bracts (B1s) were frequently not preserved with the sheets examined, despite being a key character used by Kükenthal to differentiate C. papyrus ssp. madagascariensis, and showing strong variation in dimensions, angle in relation to the rays (R1s) and of the midrib. Similarly, many sheets were absent any length of stem, the shape of which beneath the inflorescence has also been used influentially by some authors, e.g. Chrtek and Slavíková (2003).
Three anthers, themselves showing variability in size, demonstrating how the anther appendage may appear to greatly exceed the locules, or be absent. From fresh material. Anthers ± 1mm. This character is difficult to observe on many dried specimens, because the length of anthesis, in particular the longevity of the anthers is short, and because anthers drop off – they are attached by delicate filaments - or wither soon after dehiscence, therefore many do not have visible anthers.
The appearance of anther connective extensions was observed to vary even on the same sheet. However, this may be because they are prone to desiccation and change shape upon withering. Floral dissections and re-hydration are necessary to explore this character further. It is the principal character upon which both Chiovenda and Kükenthal base their monographic treatments; its omission from this study is likely to be a reason for not recovering similar taxa. There was, however, no evidence from this study that it is a consistently reliable character, and it is notable that very visible characters such as ray (R1) length and bracteole (B2) length, were not relied on, and therefore presumably considered unreliable by Chiovenda or Kükenthal.
The Rachilla (R2) wings, heavily relied on by Chiovenda but to a much lesser extent by Kükenthal, are another character which clearly varied, but relies upon destructive sampling to be consistently observed. Even when exposed on dried material, as they only occasionally are, they are invariably twisted, and therefore difficult to measure reliably. These structures are referred to in the literature as scales as well as rachilla wings, their ontology is described by Vrijdaghs et al., 2011, who considered them to be lateral extensions of the glumes which are fused to and enclosing the rhachilla. In practice they are rarely observed to be attached to the glume.
Rachilla Wings (mauve arrows) attached to Rachillas (axes of the spikelets, yellow arrows). The Rachilla wings detach from the glumes easily, since on these spikelets they remain attached to the rachillas after the glumes have been removed. Remaining glumes (green arrow) shows no indication of bearing any wings. Rachilla wing length ± 0.5mm.
Results show that there are other characters, as illustrated in the media gallery, which should also be further considered as they might reliably contribute to separation of infraspecies, including the following.
· Ray prophyll’s (P1) apical teeth present in many sheets, but not all – photographed in the Characters gallery.
· Bracts subtending the spikelets (B3s) and prophylls also subtending the spikelets (P3s) which were not always present, were not measured in this analysis but observed to be variable.
What appear to be prophylls (yellow arrows) subtending the spikelet. Sterile basal bract (red arrow) of the spikelet, which are smaller than the glumes and appear to be sterile. Basal bract ± 0.3mm. An anther appendage can also be seen (green arrow).
· Glume keels (“ridges”) in Cyperus papyrus measurements:Raw Scores – clearly some specimens had keeled glumes, others did not. Very often discoloured, as shown here.
The glume keels (“ridges”, blue arrows) are prominent on this sheet, and appear to be discoloured, this is typical.
· Glume shape and apex – needs floral dissections to be observed effectively.
· Anther length. Cf. for example photographs of S37, on which the anthers appear to be long compared to other sheets. This character was not measured.
· Stigmas and style: these were categorised to obtain an idea for how they might vary, however, clearly they did and separate characters could be defined and measured to determine if the variation is significant. Kükenthal (1935) makes use of this character to a limited extent.
· Some specimens had very thin secondary bracts (thin B2s, character 18), seemingly arising from amongst the spikes, rather than enclosing them, as the regular B2s do, invariably with strongly recurved teeth.. Their presence was clearly not consistent across all of the specimens, as shown by the results
A thin bract (blue arrow) emerging from amongst the spikes. Clearly distinct B2s (secondary bracts) surround the spikes (pink arrows).
·
See also Cyperus papyrus measurements:Scoring Notes., several further characters are discussed.
Many characters appeared to vary considerably, even on the same sheet. This is amply highlighted in the Characters media gallery. Spikelet length is a clear example, the shape and length appearing to change with maturity. This suggests that both more specific measurements are needed to ensure consistency, and also that more complex characters need to be used – to recognise this variation. An attempt was made to do this in the case of the length of the Rachillas (R2s) which were divided in to sessile, medium length and long, on the same sheets, for this reason.
Material
Some specimens were of course well collected, representative and complete. However a number were not. It is likely that the size of this species may have led to the selection of smaller, less representative material simply for ease of accommodation on preservation sheets. Certainly its size has led to the removal of proportions of the inflorescence up to ±80% in some cases, along with key characters, such as the B1s mentioned above. Another key character occasionally but notably affected is the B2s, which on at least one sheet were simply cut off around the edge of the mounting sheet.
It cannot be assumed that the sample used in this study was entirely representative, however on its evidence there were very few recent collections, 30 of the 36 are more than 40 years old, only two collected since 2000. And, despite the random selection, and deliberate policy of avoiding type material, 4 of the 36 were specimens also cited by Kükenthal (as indicated on the specimen list) and 2 even by Chiovenda (Meyers and Dinsmore 11/7/1912, Petherick, Brownell etc. 4/4/1862).
From observations of fresh material, it appears likely that informative characters might occur on parts of the morphology usually ignored in collections, notably the sheathing leaves. As can be seen from the photograph below, some plants appear to have leaf blades, perhaps only when young, despite their absence being one of the critical characters at species level in separating C. papyrus from others (e.g. Haines, Wheeler & Lye, 1983, Chiovenda 1931, Kükenthal 1935).
Plants labelled as C. papyrus, seemingly with long leaf blades on the young stems. A leaf blade is arrowed.
Two sheets, S20 Harrison 1019 and S25 Eggeling 381, were observed to have produced florets yet were apparently entirely sterile. The species is known to produce entirely sterile culms, but these do not have any spikes or other floral parts and so are a quite different phenomenon. Sterility of florets has not been commented on in any of the literature observed. In Fig. RD5, the MCA results, they can be seen to cluster together as sister sheets to one of the two main clusters in group 1. A more detailed analysis might omit these sheets to determine the extent to which they may be affecting the results. As the plant generally grows in resource-plenty conditions, usually nutrient rich marginal environments in the tropics and sub-tropics, it might be speculated that nutrient or other resource stress are not the cause of the sterility. If this is so then hybridity is a possibility. Again, further analysis is required to determine the cause and impacts of these observations.
In conclusion, fresh material, collected mindfully to the key characters, is needed for a further, more intensive study of C. papyrus.
Given the number of outliers in the Box plots (Fig. RD6), several sheets should be reviewed to ensure correctly measured, and to determine the reasons for such variation in the measurements. This may also be a result of the smallness of the sample size.
Unique Sheets
The following sheets were found to possess unique characters within the sample sheets.
S26, Wheeler Haines 4229, has bulbils in the inflorescence. Destructive dissection is needed to be certain of the morphology. However, there is no record of this in the literature, so it may be an aberration, or may represent a consistent feature, that might be found in other preserved material.
S37, Simpson 88/91. This sheet is identified as a separate group in the results, both the PCA Score Plot (Fig. RD2b.) and consistently the MCA identify this sheet as outstanding. As well as the characters used in these analyses which are clearly different about this sheet, including the arrangement of the inflorescence, extremely reduced bracts (B1s), long spikes, length of anthers (not scored) etc., it appeared to have longer, thinner anthers than generally observed.
S17, Milne-Redhead 4112. This has bracteoles subtending the spikelets (B3s) which are longer than the spikes and attenuated. This was not observed in any other material. It is also the only specimen from Angola.
S9, Quarre, 5739. It was observed that on this this sheet that the anthers appeared to be frequently entirely red, compared to the usual straw / brown colour observed on other sheets.
That such variation was observed in a small sample not only permits of the conclusion that the sample contained distinct taxa, but also that the collections at Kew, and probably therefore elsewhere, have not been fully studied and may hold more taxonomic variety.
Multivariate Cluster Analysis - Groupings
Kükenthal and Chiovenda’s subspecies descriptions, including Chrtek and Slavíková’s ssp. hadidii can be summarised as follows.
· __ papyrus
Anther connectives beyond locule apices 1-3 X anther breadth. Glumes narrowly elliptical. Spikelets 1-3cm. (Translated after Chiov., 1931).
· __ antiquorum
Anther connective not or scarcely extending beyond tips of anther locules; and shortly arched. Spikes lax, shortly pedunculate. (Translated after Kükenthal, 1935). Anther connectives not or scarcely beyond locule apices. Spikelets 5-10mm. 6-10 florets. Glumes elliptical-rotund. (Translated after Chiov., 1931)
· __ nyassicus
Anther connective not or scarcely extending beyond tips of anther locules, linear-cylindrical. Glume apices subacute. (Translated after Kükenthal, 1935). Anther connectives beyond locule apices cylindrical-narrow. Glumes more acute. Raceme lax, peduncled. Spikelets often elongate 1-3cm, many flowered. (Translated after Chiov. 1931).
· __ ugandensis
Anther connective not or scarcely extending beyond tips of anther locules, ovoid or subglobose, red. Glume apices rounded. (Translated after Kükenthal, 1935). Anther connectives beyond globoid to sub-globose. Racemes lax. Translated after Chiov. 1931).
· __ madagascariensis
Anther connective not or scarcely extending beyond tips of anther locules. Culms triquetrous. Bracts narrow. Bracteoles anthelulas barely exceeding [spikes] usually much shorter. Glumes often intensely brick red. Spikelet wings acute. Translated after Kükenthal, 1935)
· __ zairensis
Anther connective 2-3 X anther width, subulate whitish. Spikelet wings acute. Spikes conspicuously pedunculate. Spikelets 3-7mm. Glumes broadly elliptical. (Translated after Kükenthal, 1935). Anther connectives beyond locule apices 1-3 X anther breadth. Glumes broadly elliptical, apex broadly rounded. Spikelets 3.5mm. (Translated after Chiov., 1931).
· __ hadidii
Culms smooth, rounded to inconspicuously triangular. Anthela composite, many rays. Anthelulae dense, with 3 spikes, spikelets shortly pedunculate or subsessile, 6-8 (-10) mm long, pitch-black to brown. Glumes densely imbricate, adpressed, elliptic-ovate to elliptic, apex bluntly acute, violet-striped. Anther connective extending beyond anther apices, three times or four times the length of the shorter side of the anther apex. (Translated after Chrytek & Slavíková, 1977).
The groups determined in this analysis through MCA, as described in Results (page Descriptions) can be summarised as follows. N.B. group 4 is omitted as it is not considered a natural one. As discussed in the results, S17 from Angola appears unique and therefore might skew the clustering. S35 in this group is Madagascan, and notably this group falls close to sheets from South Africa in the data analysis, e.g. Fig. RD7, and as sister to Groups 1 and 2 in the MCA, e.g. Fig. RD5. It is likely, therefore to have clustered closer to the other Madagascan sheets which it superficially resembles, on further analysis. Nor is Group 6 described here, as for the many reasons discussed above, it is a single, unique sheet.
· Group 1
Culm shape unknown; rays (R1) short-mid length, many; 3-4 (-5) spikes rarely sessile, rachis’ (R2) usually mid-length; bracteoles (B2) mid-length to long; glumes mid-length; rachilla wings long; spikelets short, few flowered, short, lanceolate-ovate.
· Group 2
Culm trigonous; rays (R1) mid length, many; spikes variable, 2 -5, rarely sessile ,rachis’ (R2) often long,; bracteoles (B2) consistently long; glumes short; rachilla wings long; spikelets short, few flowered, short, lanceolate-ovate.
· Group 3
Culm trigonous; rays (R1) long, not crowded; spikes usually 4, occasionally sessile, rachis’ (R2) usually mid-length, occasionally long; bracteoles (B2) consistently long; glumes short; rachilla wings short; spikelets short, many flowered, short, lanceolate-narrowly lanceolate.
· Group 5
Culm triquetrous; rays (R1) long, rays (R1) long, sparse not usually crowded; spikes few, often sessile else rachis’ (R2) mid-length; bracteoles (B2) various; glumes long; rachilla wings short; spikelets long, many flowered, lanceolate-parallel sided..
Kükenthal regards lax spikes and short peduncles (=R2s) as reliable notably for ssp. antiquorum, this accords with the findings for Group 1. These characters alone, however, should not be relied upon. Conspicuously long peduncles are notable in ssp. zairensis, which accords with Group 2 from these results.
Kükenthal also uses spikelet length as a reliable character in some cases. It was observed to vary considerably on the same sheet in this study. This character seems certainly to vary with the relative maturity of the spikelet, and possibly according to position within the spike and with the spike’s position in the inflorescence. Therefore, more sophisticated and hence more consistent measurement, would be necessary to determine if it is reliable.
Material from the reportedly introduced populations in Sicily and Israel do not stand out as separate groups in any analyses. This upholds their status as introductions.
However, it can be speculated that group 5 has affinities with ssp. madagascariensis, because of the long glumes, and long, narrow, sessile spikelets which make it stand out, as well as the sharply triquetrous stems described by Kükenthal, the absence of other key characters of the anthers and bracts (B1s) detracts from this conclusion, but given the geographic source of the sheets in this Group and the known high endemism on Madagascar (84% of species, according to Buerki et al. 2013), it is likely to be upheld in a more extensive analysis.
Results of measurements and data analysis show there are clearly one or two separate groups from Madagascar, possibly including South African and Mozambiquan material. Further sampling and analysis is needed to determine the taxon delimitations and whether one or two are present separate from the bulk of C. papyrus, i.e. whether S37 is part of the Madagascan taxon delimited by Kükenthal an unnamed taxon, or even a misidentification.
Conclusions
Based upon measurements and observations clearly there is enormous variation within C. papyrus. This has led to the delineation of subspecies or species by Kükenthal and Chiovenda respectively. It was not found conclusively in this study that the variation is discontinuous, and clearly some characters, e.g. ray (R1) length and bracteole (B2) length vary, probably continuously. However, sufficient evidence of consistent variation was identified in the PCO and MCA to warrant further investigation. A larger sample is needed, although this sample compares to that of Chiovenda, clearly it is inadequate to represent the species, given its very broad variation and geographic distribution, and in places such as Angola and Madagascar, likely geographic isolation. With additional specimens and better material, data analysis in more depth can be undertaken using a better considered character set. This analysis does give good pointers towards characters which may be useful, albeit they were found as the study progressed and not in time to influence its outcome.
It is likely that following a full revision of this species that some of the infraspecies published will be upheld, but also that new taxa, at either infraspecies or species level may be segregated. Clearly molecular analysis will have a big role in supporting and enhancing the morphological evidence that would be obtained.
Overall there was too much variation in the quality of collecting in the sample, resulting in potentially atypical material and the absence of key characters for conclusive results to have been obtained, even with further data analysis.
Anther connective shape and exsertion beyond the anther locules is the most important character used by Chiovenda and Kükenthal to differentiate taxa. It was quickly observed, as confirmed by the images in the Characters media gallery, that in dried material this character is both difficult to observe and variable on the same specimen. In the Flora of Palestine, Feinbrun-Dothan (1981) also rejects this character as inconsistent. Re-hydration of the material might be undertaken to determine if the character is consistent when scrutinised, and informative, as clearly it does exhibit variation. However, it is not certain that its availability would have produced more reliable results, nor that its variation would reflect that of the other major morphological characters such as secondary bract (B2) length.
Cyperus; Cyperus papyrus
Cyperus; Cyperus papyrus