AGROPYRON CANINUM AND ITS NORTH AMERI¬ CAN ALLIES. ARTHUR STANLEY PEASE and ALBERT HANFORD MOORE. [Reprinted from Rhodora, Vol. 12, No. 136, April, 1910] A AGROPYRON CANINUM AND ITS NORTH AMERICAN ALLIES . 1 Arthur Stanley Pease and Albert Hanford Moore. The genus Agropyron is, as is well known, a very puzzling one. Two recent revisions 2 have, in the opinion of the writers, not entirely cleared up the difficulties presented by the relatives of Agropyron caninum (L.) Beauv. The difficulty of indicating the rank and rela¬ tionship of the members of the group by a series of independent species, as has generally been attempted, arises from an extraordinary amount of intergradation. Agropyron caninum passes, by the loss of its awns, to A. tenerum Vasey; the latter to A. violaceum (Hornem.) Lange, by the shortening and often thickening of the spikes; this again, on the one hand to A. violaceum (Hornem.) Lange var. latiglume Scribn. & Sm., by the flattening out of the keel and the presence of pubescence on the spikelets, or to A. violaceum (Hornem.) Lange var. andinum Scribn. & Sm., by the presence of awns. Forms in which the awns are strongly bent outward have been described as Triticum caninum L. var. Gmelini Ledeb., which may also be regarded as Triticum 1 The authors desire to express their thanks to the staff of the Gray Herbarium for the privileges and valuable suggestions received from them. 2 F. Lamson Scribner and J. G. Smith, Native and Introduced Species of the Genera Hordeum and Agropyron. Studies of American Grasses, U. S. Dept. Agr. Div. Agrost. Bull. no. 4, 25-23 (Feb. 6, 1897); C. V. Piper, Agropyron tenerum and its Allies, Bull. Torr. Bot. Club, xxxii, 542-547 (Oct. 21, 1905). 62 Rhodora [April or Agropyron caninum with long, recurved awns, as, indeed, Ledebour considers it. Again, by a lengthening out of the awns, there is an imperceptible transition from A. caninum to A. Richardsoni (Trin.) Schrad., as defined by Scribner and Smith . 1 It is also possible to view A. violaceum (Hornem.) Lange, together with its varieties lati- glume and andinum, as short-spiked representatives of A. caninum. Another noteworthy feature of the group is a striking parallelism, in that almost every variety has a smooth and a hairy development. The Agropyra in question belong to the non-stoloniferous group. While this distinction, based on the presence or absence of stolons, may seem an artificial one, it is so generally considered important in the Gramineae and Cyperaceae that we cannot with Ascherson and Graeb- ner subordinate A. violaceum to Triticum repens , 2 nor can we follow Beal , 3 who describes A. repens (L.) Beauv. var. tenerum (Vasey) Beal . 4 The earliest specific name in the group is A. caninum (L.) Beauv., which goes back to the name Triticum. caninum L., of the first edition of the Species Plantarum. Linnaeus here discusses two species, T. repens and T. caninum. Of the former he says, “radice repente,” of the latter, “calycibus subulatis quadrifloris aristatis.” The authors of the synonyms of T. caninum cited by him all indicate plants with aristae. The first synonym given is Gramen caninum non repens elatius spica arisiata Moris . 5 The two remaining synonyms contain the word,“repens,” and the accompanying descriptions state that the plants are stoloniferous; the species is therefore a mixture, as so many of the Linnaean species are. The first mentioned synonym and the type-description are clear, and must hence be taken as conclu¬ sive. Morison says that the stem of his plant is glabrous, but is not explicit as to the leaves. Buxbaum repeats the description given by Morison. Both give figures — Buxbaum’s being especially good — each of which plainly represents a glabrous species with aristae of medium length, such as A. caninum has been held to be. Linnaeus was not satisfied with his first treatment, so that in the Flora Suecica 6 1 U. S. Dept. Agr. Div. Agrost. Bull. no. 4, 29 (Feb. 6, 1897). 2 T. repens L. B. T. biflorum (Brign.) A. & G. A. virescens (Lange) A. & G. Syn. Mittel- europ. Flora, ii, 1, 654 (Dec. 10, 1901). 3 Grasses N. A. ii, 637 (1896). 4 It is perhaps worthy of note that many of the variations described below, based upon length of spikes and of awns and presence of pubescence and glaucousness, are paralleled, in greater or less degree, by variations in A. repens (L.) Beauv. 6 PI. Hist. Univ. Oxon. iii, 177, t. 1, fig. 2, sect. 8 (1715); Buxb. PI. minus cogn. circ. Byzant. Orient. Observ. cent, iv, 29, t. 50 (1733). 3 Ed. II, 39 (1755). 1910] Pease & Moore,— Agropyron caninum and Allies 63 he calls the plant Elymus caninus, a name taken up by a number of subsequent authors. Under these three names, Triticum caninum , Elymus caninus, and Agropyron caninum, our plant is discussed in a multitude of European and American works. Nowhere does there seem to be the slightest difference of opinion as to its nature. The name Triticum sepium Lam. 1 2 seems to be a synonym of Agropyron caninum, and has generally been so regarded. The name Festuca nutans Moench? also seems to be properly included here. Moench' himself says of his plant, “An Elymus caninus Linnaei.” Very closely related to A. caninum (L.) Beauv. is A. tenerum Vasey. It differs from it in no important character, except the absence of awns, and this is not constant, for specimens are not rare which it is difficult to place. Some attempt has been made to distinguish eastern plants from it on the one hand, and from A. violaceum (Hornem.) Lange on the other. Scribner 3 unites them as a species under the name of A. Novae-Angliae. An extremely painstaking examination of the subject has led us to agree with Hitchcock 4 that'this species cannot be. clearly separated from the A. tenerum of the West. It may be in¬ teresting to compare what Scribner says: “In regard to the Agropy¬ ron for which I have proposed the name Novae-Angliae, I must think it quite distinct from the A. tenerum of the west, although it suggests that species somewhat in habit. It looks more like a caespitose form of A. repens. I believe it has been included under A. violaceum by some authors, but it certainly is distinct from that species, as repre¬ sented in our herbarium, from northern Europe and from the Rocky Mountain region.” 5 6 “The plant described in Gray’s Manual, ed. 6, 672, is the western form which is doubtless true Agropyron violaceum, Lange, while the variety mentioned ‘with longer usually pale narrow spikes and attenuate often long-awned glumes,’ is the plant here de¬ scribed as Agropyron Novae-Angliae.” 6 Thus it is seen that Scribner himself calls attention to the affinity of the eastern plant. Through it, A. tenerum Vasey passes readily to A. violaceum (Hornem.) Lange. Scribner and Smith 7 have described Agropyron violaceum (Hornem.) 1 Flore Frang. iii, 629 (1778); Encycl. Meth. (Bot.) ii, 563 (1786). 2 Meth. PI. Hort. Bot. Agr. Marburg, 191 (1794). 3 In Brainerd, Jones, and Eggleston, FI. Ver. 103 (1900). 4 Gray, Man. ed. VII, 167 (1908). 5 In Brainerd, Jones, and Eggleston, Flora Vt. 9 (1900). 6 L. c. 103. 7 U. S. Dept. Agr. Div. Agrost. Bull. no. 4, 30 (Feb. 6, 1897). 64 Rhodora [April Lange var. latiglume, the salient feature of which is wider glumes covered with a fine pubescence. In other members of our group the glumes are merely minutely ciliate on the margins. In most of the group it is difficult to measure the glumes, because of their shape, for they are folded strongly so as to have a prominent keel. In A. viola¬ ceum var. latiglume the keel is much flattened or almost wanting; that is, the glumes are, as it were, unfolded, thus giving the appear¬ ance of width. This fact is certainly more striking than any actu¬ ally greater breadth, even if that exists. The variety was described from the West, but all of the Arctic material in the Gray Herbarium, except one violet-spiked specimen of A. violaceum, answers to the description of it. This suggests the bare possibility that A. violaceum (Hornem.) Lange var. virescens Lange may belong here. The intimate relationship between A. violaceum (Hornem.) Lange and A. caninum (L.) Beauv., referred to above, is brought out by Beal, 1 who says, “Perhaps this [A. violaceum ] is a northern or alpine form of A. caninum R. & S., from which it differs in having a culm 30-50 cm. high, leaves 3-4 in number, 2 blades proportionately shorter, spikes 3-8-12 cm. long, narrow, strict, more or less tinged with violet. Spikelets 3-5 flowered; floral glume wider than in A. caninum, not so firm, nerves more prominent, awn short or none; palea shorter than the floral glume, pectinate-ciliate.” Even these differences do not all hold. A violet tinge is occasionally found in A. caninum, while entire absence of violet color is common in A. violaceum’, the palea- and glume-characters are not good. Triticum violaceum Hornem. 3 is the first published name, but the first varietal appella¬ tion, which we believe expresses the rank best suited to this plant, is Triticum biflorum Brign. /? Hornemanni Koch. 4 Plants with greener, more stocky spikes have been distinguished as Agropyrum violaceum (Hornem.) Lange ft. virescens Lange. 5 This tendency is the common one in the White Mountains, while the other is known from the Arctic Region and the West. However, the White Mountain development is not unknown in the West, where one also finds the slender-spiked, narrow-leaved plant of Greenland, probably 1 Grasses N. A. ii, 639 (1896). 2 This character we find to be inconstant. 3 Flora Dan. xii, t. 2044 (1834). 4 Syn. Flor. Germ. Helv. ed. II, ii, 953 (1844). Koch bases his variety directly upon Triticum violaceum Hornem. 5 Consp. Flor. Groenf. 155 (1880). 19101 Pease & Moore,— Agropyron caninum and Allies 65 representing the type, 1 together with a thick-spiked plant having a strong violet color. Every variation in leaves from revolute to flat and broadish is to be noted. Not one of the members of the group under discussion fails to show an indefinite violet tinge at times. The color shades off into the colorless or green condition too imperceptibly to give rise even to color forms. The tendency is a very common one throughout the Grasses, so that it does not seem noteworthy. Following Piper, 1. c., Hitchcock, in the seventh edition of Gray’s Manual, 2 uses for Agropyron violaceum (Hornem.) Lange the name A. biflorum (Brign.) R. & S. The original publication of Tritieum biflorum by Brignoli was not accessible to us; but fortunately it is copied by Roemer and Schultes. 3 The conspicuous point in the origi¬ nal description, reiterated by them, is that the glumes are 3-nerved. The glumes of our American plants are 5-7-nerved. This distinction has been held by monographers of the genus to be important, yet, although Hitchcock credits our plant with having 5-7 nerves he calls it A. biflorum. Another, not insignificant objection to accepting the name is the peculiar range. Roemer and Schultes note it from, “Monte Matajur, 4 ’’ and while later European authors extend this range a little, all that we have found agree that it is a quite local plant. 5 European authors, who should be more familiar with A. biflorum than the meagre representation of rare European plants in American herbaria permits us to be, do not generally hold the view 7 that A. biflorum and A. vio- laceum are identical, but only closely related. Ascherson and Graebner, for example, treat the latter as two varieties of a subspecies, biflorum, of Tritieum repens L., A. virescens and B. Hornemanni. Nyman 6 believes that Tritieum biflorum is distinct, and says, “gramen rarius.” Ascherson and Graebner are of the opinion that Nyman’s plant is virescens. The following quotation from their Synopsis is indicative of the uncertainty which surrounds the name T. biflorum: “Eine sehr kritische Pflanze, die trotz ihrer fast stehts sehr eigenartige Tracht kaum durch scharfe Merkmale von T. eu-repens getrennt ist.” These authors also note the three nerves, “Hiillspelzen lanzettlich (wenigstens 1 At least, there is a specimen of this phase in the Gray Herbarium which was collected in Greenland by J. Vahl, the collector of the type. It may well be the material of the type collection itself. 2 Ed. VII, 166 (1908). s L.— R. & S. Syst. Veg. ii, 756 (1817). 4 Near Gorz. 5 The reference of it to Syria by the Index Kewensis is probably a mistake. 6 Consp. Flor. Europ. 841-842. 66 R hod ora [April die untere) 3-nervig.” In view of all this, it does not appear justi¬ fiable to consider the name T. biflorum synonymous, by the Vienna Rules, with A. molaceum. Furthermore it is not necessary to do so, because, for us, the first name which can be taken up in the varietal rank, is, as we have shown, Triticum biflorum Brign. [3. Hornemanni Koch. Transitional material to Agropiyron tenerum Vasey is especially common from the Gaspe Peninsula. In fact, if this region alone were taken into account, one would be inclined not to make any distinction at all. A co-type of A. brevifolium Scribn. 1 needs a word of comment. It differs from the ordinary forms of A. molaceum only in having many sterile shoots; the length of the leaves is unimportant. The larger / number of shoots hardly constitutes a character, so that the name A. brevifolium may well be relegated to synonymy. The next variety which needs to be discussed is the one described in correct form under the genus Triticum as T. caninum L. var. Gmelini Ledeb. 2 Oddly enough, the correct combination has been actually used, though not technically made, for in the Bulletin of the Torrey Botanical Club, 3 Vasey speaks of Agropyrum caninum (L.) Beauv. var. Gmelini Ledeb., a combination not made by Ledebour, and so far as we are aware, not formally published by anyone. According to the Vienna Code, a combination is not considered made unless reference is given to a previously published name, which in this case was not done. The three last mentioned plants may best be treated as parallel varie¬ ties. There exists a pubescent form of Triticum caninum var. Gmelini. The type of Agropyron Gmelini (Ledeb.) Scribn. & Sm. var. Pringlei Scribn. & Sm. in the Gray Herbarium, as well as most of the speci¬ mens so labelled, is evidently this form. No other constant difference from A. Gmelini has been detected. A specimen in the Gray Her¬ barium labelled A. Scribneri Vasey, which also does not differ essen¬ tially from our plant, is noted, on the sheet, by the word, “Type.” In fact, the affinity is recognized by Vasey, who says of it, “It is per¬ haps the A. caninum, var. Gmelini, Ld.” As there may be some doubt about the identity of A. Scribneri, since the specimen labelled 1 U. S. Dept. Agr. Div. Agrost. Bull. no. 11, 55 (July 20, 1898). 2 Flora Alt. i, 118 (1829); figured, Ic. PI. Nov. imperf. cogn. Flor. Ross, iii, 16, t. 248 (1831). 3 x, 128 (Dec. 1883). 1910] Pease & Moore,— Agropyron caninum and Allies 67 ‘type’ is not technically such, Vasey not having cited any type, it seems wisest to derive the name of the form from A. Gmelini var. Pringlei, the identity of which cannot be questioned. The last plant to be considered is that called by Scribner and Smith Agropyron Richardsoni (Trin.) Schrad. The evidence for identifying this name with the plant in question is extremely weak. The original description of Triticum Richardsoni Schrad. 1 reads as follows: “Agro- pyr. Richardsonii Ind. sem. H. Berol. 1832.— America borealis arctica ? A- — A Trit. repente vulgari differt foliis firmioribus serius involutis spica graciliori; spiculis minus distichis paucifloris, floribus longius; aristatis.” It is true that the use of the expression, “Trit. repente,” may not be accurate, but is there any reason for assuming this ? Surely, “longius aristatis,” is a mild statement for the longest awns in our group. Besides the published evidence, there is some supposed further documentary evidence. Scribner and Smith 2 say, “ Triticum richard¬ soni Trin. 3 in Reliq. Scrad. [sic.], Linn., 12: 467 (1838), according to a specimen from the St. Petersburg Academy in the Gray Herbarium.” The only specimen of this species from St. Petersburg in the Gray Herbarium is labeled: “Ex herbario horti Petropolitani, ‘Triticum Richardsoni’ = T. violaceum, Hornem.” The expression, “= T. violaceum, Hornem.,” is added in the handwriting of Dr. Gray, and the quotation marks seem to be in an ink of the same date, leaving “Triticum Richardsoni,” carelessly written in an unknown hand, as the only original label. Under “A. richardsoni,” 4 Scribner and Smith cite as seen in the Gray Herbarium: “British Columbia: Rich¬ ardson, type collection.” A careful search has revealed no such specimen. Probably “ Gray Herbarium” is an error. There is, however, a good name for the plant under discussion, namely, Agropyrum caninum (L.) Beauv. var. unilaterale (Cassidy) Vasey; 5 A. unilaterale Cassidy. 6 Cassidy’s description follows: “Flowering glume finely scabrous, flattened and obscurely nerved below; apex rougher, acutely two-toothed, the mid-nerve terminating in a rough, straight awn 2 inches long. 7 Banks of streams 7,000 to 1 Linnaea, xii, 467 (1838). 2 U. S. Dept. Agr. Div. Agrost. Bull. no. 4, 29 (Feb. 6, 1897). 3 There is no evidence that Trinius is the authority. 4 The original authority for A. Richardsoni is Hort. Berol. It appeared as a nomen nudum in the Ind. Sem. Hort. Bot. Berol. 1831 Collect (1832), a copy of which A. H. Moore was permitted to see at the Berlin Botanical Museum. 5 Contr. U. S. Nat. Herb, i, 279 (Oct. 31, 1893). 6 Colo. State Agr. Coll. Exper. Sta. Bull. no. 12, 63 (July, 1890). 7 We wish to call attention especially to his measurement of the awns, which is even rather extreme. 68 Rhodora [April 8,000 ft. [Colo.].” This description leaves little doubt as to the mean¬ ing of the name, and, indeed, Scribner and Smith consider it a synonym of A. Richardsoni. As the first varietal name Vasey’s combination is the correct one. The following quotations will throw some light on the nature of A. caninum (L.) Beauv. var. unilaterale (Cassidy) Vasey, and also upon its relationship. Ramaley 1 says: “From this description [Vasey’s] and from an examination of specimens distrib¬ uted by the Department of Agriculture it appears that this plant is very near to A. violaceum (Hornem.) Lange, forma caninoides 2 described below; it is probable that through these formae 3 the two species 4 are related.” Ramaley, having apparently read only Vasey’s brief description, 5 assumes that A. caninum var. unilaterale differs only very slightly from the type. Scribner and Smith, under A. caninum (L.) Beauv., write: “Forms of this, with unilateral spikes, have been referred to A. richardsoni , which has awns three or four times as long as those of caninum, and the flowering glume bidentate below the origin of the awn. Slender forms have been referred by collectors to A. tenerum Vasey, and forms with short compact spikes and short awns to A. violaceum Vasey.” Scribner and Smith 6 describe Agropyron tenerum Vasey var. longi- folium. We have seen no specimens of this, and are therefore unfor¬ tunately unable to discuss it. We may now proceed to the taxonomic treatment and diagnoses. We desire to call attention to the fact that the measurements in all cases represent the usual or average limits, and that measurements may be observed which lie above or below them. For example, if the awns are said to be 1-3 cm. long, some awns may be found 0.8 or 4 cm. long, but this would not apply to more than a small proportion of the awns in the given spike. Approximate averages are given whenever possible, to show the proportional size of the spikes, which is an important character. 1 Geol. Nat. Hist. Surv. Minn.— Minn. Bot. Studies, Bull. no. 9, 3, 107 (June 9, 1894). 2 We believe this form to be inseparable from A. caninum. It does not seem best to retain names for intermediate forms of this sort, where no sharp line exists, for there is no more reason for using one name than several. 3 A. violaceum (Hornem.) Lange f. caninoides Ramaley and A. violaceum (Hornem.) Lange f. violacescens Ramaley. The latter is Ramaley’s name for A. caninum var. unilaterale. 4 Ramaley means here A. caninum and A. violaceum. 5 Vasey does not mention the awns. 6 U. S. Dept. Agr. Div. Agrost. Bull. no. 4, 30 (Feb. 6, 1897). 1910] Pease & Moore,— Agropyron caninum and Allies 69 KEY TO THE VARIETIES AND FORMS. 1 a. Awns mostly under 1.5 cm. long, or none. b. b. Awns mostly 7-15 mm. long. c. c. Plants glabrous.1 .A. caninum c. Some or all of the leaves or sheaths hairy, d. d. Plants not glaucous.2. A. caninum f. pubescens d. Plants glaucous.3. A. caninum f. glaucum b. Awns none, or not exceeding 6 mm. e. e. Spikes generally long and slender (average length about 12-14 cm.): usually lowland plants, f. f. Plants glabrous.4. A. caninum var. tenerum. f. Leaves, sheaths, or both provided with hairs, g. g. Plants not glaucous . 5. A. caninum var. tenerum f. ciliatum. g. Leaves and sheaths glaucous. 6. A. caninum var. tenerum f. Fernaldii e. Spikes generally short and thick (average length about 7-8 cm.): usually alpine or far northern, h. h. Glumes with a fine pubescence. 7. A. caninum var. latiglume. h. Glumes minutely ciliate on the margins, not pubescent, i. i. Leaves glabrous .... 8. A. caninum var. Hornemanni. i. Leaves pilose above. 9. A. caninum var. Hornemanni f. pilosifolium. a. Awns mostly 1.5 cm. long or over. j. j. Spikes short and thick, average length (exclusive of terminal awns) 5-7 cm. k. k. Awns ascending, not strongly recurved. 10. A. caninum var. andinum. k. Awns strongly recurved (often bent out nearly at right angles). 1. 1. Plants glabrous.11. A. caninum var. Gmelini. 1. Leaves (rarely sheaths also) provided with hairs. 12. A. caninum var. Gmelini f. Pringlei. j. Spikes (inclusive of awns 2 ) 7.2-20 cm. long. m. m. Plants glabrous.13. A. caninum var. unilaterale. m. Sheaths and often leaves provided with hairs. 14. A. caninum var. unilaterale f. ciliatum. 1. Agropyron caninum (L.) Beauv. Spikes long and slender, 7.2-21.6 cm. long (exclusive of the terminal awns), 3-10 mm. wide (exclusive of the awns); awns 5-25 mm. (mostly 7-15 mm.) long, 3 generally spreading, not rarely ascending (seldom giving the spike a unilateral appearance).— Agrost. 102, 146 (1812); L.— R. & S. Syst. Veg. ii, 756 (1817); Reichenb. Ic. Flor. Germ. Helv. i, 29, t. 19, fig. 1381 (1850) ; 4 Scribn. & Sm. U. S. Dept. Agr. Div. Agrost. Bull, no. 4, 29 (Feb. 6, 1897). Gramen caninum non repens elatius spica aristata Moris. PI. Hist. 1 Any key to this group would fit only the more typical cases, because of the intergrada¬ tion. 2 The awns are included because their ascending or approximate character makes them an integral part of the spike. 3 If the low measurements (5-6 mm.) hold throughout a spike, the specimen will be better treated as no. 4. 4 Color of the awns purplish. 70 Rhodora [April Univ. Oxon. iii, 177, t. 1, fig. 2, sect. 8 (1715); Buxb. PL minus cogn. eirc. Byzant. Orient. Observ. cent, iv, 29, t. 50 (1733). Triticum caninum L. Sp. PL ed. I, 86 (1753); Zinn, Cat. PL Hort. Acad. Agr. Gott. 67 (1757); Sm. Flora Brit, i, 159 (1800); Host, Ic. Descript. Gram. Austr. ii, 20, t. 25 (1802); Engl. Bot. t. 1372 (1805); Flora Dan. ix, 1, t. 1447 (1818); Sm. Engl. Flora, i, 184 (1828), etc. Elymus caninus L. Flora Suec. ed. II, 39 (1755), Syst. Nat. ii, 879 (1760), Sp. PL ed. II, i, 124 (1762); L.— Willd. Sp. PL i, 469 (1798); Sehkuhr, Bot. Handb. i, 56 (1791), etc. Triticum radice fibrosa , foliis hirsutis, locustis quinquefloris, aris- tatis Haller, 1 Hist. Stirp. Indigen. Helv. 212 (1768). T. sepium Lam. Flore Frany. iii, 629 (1778), Encycl. Meth. (Bot.) ii, 563 (1786); Lam. & DC. Flore Fran