jgflfiwi as 30308 gnaw-.03 3. mfiuInuflmmoIérawfiofi... I . . ......t30...0¢0. ......“ ..mfl3m_3.0...0 ....I. .. . . .. . 3.. . . ._ .. . . I I I. I 0 . .3 u 0 OIJOW 03:v.i3 Cl 0.03. 403(3...3. 3I0... ....I I _ I .I.... ...I. .19.... .... ......I..._ 3:..I..-._........M...I..w... .u. .322... .......u..3_ : . . . .. . . . OI301INII1‘32HQ3!”033V33. .0o ‘(III3330 0 .3... 3.30.fl043. 30*: I... .. 33v.0 .4. . 3.3. 3 0 .3 . 09 3 I 3 ...... .....p. ......i. ...._..I._ .354... ...... 3.2312... 2 .. . . . ..LI.0.....0...I.0. .....33. .I... 03.... 0.3.003} .. .33I4103 ....33‘3303 I. . ... z. . .03 3 _I...I 3 I 033._3. 30 '0 303.. ...I *5” Va. 0 I 3.3.133 3.4: 0.0 . 3 3. J .A. Q. 3 .30.... ..0/ . I O3) VIIIP.3 3. I3. 433 ‘. I . 0 In. ..q cu ..... MIT?“ _ .. 2‘ I .. 3 .... 30.3.... 0.9.0.3.. . . . . . 03 . .. 0 e . . . .3 I c~3if C33. . 3. 0.0 ...l?.% ...:9333 mi 3.. .0... .3 .J 3403' 3 0 .3 3 A I I I . D .3. .3 . 3 .0 3 3 I . I I . )fi.......bo..l. 3L 3_ 5&3 “3.3 asunaw.) 3...3..... I3§334mkfi3wv .. I I I . I10 " t ...... . I I I k I .... . I 3’3 0.9%”...3 .‘03I3 33.0.0 ‘_0 .J‘3 .a 033.30.30c‘I \ 3 .u 3.3... . .I01.3J.‘_°v.v..h¢ . I. I3 - ......7 I IO. 03 0’ v 3 . 0 Q. 3 ’1 . 3 I O r¢.0CI'.III0.. 3 3.1 3.35.0 rrIoTs .3..3... 0.33.00.33.1I-I. .3 3 .0Ifi30 3.3033330... .wv ‘3IO33330VC_3 ... . 0003170.... .3 I 3 . 0 O I. _ ..I._ 1.0.. ... -.I.Ia...m ~300A"...I I..I.0.l3 357......m... .. . 631.19.“ 3315...... III......3I. ...0. ..III . .. . I I . . . .I .r3. . . 0 . .l . I . _ I I I I ..0...3...3I. .33 3 ...OIDI is... ..33. 330333.? .3. 3......033 .3433.E0...3..- W1.’J.~.....,T.'o.¢...o. Pym. .. II.I . 3 3 I . I I Z .3 I. . 0 I 0 0 I . I "3. (Jo 0'. 30 ‘0030.‘3.3 . II‘O‘ J 3 '3' I3 .0 O 3.90.30...‘ . .330 3.30 ... . I 3 .3 3 3! 3 ... . I I I 3‘. 3'.....QI0U13I..03M.030.........I 000. .03.. I. . 9 0 I 0‘3 ..0. $130300 I .kotwo’hc . _ _. 3I3 I . .»0..3 7...... .7 393...? ......I ,.. ' I‘ 000 IIIP..§0030n I' ‘. I.3I.IQ..330 . _ . I . I . . 0 ..3 o . .0 35.. .9...rIIo_... .. ,. ... . . O 0. I I r. I Q \II. .I 0. .II ,3- I 3 . 0A II M3 Isiaufl. W00 3 .0 00.30 3 ‘30-.- ...,0I_o.03r3I3I _ I... . III 3 . . . I 3 .9. 0 . 3. I 353 . I O 3 I _ ... O 3 I. _ I0. 4 .- 3 I. I 0. 0 I F I Q I... m ... . I . ' 3. 0.0 0.. . '.I.00 III . ..(I .. 0 I k “3 31.0 I, 0. . .....30 r: . . _ .. b. 3. 3. 30 ‘ I‘ll ' I'I. II . I ... I 0 w . 3 ... 3 0 _ .IV .v ‘3‘ I . I. P . . .. 0‘ . "O . .I I. II ‘I I I . . I Q I 0 II O O. I I 9. 0. : . 0. I .. "0.. I I Q.. I I 3 O Q 3 I I I I ..I3I3 3 II. 03.. & . I , I. - O. I 3... O Q I . O ' | a - . . . o I . 0 l I m - .. s m“ _ . . , .0 w . . 0 . . 3 I . I III . 3 I 0 I I u 0 09! Q W I I r I I . - . . -. - I . . . . O O. O ' _ I 0 . 00 II 0 I I 0 3. . I - _- O I . D. O . ‘I . I 3 O 0 . O ' Q 3 I . . . O. 3 . . 3. - - _ . w . . I V I. l I . 3 0 I . '.I . III..~”0IIII0m-~ow0~- u 3-3-33. .... Q i .D‘IIIIITIIII 0‘.. .. I 30. 'Y ‘ 3M.S.~' 7 L D eof V 0 I o 0 3 égre, Tatum ' I 1 .- 0 E. 9 . av. .t . I... .. I Om”...fill_ll. - e..-mw In... .M I 33333 .3 . .-. 33 ..333 ..I 3.— 3.3‘W33.I333.03I33. 3 . 3. . II0. 0 0.. I. C 330 I . 0 0 I .3 I. . . ........IrI...II_I33.I.3 w... I. . .0 3 I «3.33.3 .. V :4... .....I. :3 ... -I 2:25: 1.3.99.3 .....35..I0I.n,.=.r. .....f.....!.........£.... _ .IwfiIIanMnIIrJERIhrfif 535333174... l..:§u:......v..0..so~m§fih.3 ...... at . 31.0...{Kahiyqfl3g‘ I I '4! I’ . I I . I3 I. .U 3 I 39.! . ... .... t .P. _. 0.Io .Iy! .. I .. 3 . .3 3. 0. .I .0 I 3030” I .0 I00 I 33 3 r 1w .0. ,3 3.. 33.. H3 “A3,; I 3. . .3. . . _ 013’ 0. 03313133 .. .... ..I......;..I ... .3). II... I. I3 I; .‘03; 0'80“? . , I. _ .. .VII 0. .0 I ‘3I3II. . :3. 0. I. .0.3I_3. .3 33. O3 0'... 3h. I. J I. II ..I . III. 0 .30.\3 It.0 ‘10:”. I. I 0.... I. . I 1330-.00...II3IIr:. 0 I... on? ”wilwlli‘. .‘IItil II . . 3 .I II. . I .. . r . 3) 2.1034100: J0 “(VIII-0t I .0... . .I. 3I 3 I 0...]3 3~.I III IIolII3330 (-0.0. I. 0.- (IOII I I .. I! . I I. .. . ..9. 30- 7:433 IooJoI.Y3I.M 3.I§$\:0 {— , 4:1 .. .... I .r. I. :1 I ...?)I 01.50.... IFIHIIIIIP: 0t: . I _ I r I ....37 .3 31030 "xu30mb3.. ”fil‘ .‘OML‘ObONI” 3 I '10 .‘ 0 TI I . 0 0.0 I 0.. II 3.30 I . 34 I I . I. 3 .3 I I I .lp , 3 I I a 3. .. I .. 00". I I .II .Iohi‘. I 3'I..3. 0 3.33 . I I .I3 '13. 330l_0I.v 3 . C 03.. . I I. ’0 P. ' I .. II 7 I .l . I .I! I I 3. 0. 3.0.0 I 3. ' I. . 3 I I 0 . . I It' teI I30!" 'I‘ '30 3" I. 3. .I .IOI II 0.0I O r .. a I I 0 01;... 4 IHIYIIF WIMOI_ I‘.3!|3.‘-II(&..3'J33.I3 II. Q I I ...Is .3O3. DI . . .. .I. I I .I I .I 3.5!. 69.16.... . (IIIIJNI I I 3 P . oI I. I l .' a .I 0! .3...)-III33 I)! I» ’3. I II II I I I I 0 .30 III. I '.I 0.I.lf.003ll 3.: 0 I. I I I D I I I I .I 0. 33' .. . . .3! I 3 V I I . .II. . ..II III I «III.- ..M..... . .1. ((5.03 . A I .. II I .0 0A. ‘I053 .03th ..l0IiH3-‘ 3 3 3. I .. -- ....0 v.3 0.. I‘lI33I .‘4 .III 30 L I I ‘0 . I I .\ .3 II 15:0. .Il II.‘.0v’I(“ 0;, 33 I I . II ’ I 3 I..0 3.. . 0:0 "I0‘.. 0 I. I 3 I I0.000 0’ ‘0!” 3.. :0 I 3 I I ..I ..I0. .I . Aff. 0 .. I I If I . I I I I .III III..I|JI(.I O VIII}: I 3. . _ 0. 3 0 III) 0‘ II 0 r A I .. a ‘3IUPrI... I‘.‘3H._OII_‘ ‘0IO' (it... 3 . 3 I ..I . . 0 000.3".- III-II.) I I». I I. '0.I0It_t’I0.a. IOIOFAIIII ’006III _ o.IIII I .. I . 3 . 30 I OIIII I 0 I 0 I I I0. .I '33.. .Il'IhI'I .I'flo- II LO‘IIIo I 3 Or.’0..ll3.I. I I I I . .. I3 I III. '(OIIH' 0.0. II 0'3 .. I I II I I .-. OI I I II I .’ ' r I I II ,0 l3. 0. I0 I I I. III tr 14 I I I r . 0 I .. I I I . 3 I I 0 .I00 ‘I .3 ‘ICI I I I 3. .. .1 I‘ I000“; ’l}o00 1 | I I! I .A 0... I0‘0 I ‘0 ' I . I (I .3 DUI. I I 0.. clI' I 30. ‘ bI I 0 I 01’ I I 3 .0. I 0- I- ' 0species on the Atlantic Coastal Plain and one in montane areas of Mexico. The monotypic section Arseno- % includes only Lyonia ligustrina, a widely distributed and variable species of the Coastal Plain, Piedmont and mDuntains in the eastern United States. Section garia inCludes two species, both of the Coastal Plain, although one also occurs in Cuba. Numerical taxonomic and chemo- 8ystematic investigations support the recognition of the three sections. The phytogeography, ecology, floral andVegetative anatomy and pollination biology were inveStigated. The genus is distinguished from related genera by its axillary inflorescences from the wood fine previous seasc r32: buds, and cons, :5 Lie fruits. (5(2) Walter S. Judd Ch (9 (If the previous season, two large outer scales on the ‘winter buds, and conspicuously pale, thickened sutures of the fruits. THE SYSTEM n N.- ‘ 1“ Parnell? Depdrme'lt THE SYSTEMATICS OF LYONIA (ERICACEAE) IN NORTH AMERICA BY ;\ \‘I \ —.‘ -|. Walter Sikhudd A THESIS Submitted to Michigan State University in partial fulfillment of the requirements for the degree of MASTER OF SCIENCE Department of Botany and Plant Pathology 1974 Ivish to 8X} ““f “'5‘ ‘ I. V‘V I I .essor. Dr. zinc-suragement (3:13 :25 and Stephen " :n; u. :rzticism concert. :' 3-50 to Drs. Jareze ~L‘.J. Furlow for t1”. D r. Barbara I p. “'45-? gave the late .\a and relate: 532:1“ - - rm 0f chhlga: E5 1" 2 3'5: “{C is:- . y “a . 5 are listen fittfi" ACKNOWLEDGMENTS I wish to express my sincere appreciation to my xmajor professor, Dr. John H. Beaman, for his guidance and encouragement during this study. Drs. Aureal T. (Iross and Stephen N. Stephenson provided helpful comments and.criticism concerning the manuscript. Special thanks go also to Drs. James W. Hanover, Garrett E. Crow, and John.J. Furlow for their suggestions. Dr. Barbara Palser of Rutgers University gra- ciously gave the late Dr. Kornelius Lem's collection of Lyonia and related genera to the Beal-Darlington Herbarium of Michigan State University. These specimens were a valuable aid to this investigation. The support of a National Science Foundation pre- doctoral Fellowship greatly facilitated the progress of my research, and funds for fieldwork were provided by the Department of Botany and Plant Pathology, Michigan State University. I also wish to thank the curators of the herbaria from which specimens have been used for this study. These herbaria are listednear the beginning of the Taxonomic Treatment . ii A special the. :iersméing, patier. 11;;Izout the stuiy. A special thanks goes to my wife, Beverly. Her understanding, patience, and help have been essential throughout the study . iii 13? OF TABLES . O nu.- A n?!“- ..;. J? MULRES. . fic-a-‘QFVA - ~‘.:....,‘\,‘ng‘ . w . 37331 CAL ACCOUNT . u ‘- ..‘rw- -14.. REJIIORSHIPE EYES-SPA?“ AND E 1.322% FRUIT MOP 3:1,. «avrescence . . :pmgenetic 13338:. . n O .a;yx , Ctrslla . ‘ 5‘:~3ecllum . eg'neecium. . :riit . 'I . ~~ HON BIOLOGY 1:233. TABLE OF CONTENTS LIST OF TABLES . . . . . . . . . . . . LIST OF FIGURES. O O O O O O O O O O 0 INTRODUCTION . . . . . . . . . . . . HISTORICAL ACCOUNT. . . . . . . . . . . GENERIC RELATIONSHIPS. . . . . . . . . . PHYTOGEOGRAPHY AND EVOLUTION . . . . . . . FLOWER AND FRUIT MORPHOLOGY. . . . . . . . Inflorescence . ‘Morphogenetic Cycle Flower. . . . Calyx . . Corolla . ‘Androeciwm Gynoecium. Fruit . . POLL INAT I 0N B I OLOGY O O O C O O O O O O EPIDERMAL APPENDAGES . . . . . . . . . . Key to Sections and Species of Lyonia Based on Trichome Type. . . . . . . . . . . ECONOMIC IMPORTANCE . . . . . . . . . . NUMERICAL TAXONOMIC STUDIES. . . . . . . . Phenetic Studies . . . . . . . . . . Phyletic Studies . . . . . . Comparison of the Phyletic and Phenetic Methods iv Page vi viii 17 23 40 40 40 42 43 43 S4 55 55 59 62 64 75 77 78 130 141 '32"; ATIC STUD lathe-is and Mater; Results and Discus :. ‘Sl’flC TRE‘ATYEXT "axis Rattan. :2; tcSections, E ":i’ v lial O - . :geczzens Examine lien; m. a hut. sect l. LVCnia Sr 2. 1.;"05‘ a T“ x... YRQLLTSOI 3. L‘ieria f: \ Mitt. Sec Ssdt. “CV. e ' LVQni . ‘ a ] 9.. a r “ERRED: Page CHEMOSYSTEMATIC STUDIES . . . . . . . . . . 143 Methods and Materials. . . . . . . . . . 144 Results and Discussion . . . .. . . . . . 147 TAXONOMIC TREATMENT . . . . . . . . . . . l7 8 Lyonia Nuttall o o o o o 6 o o o o o o 1.78 IKey to Sections, Species and Varieties of L onia C O O C O O O O C O O O O O 180 Spec1mens Examined. . . . . . . .. . . . 182 Lyonia Nutt. sect. Lyonia . . . . . . . . 183 l. L onia squamulosa Mart. & Gal. . . . . 183 2. L on1a fruficosa (Michx.) G.S. Torrey in RoBinson . . . . . . . . . T- . 195 3. Lyonia ferruginea (Walt.) Nutt. . . . . 215 Lyonia Nutt. sect. Arsenococca (Small) Judd, Stat. nOVO O O O O O I O O O O O O 228 4. Lyonia ligustrina (L.) DC . . . . . . 228 4a. L onia ligustrina (L.) DC. var. Iigustrina. . . . . . . . . 231 4b. L onia’li ustrina (L.) DC. var. o 1081 ora Michx.) Fern. . . . 258 ‘Lyonia Nutt. sect. Maria (DC.) C. E. Wood J. Arnold Arb. 42:48. 1961. . . . . . . . 281 5. Lyonia lucida (Lam.) K. Koch . . . . . 281 6. Lyon1a mariana (L.) D. Don. . . . . . 301 LITEMTURE CITED 0 O O O O O O O O O O O 321 CO‘Z-Tvarauve trq authors , CCZParison of LiSt of 01:ch 5 each , Characters Us (Phenetic HG J5? anal‘i: loa‘chdr al ESRGUP anal}. lZB-Chara 3390;? anal! “Titan 38mg? anal flower-f HERO?” aha? 10. 11. 12. 13. 14. LIST OF TABLES Comparative treatment of Lyonia by various authors . . . . . . . . . . . . Comparison of the sections of Lyonia . . . List of OTU's of Lyonia, giving the code name eaCh O O O O O O O O O O O O 0 Characters used in numerical taxonomic study (phenetic method). . . . . . . . . Matrix of error potentials, set. 0 O O O O O O O HGROUP analysis: 108-character HGROUP analysis: 128-character Matrix of error potentials, set. 0 O O O O O O O HGROUP analysis: Matrix of error potentials, vegetative character set . . . . . . HGROUP analysis: Matrix of error potentials, flower-fruit characters. . . . . . . HGROUP analysis: Matrix of error potentials, floral characters. . . . . . . . . HGROUP clustering of characters of Lyonia. . Factor analysis of loo-character set . . . Characters used in phyletic study (Wagner me thOd) o o o o o o o o o o o 0 Character divergence values for each taxon, used to construct Wagner tree. . . . . Matrix of character divergence used in Whiffin- Bierner method. . . . . . . . . . vi of Page 4 13 80 81 104 105 106 107 108 119 128 131 133 138 I‘i‘ . . (D o r o . - o . . . '-.‘ u I." o 3‘ 6“ c graphic 10C? foliage 537% com. ands V“ Rf values and i I v ‘ m L 00.13 Total listing C in each POISJ studied Characteristic species of later of cor: of Lvon' . . 1a 9 Chemical simii in North A Compounds se; lioustrira .oiiosn‘w ‘. “'1 - u agnostic c< Comparison 0 erruc‘ .ne comparison c llQ“S '. 1] Table Page 15. Geographic location of specimens from which foliage samples for analysis of phenolic compounds were obtained . . . . . . . 145 16. Rf values and color reactions of the phenolics in Llonia O O O O O O O O O O 0 148 17. Total listing of phenolic compounds present in each population of each species studied . . . . . . . . . . . . 153 18. Characteristic phenolic compounds for each species of Lyonia . . . . . . . . . 154 .19. Number of compounds in common among the taxa of Lyonia studied . . . . . . . . . 167 .20. Chemical similarity between species of Lyonia in North America . . . . . . . . . 168 21. Compounds separating L. ligustrina var. li ustrina from L. ligustrina var. foIiosirora. . . . . . . . . . . 170 22. Diagnostic compounds for section Lyonia. . . 171 23. Comparison of Lyonia fruticosa and L. ferruginea . . . . . . . . . . . 210 24. Comparison of L onia li ustrina vars. ligustrina an oliosirora. . . . . . 256 vii "Grid distribu Possible disu lic'JStrlna \ Drawing of flo Drawing of flc Drawing of fl: Drawing of £11 Drawing of fl Variatioh in cease in E VariatiOn 1 COIQr. . variation in LIST OF FIGURES Figure l. WOrld distribution of Lyonia . . . . . 2. Possible distribution patterns in Lyonia ligustrina . . . . . . . . . . 3. Drawing of flower of Lyonia ligustrina. . 4. Drawing of flower of Lyonia ferruginea. . 5. Drawing of flower of Lyonia fruticosa . . 6. Drawing of flower of Lyonia mariana. . . 7. Drawing of flower of Lyonia lucida . . . 8. Variation in capsule morphology and dehis- cence in Lyonia . . . . . . . . 9. Variation in glandular hairs in L onia; darker shading indicates a reddish color. . . . . . . . . . . . 10. Variation in size and shape of ferrugineous scales of Lyonia ferruginea . . . . 11. Longitudinal views of hair types in Section Lyonia. . . . . . . . . 12. Hair types in Sections Maria and Arseno- cocca. . . . . . . . . . . . .13. HGROUP phenogram of Lyonia based on 128 character set . . . . . . . . . 14.. HGROUP phenogram of Lyonia based on 108 character set . . . . . . . . . lS. HGROUP phenogram based on vegetative char- acters of Lyonia . . . . . . . . viii Page 26 34 45 47 49 51 53 57 68 70 72 74 95 97 99 ‘. :.;:e . a c». u p - (‘I A. - . L- - HGROL'P pheno;ra acters of L? 3290:? phenogra 0f Lyonia based on 10 . N 35932"? relatio: °n Vegetati‘ - HGRC'JP relatio flower-frui ThreeLd i me n 5 1C snips anon: data from c F 4 : .actor analys based On v based On f LVOnia Figure Page 16. .HGROUP phenogram based on flower-fruit char- acters of Lyonia . . . . . . . . . 101 1]. HGROUP phenogram based on floral characters of Lyonia . . . . . . . . . . . 103 18. HGROUP relationships between OTU's of Lyonia based on 108 character set. . . . . . 110 13. HGROUP relationships between OTU's based on vegetative characters of Lyonia . . . 112 ML. HGROUP relationships between OTU's based on flower-fruit characters of Lyonia . . . 114 21. Three-dimensional model showing relation- ships among OTU's of Lyonia, based on data from cluster ana y81s. . . . . . 116 22. Factor analysis: Principal Factors I and II based on vegetative characters of Lyonia . 122 23. IEactor analysis: Principal Factors I and III based on vegetative characters of Lyonia . 124 24. Factor analysis: Principal Factors I and II based on flower-fruit characters of Lyonia . . . . . . . . . . . . 126 25. Index of divergence of L onia with code letters of characters 1n specialized State. 0 O O O O O O O O O O O 135 26. Phylogenetic trees of Lyonia produced by Whiffin-Bierner phylet c method . . . . 140 27. :Representative chromatogram.of Lyonia fruticosa . . . . . . . . . . . 156 23. Representative chromatogram of Lyonia ferruginea . . . . . . . . . . . 158 29- Representative chromatogram of Lyonia ligustrina var. ligustrina. . . . . . 160 30- Representative chromatogram of Lyonia ligustrina var. foliosiflora . . . . . 162 ix . . :.;re 31. Representative 1:; . Representative mariana . a Phylogenetic t: pounds dist: :2. Representative “ “‘- a u 1.. Listribution o 1“ Rexesentative 15. Dis tribution c 8.‘ c “it Curvatur. co 8a and E C. . A Cofiearison LVon' Q E ,r, L Figure Page IL Representative chromatogram of onnia lucida . 164 32. Representative chromatogram of onnia mariana . . . . . . . . . . . . 166 33. Phylogenetic tree showing the phenolic com- pounds distinctive for each taxon of Lionia O O O O O O O O O O O O 174 34. Phylogenetic trees based on chemical data, using Whiffin-Bierner method . . . . . 176 RL Representative specimen of onnia squamulosa . 187 ML Distribution of onnia squamulosa. . . . . 189 EL. Representative specimen of onnia fruticosa . 199 38. IDistribution.of onnia fruticosa . . . . . 201 39. Leaf curvature and position in onnia 35213- 2222 and E. ferruginea . . . . . . . 209 40. .A comparison of the flowering periods of Lzonia fruticosa and 2° ferruginea . . . 213 41. Representative specimen of onnia ferruginea . 219 42. Distribution of onnia ferruginea. . . . . 221 43. .Representative specimen of onnia ligustrina var. ligustrina . . . . . . . . . 234 44. 1Distribution of onnia ligustrina var. ligustrina . . . . . O C C O C O 236 45. .Representative specimen of onnia ligustrina var. foliosiflora . . . . . . . . . 263 46. Distribution of onnia ligustrina var. foliosiflora. . . . . . . . . . . 265 47. lRegions of overlap in distributions of L onia ligustrina vars. ligustrina and foIiosiflora. . . . . . . . . . . 279 48. Representative specimen of onnia lucida . . 286 F;;;:e 4%. Distribution 0: it. Regresentative Distribution 0: _._.L -‘ ‘ . Figure Page 49. Distribution of onnia lucida . . . . . . 288 50. Representative specimen of onnia mariana . . 305 51. Distribution of onnia mariana. . . . . . 307 xi .__.____ am -~ , The Ericacea' fillies, but infraf iffzcult to detemi. 255 in point. Andrl \ I:::.‘ie diverse arra :s:e is still much "4: should be recor iiiinctive gene r a «s see ll CORtI‘OVEI 5 fine Coastal Plai '- in some locali' ‘tI-r 75:321. 1011, The e .. C19 species 0f 1' IQSea sewn to the F1!” :3 ‘i‘svxn . 3.31:4 ~a in COn'JEn‘ we INTRODUCTION The Ericaceae are one of the more natural plant families, but infrafamilial limits of its taxa are often difficult to determine. The tribe Andromedeae are a case in point. Andromeda was long used as a catch-all for the diverse array of species in this tribe, and there is still much confusion as to the number of genera that should be recognized. Lyonia is one of the more distinctive genera in the Andromedeae, but its definition has been controversial. It is abundant in many places on the Coastal Plain of the southeastern United States and in some localities is a dominant component of the VGQEtation. The extreme variability shown by several 0f the species of Lyonia has long been a source of con- fusion and has been a stimulus for my interest in the genus. My research on Lyonia was initiated as a contri- bUtion to the Flora North America data bank. Because of the suspension of that Program, this treatment is pre- 8eI'lted in conventional form; but hopefully knowledge °f the North American species is now more readily ac(lessible for a computerized taxonomic information 72's. of the future. use 2min Wed and e . . . - I anus-c in the field. nae: ision to re: 2 1.5.. ml SPECIES. 1:21:21”: studies we: :52: heavily on sta: $223.“. only the sp- ‘ El" ., it is hope ’55 can eventually “‘3 PIO"ide a has ~::‘.C‘.‘..t West India system of the future. Over 4,000 herbarium specimens were examined, and each taxon except 1;. squamulosa was studied in the field. Fieldwork was a fundamental factor in my decision to recognize L. fruticosa and L. ferruginea as distinct species. Both the chemical and numerical taxonomic studies were helpful in reinforcing conclusions based heavily on standard morphological comparisons. Although only the species native to North America are included, it is hoped that an investigation of the entire genus can eventually be completed. The present study should provide a basis for interpretation of the more difficult West Indian and Asian taxa. ‘ u - Both Lorna 1. Embed by Linnae‘u‘ ';'V \ "qo ;..a was out m l as glazed in Vaccii \ :Lz'al resemblances ‘a of nisunders ta at: :4 ' *5 «meat ity of Lil 2:1;1'2t it was a l: 333: Pu: sh. lBlS; lite: s correctly 2 LN a. the err x-omed l ‘ q «ray, lBBB; Sr: ‘35! botanists 2;: - SPECIES , th ':: ‘ “‘lptiOn Of HISTORICAL ACCOUNT Both Lyonia mariana and -I_.. ligustrina were described by Linnaeus in his Species Plantarum. Lyonia mariana was put in the genus Andromeda but _I_._. ligustrina was placed in Vaccinium probably because of its super- ficial resemblances to that genus. This led to a great deal of misunderstanding by later botanists concerning the identity of Linnaeus' Vaccinium ligustrinum. Some thought it was a blueberry (Marshall, 1785; Willdenow, 1799; Pursh, 1816; Nuttall, 1818; K. Koch, 1872), while others correctly concluded that its true affinities were with the Andromeda group (Elliott, 1817; DeCandolle, 1839; A. Gray, 1886; Small, 1914; Wood, 1961), Table 1. Most early botanists used the name Andromeda paniculata for this species, thinking that the A. paniculata of Lin- naeus (1753) referred to this plant. But Linnaeus' description of A. paniculata, which is quite ambiguous, probably refers to Leucothoé’. racemosa. Another nomenclatural mixup concerns the identity of Walter's (1788) Andromeda ferruginea. Pursh (1814), Elliott (1817), and Nuttall (1818) incorrectly took his ltl‘l‘l ILIIII'IIIIIII “A n ‘1‘ "C ~~ L» Illlllqlull' l‘ll‘lllvlllll AQ—zns 5. ~35 ..IIII .Illllllllllll *— C: a» ‘3IZ...—I 2.3K : )..:'_.n a «I :eauwv «Arnw. .... . ~23 -E£ILEZ I11. .IHUIoSI I3:u.~0> NJ Q‘Zufflnu. hO «a n\ a» .30.).th «COEUIIHU I)welhl§EOb CSXIL. ll. - H E4‘<.h II II II II II II II mmmmmmmmmm .q Asmooawsum I macaw .um> dosh Ammcwmsuuom .<~ I snuom .<~ INSHHOM .4~ I someoupcé uposouvcd mpoeonond II II II II snouaosuu .A «acoomwummun l I .um> aocH I. MOfiH Am amen .4 Am amau .4 Imsuuom .<~ Admswmauumm .4~ Imsuuom .K~ mumEouccd moofiouvad mvwsouvcd moofioupqd someouqu II II mondmsuuou .A I Asmocaoum .m as Ammo sou . 0cm can muaflm madam dam an chow 0H” escafiw scans»: 4H .¢~ Iomcom . ~ loanum .<~ Ioacow . ~ Icons .K~ Ioacdm .4~ In HH .>~ I upoaouucd svoeoupsd noofioupnd avoaounqd upoEouvcd upofiouvc¢ aswswooa> dddh¢¢flflda .A Aumduac .m~ Ramayac afly “ammum: .wu .uuouduou awy Amman“: .my Aommwm: .N~ I upmfioupqd upoaouvcd upmeoupcd moosoupcd someouvsfl mvoaounnfl II duaoad .q Amcmaume .m~ “manauua .fl~ Annuaume .fl~ Aucuauue .fly Amcuauua .m~ Annuauda aw Accoanda xfl~ I upmaonpcd mumEoupc¢ apoeoupc< momfioupcd mooEOHUQ¢ opoeouvnd noosoupad acuauufi .q reams. .eamac .momac Assad. .mohas .mmpav .nmpas nexus unawaam cause nuance: sosocnawz hands: Haucuunz uaosaswq .uuonusn unowun> in «HCONA uo vacauaouu o>wuoudmeoo .H man‘s .4 I . P . ,r ..r..— u . ...m.-—~_ . a... “P... . .. .. .r. . , . r. .... . . . _ Ink- LI . . I v I IJ .I .c...... . .r : "Lu . CALCN .22: Tris...“ Q2.C;W~ p.._lE..~TC< p.72..— up}: as .... . .. 2. 23...: Six: .26.: 21:: 1:3: .....K_u~_.:..z: :22... .x ..L T...:p...:.: :2... ... ~25 2.2 cones “4-1—. H U carJ o N iw~au3; Ore‘ _ . sided I I "file . S With 1&5 ‘ .t: CS ar . “tir re ‘ or less 11 the absence of ferruginous scales, the presence of long club-shaped multicellular hairs, and by the depressed globose capsule, not prominently angled, with less thickened sutures. The leaves are nearly always decid- uous. This variable species is distributed from Maine to Florida and eastern Texas. Section Maria (Small's genera Neopieris and Des- mothamnus) includes both A. mariana (the type species of section A3333) and A. lucida. This section is character- ized by the flowers borne in axillary clusters on the branches of the previous season, by the noticably appen- daged filaments, the ovoid or globose-ovoid capsule with moderately thickened sutures, and the elongate flowers with lanceolate calyx lobes. The leaves are entire, evergreen or deciduous, and glabrous to quite pubescent (usually with small, stalked glandular hairs). The two species are native to eastern North America, primarily on the Coastal Plain, although A. lucida is also found in Cuba. The fourth section, Pieridopsis, includes only Asian species. Those species have the flowers in axil- lary, one-sided, usually elongated racemes and have capsules with less strongly thickened sutures. The leaves are entire, persistent or deciduous, and glabrous or more or less pubescent. Species in this section can :2 ions”. from Kashm; :i central China tr U H“: Wood (1961) :‘A 5~ .Tari g ‘ , *aalns ' CiaCEd SPECIEC :esrs o. the ‘1 32:51qu L. 113”: . \¥ ..a: “r. ' 1‘3 Cf I' “*6 r. ‘U‘. “323: a w to a ‘5‘ t3 12 be found from Kashmir through the Himalayas to western and central China to central and southern Japan and Formosa. Wood (1961) included A. ligustrina with A. lucida and A. mariana in section Maria. He states: Section MARIA (DC.) C.E. WOOD (Xolisma sec. Maria (DC.) Rehd.), as defined here, includes only three isolated species which are placed together on the basis of the filaments appendaged near the summit, the non-lepidote pubescence, and the axillary corym- bose (fasciculate) inflorescences. I consider A. ligustrina to be distinct enough to warrant a separate section. It differs from both A. mariana and A. lucida in the following points: (1) its long club- shaped glandular hairs, (2) flowers in pseudoterminal panicles, (3) long pointed buds, (4) small urceolate flowers, (5) small subglobose pubescent capsules, (6) serrulate to nearly entire leaves, and (7) only small inconspicuous spurs near the summit of the filament. The results of numerical and chemosystematic studies of the genus in North America also indicate the presence of three distinct groups, corresponding to the three sections described above (Table 2). Rehder (1924) used the name Xolisma instead of Lyonia. Rafinesque (1819) proposed the name Xolisma for the Lyonia of Nuttall because Rafinesque had given the name Lyonia to another genus, but that was only a new name to replace Polygonella Michaux, which Rafinesque ll I . Inez—1:. ~.§.A¢..II.._.AU :1...»— 2.«) I...¢A._ ._ :..1£ .1.-.111. .. I I I‘ll-'1!lll‘lllill IIIIIIII‘I Ill“ I I I'......I:...u.....:owll.~l¢ a #3.: I...m..—.r.,v&. . .190: III.-. .I Fu‘—~.AsfilNl lU-J‘a .I.A...Qx.~ LO ICOAu ...l I ., Cu h... C3I-l350 . L .N ...AAESA 213 msosvomo mo>owA .m moosvooo no coououo>o mo>mmA .m cooumuo>o um>qu .m was» muMAooous «AAouou .5 vo>0 so no no oAuocAAmo oAAouou .5 oquooouo «AAouou .h wmoAAm .00A23 mAAouoo .m msoundAm .ouAnz uo chm mAAouou .w ouooAQOA .ouAss oAAouou .w AmvoAomwm omonfihuoo mo moAOAommM noAOAounm mofioomuv mvoAcmm GA ocuon mno30Am .m muonfihuou cA canon muozoAm .m omonfihuoo GA ocuon uuozoAm .m no>Am> oAaomoo mo no>A¢> Scum mo>Am> Eoum uAcs a mu chusuumoo Spun uAcs n no ocAumHumon AAA: 0 no chvmuumon no: mAAusms .oocoonnu moususm .v nos mAAwsms .oocoonsu moususm .v .UocoxOAnu .nouuusm .v 133.? AAMHm ousmm .m ucocAEoum musmm .m muo> HOV ucomnd «Adam .m cornea» sAaoAAunu OAMAduumo cAmMME wooA .N muAucw cAdeE umoA .N no ouAuco :Amuda mooA .a mqun AMA rachAm commemlnsAo .mmumA AAA: nucoAm .A mqun udAsocmAm AAmEn AAA: mucuAm .A qumou ouovAom AAA! nunsAm .A oooooocomud .uoom sAusz .uoom oAGONA .uoom .quONA mo uaoAuoou and no sonAummeoo .~ MAAAH :siiered incl???”I :ralié nape and a s 327.13 was prOPOSEd I‘ -.~..:pi.aceae (CW-8' Etna of Nuttall: E.L;c:t is the olde- snepiadaceae, whi Lueip the next 01 l2ztall‘s Lvo ' nia. nrw ' - - onia Elliott Lvon' 1a was ‘: Verb I «- Jy. .h Alfierlca ' .«Owers. Barty- till. l'sr‘; ~ “\rda Sevev "3°31 shrubs w 4“" “meta GeOrgl ti 8‘“ c r»Cla11 Y at W". ‘1 k ‘“ Ca iec “Slanabyetrad V“ It’s-rent. ‘1.th " ab C300 egetablée mtks dwflhe ‘ C‘Jti‘ rich 5:.ILQC of 14 considered inappropriate. Therefore Lyonia Raf. is an .invalid name and a synonym of Polygpnella. The name Lyonia was proposed by Elliott (1817) for a genus in the JAsclepidaceae (Cynanchum). This name was published before Lyonia of Nuttall; thus Rehder concluded that Lyonia IElliott.is the oldest and valid name for the genus of .Asclepiadaceae, which made it necessary for him to ‘take up the next oldest name, which is Xolisma, for 'Nuttall's Lyonia. Lyonia Nuttall has been conserved (yver Lyonia Elliott (Kew Bull. Misc. Info. No. 3, 1940). Lyonia was noted very early in the exploration <3f North America, often being taken back to Europe where .it was grown in gardens because of its beautiful foliage Iand flowers. Bartram (1791) refers to both A. ferruginea and A. lucida several times in his Travels, indicating that both shrubs were common elements of the vegetation in southern Georgia and Florida even as they are today. His description of the“ plains near the Suwannee River .are especially interesting: The chief trader, intending to show me some remarkable barren plains, on our return to our encampment; about noon we set off, when we came ‘within sight of them, I was struck with astonishment at their dreary appearance; the view Southerly seemed endless wastes, presenting rocky, gravelly and sandy pine barren plains, producing scarcely any vegetable substances, except a few shrubby, crooked Pine trees, growing out of heaps of white rocks, which serve only to perpetuate the perse- cuting power and rage of fire, and to testify the aridity of the soil; the shrubs I observed were chiefly the following, Myrica cerifera, two or three varieties . ferruginae, And: irangula, Siderc Ilex nyrtifoliu: Cassine, and a . green and decrd heartiful. . . William Bal rilr. Darlington‘ 12 or another, t. : 333:1“.- Anerica e .. a ‘ ' ..a ‘F ‘ Animator: , lE . ‘1».- r ...;a includes 2 ...eot Eire in By ‘ar th .. ~ ‘ e no : gr H L) r1 (0 (D (+01 01 f 1::O"per . 1a" Chiefi lS-he‘ it ' VI th‘ n'. . x: ‘u ' :s' ‘ A "VP are . CR“; :1 -““l 15 three varieties. . . . Prinos varieties, Andromeda ferruginae, Andr. nitida, varieties, Rhamnus frangula, Sideroxilon sericium, Ilex aquifolium, Ilex myrtifolium, Empetrum, Kalmia ciliata, Cassine, and a great variety of shrub Oaks, ever- green and deciduous, some of them singularly beautiful. . . . William Baldwin in his letters to Dr. Muhlenberg and Dr. Darlington, from 1813 to 1817, refers, at one time or another, to every species of the genus native 'to NOrth America except A. squamulosal His description (12 Darlington, 1843) of the Pine Barrens of southeastern Georgia includes some interesting comments on the impor- 'tance of fire in such habitats: By far the greatest quantity of land along the sea board is low Pine barrens, covered with Pinus palustris (long-leaved Pine), principally . . . Undergrowth, Chamaerops serrulata or Saw Palmetto, --with some fine shrubs,--as Andromedas, Bejaria, etc. This is the most sterile offallidescfiptions of land. . . . There is as yet, but little naked sandy desert; but should the weather continue, a ew years longer, as dry as it has been . . . and fires should rage as extensively, destroying the vegetation,--a large portion of the maritime part of Georgia would be rendered like the deserts of Arabia! Were I a member of the Georgia Legislature, my most strenuous exertions would be made to pre- vent, by law, the burning of the forests,--which impoverishes the land, and does incalculable mis- chief, without a single advantage resulting from it. Yet many of the stupid people do it, to destroy the rattlesnakes--make the grass grow--and I believe for the fun of looking at it. this judgments concerning the effect of fire on the vege- tation are slightly exaggerated! Actually periodic tnxrning promotes such a pine-scrub vegetation. Lyonia :mariana and A. fruticosa both are probably fire induced e‘iprosper in so Exes (Iltis, per userxation) . 16 and prosper in such dry sandy areas swept by periodic fires (Iltis, personal communication, and personal observation). n' . ine Fur-:1: 4.:-.-_ .i- ‘ "-5.3‘ut group. “P5353 35 one I 15 early decal Pun. ' " :9 gene-r ii‘th‘b'la '7 \I he. \ it" A . {I an“ 501 3 \ Rafine .... actual ( .""‘n .‘ “I; 8:101:30 “is (that GENERIC RELATIONSHIPS The Andromedeae have long been a taxonomically difficult group. Most of the species were originally treated as one genus by Linnaeus and other authors until the early decades of the nineteenth century when many segregate genera were recognized. Today the tribe includes the following genera in the eastern United States: Andromeda, Leucothoe (including Eubotrys) , Gaultheria, Zenobia, Lyonia, Pieris, Cassandra, Oxyden- c_l_r_‘_1_1_m_, and Epigaea (Wood, 1961). Rafinesque (1836) recognized the artificiality of Andromeda as circumscribed by the botanists of his day. He asserted that: ”GENERA ARE NATURAL, and that every actual Genus that is unnatural, arbitrary, or polymorphous, IS NOT A GENUS, but an incorrect and artificial aggregation of aliens!" McVaugh's (1945) first recommendation in the recognition of segregate genera is that primary consid- eration should be given to strong morphological char- acters (that is, ordinarily, qualitative characters, or those involving changes in? plant parts, or the 17 sauce or a‘oser rarer than to t] :‘tlving change 2:: of parts. ‘1'?! there a: ifferentiating 3:5: recogniz. :e ‘S W“ A t—‘yh a nil -t‘..a1 qroa .JCSQA J fifth-,3 ”II" V' m, 28.. 3‘ :i‘fiu “‘31 aha I‘.‘ ‘ss *- ‘ . \~Jls‘ 10 Ana ... 5‘ ' ‘tate .‘ lZed (.“ ...s and 1 “3:5 . 8017‘s; ‘2 3“. I N‘s: 18 presence or absence of some distinctive attribute), rather than to the weaker characters such as those involving changes in number, shape, position or attach- ment of parts. One of the problems in the Andromedeae is that there are few strong morphological characters differentiating the major groups. Thus the number of genera recognized has varied widely and often seems to be as much a matter of personal opinion as the result of natural groupings of taxa. David Don who in 1834 proposed one of the first attempts at a natural arrange- ment of the genera in the Ericaceae stated: .As happens in other very natural families, the Icharacters of the generic groups in the Ericaceae are not so strongly marked as in those that are less so; but we are not on that account to give ‘up the idea of dividing them. . . . His treatment (1834) is quite modern, including the following genera in the Andromedeae: Andromeda, Cassiopg, Cassandra, Zenobia, Lyonia, Leucothoe, Pieris, Phyllodoce, Bryanthum.and Daboecia. It is not within the scope of this revision to consider the validity of the various segregate genera of Andromeda. Lyonia was one of the first segregate genera recognized (Nuttall, 1818) . The genus is quite homo- genous and is separated from all other genera in the Andromedeae by a definite combination of characters, although some of these characters appear individually in other genera. Thus Lyonia fits McVaugh's third :ezzzerdation 't :3 supposed 9933 3.2.5 another. b 13:7."2 seems to 0 12:3 of the gene: :arest relative :32 can also be \3 (Hood, .‘Iuch conf sateen ironic: an ‘1. ...:‘e grOlJp i it: A.‘¥ " N ““buOdy (192? o .l‘ n «.225 between at. ':«S) and those “=5 Spurs). 'r z 55" v- ' -.-3.ating pl 0 \' fires ‘ 0f fllar-e "i:: I 19 recommendation "that the most important criterion of any supposed genus is not the width of the gap between it and another, but its own biological unity.” In fact Lyonia seems to have more distinctive characters than many of the genera recognized in the Andromedeae. Its nearest relative is probably Pieris although relation- ships can also be seen with Zenobia, Gaultheria and Cassandra (Wood, 1961). Much confusion has existed as to the differences between Lyonia and Pieris with some authorities putting the whole group into one genus. Matthews and Knox (1926) and Anthony (1927) have doubted that any real distinction exists between appendages on the back of the anthers (spurs) and those at the summit of the filament (also called spurs). This difference was used by Rehder (1924) in separating Pieris from Lyonia. These spurs are out- growths of filament or connective tissue and vary con- siderably in size. In position they range from insertion on the free portion of the filament to insertion beyond the point of union of anther and filament, thus appearing to be attached to the anther lobes (Matthews and Knox, 1926) . Therefore they concluded that these structures are simply emergences, and cannot be employed as a char— acter of generic importance. Their contention as to the lack of a real difference in the position of the appen- dages in these genera is only partly true. There does see: to be a wide as :cnfused beca; :zegration tiss' Ergir. of the as -. ¥ ‘0 U A“ P e “31159.54 ”.5? .35 34" k e hue tr‘i . r :‘F. 1 1.! I‘ t. J. 20 seem to be a wide variability in position, but in general the spurs of Lyonia are at a lower level than those of Pieris. The appendages of these two genera can never be confused because those of Pieris lack the white dis- integration tissue which is always present on the upper margin of the appendages of Lyonia. Even when the fila- ments of Lyonia lack the spur-like appendages near the summit, a line of white disintegration tissue is present on each anther-half, forming an inverted "V" at the junction of filament and anther (Wood, 1961; Palser, 1951). The appendages of both Pieris and Lyonia should not be confused with the awns present on the anthers of genera such as Zenobia, Gaultheria or Leucotho'é. Anthony (1927) also disagrees with Rehder's (1924) use of the thickened sutures of the capsule as a unique characteristic of the genus Lyonia. He stated that in the description of the genus Pieris in the Genera £132.- tarum (Bentham and Hooker, 1876) the capsule valves are described as having margins often thickened, and there- fore concluded that thickened capsule valves are present in both groups. Actually nothing is farther from the truth! Bentham and Hooker's treatment of Pieris includes Lyonia mariana, 2° lucida (in sec. M) and I": o_vg_1_._i_- folia (in sec. Eupieris) . All these species actually belong in Lyonia, and when the transfer is made the homogeneity of each group is greatly increased and the :stinction be1 aims is valic :9 also diffe: ..f.l:'.llty be ‘ I cons inc-....bt 5:031 l o: ‘ k . ‘ ‘An.“ ‘ v- J:::Ju' the t“. 1‘: .3. h.‘ bcne We 1a SOme b zaler 99%” .- I we :istributi Cite»: sh OUt e‘ 21 (listinction between thickened vs. nonthickened capsule *malves is valid. The inflorescence and leaves of Pieris ‘are also different in general aspect and can only with difficulty be confused with those of Lyonia. I consider Lyonia to be a natural group of species distinct from Pieris and set off by the conspicuously ‘paler, thickened sutures of the fruit, the exclusively maxillary inflorescences from the wood of the preceding season, the two bracteoles at the base of the pedicels, ‘and the two large outer scales of the winter buds. Some botanists have divided Lyonia into several smaller genera (Britton and Brown, 1913; Small, 1914). 'The distribution of characters used in such cases is so reticulate, when all the species are considered, that the segregates are essentially based on single character (tifferences. The temptation to do this is sometimes great as the group is an ancient one and many of the intermedate species are evidently extinct. Yet as pointed out earlier by K. Koch (1872) , the result of such a practice in the Andromedeae would probably be the recognition of almost as many genera as species! In summary, the group seems to be most natural and clearly distinctive when separated from E23135, and not divided into several small genera along sectional lines. Palser (l9 oray‘n on: V hr ....1. nanes .‘ue. :‘ze: stud', reco :fizese taxa by q szarizing relatl 22 Palser (1951), although she used the segregate generic names NeOpieris and Xolisma as well as Lyonia in her study, recognized the similarity in floral anatomy of these taxa by grouping them together in her table .summarizing relationships within the Andromedeae. IE2 As :- :--:;i- '5' D. U n 4 .e s s m a a w. .1 u. e s e s .m .... d n e x s a e b n l l. r A. O .1 LL .3 ¢ 5 .t k1 C U. ml 3 m .. a S S +u n . S a . o .4. .. . a. 3.. 3 A. .3. E . Aw ...—...; R ....u \nu u:- ‘s: ‘9.» ... u .C. s » a: a. s s . C \ ... L. p: as ... ... a . . a: .. .. ...: .. . .... ..r. 2:. .... .. t. t... a .. 3. .... .rr: 3.- PHYTOGEOGRAPHY AND EVOLUTION As treated in this study the 30 to 35 species of Lyonia are classified in four sections. Section Pieridopsis, comprising three or four species, occurs in Asia from Kashmir to Japan, and south to Malaya. The greatest diversity of the genus, although not the greatest number of species, is in eastern North America where three sections are represented. The monotypic section Arsenococca is found only in the eastern United States. Section _M_a_r_:i£ includes two species occurring on the Coastal Plain of eastern North Merica. One of these species, Lyonia lucida, is also native to Cuba. Section Lyonia is the largest subgroup of the genus, containing perhaps as many as 25 or 30 species. These are mostly natives of the Greater Antilles, although one is found in eastern and southern Mexico and two occur in the extreme southeastern United States. Most of the species in the southeastern United States are limited to the Coastal Plain; only _I_._. £19.29... trina is also common in the Piedmont and mountains. In more tropical areas such as Mexico and the Greater 23 :zilles the genus insulting isola was areas may h 13’." 1‘ .,....-ion in secti z.“ua 1S 5 i:::_:cr.ent of the nu“ ‘f‘ ...: ..c .Vr «.est whiz! Iriierr. regions .1 I I :ZILL'Jtl’Qn pane 5:55 a Tertiary r ..... 1191’ Genera \' Ca "... C ‘ a: « to en Esp-\‘Ber‘en S C 1"" :10 0L 733‘33.-n.h93 l {ksa ACCOra :“Ses genetic T: ‘3!“ OUId ~15: Mia?! ‘ ‘ ‘ “2': 0 law‘ a?“ g OCC ‘- 24 Antilles the genus is limited strictly to montane areas. The resulting isolation of populations in various moun- tainous areas may have been one factor in the extensive speciation in section Lyonia in the West Indies. Lyonia is a very ancient genus and was probably a component of the Arctotertiary Geoflora, a great meso- phytic forest which once extended around the globe in northern regions.l Evidence for this is seen in the distribution pattern of the genus (Figure 1). Lyonia shows a Tertiary relic disjunction pattern between eastern North America and eastern Asia. This pattern was first brought to the attention of American and European scientists by Asa Gray through a series of papers published between 1840 and 1878 (Wood, 1972) . Many other genera show similar patterns, a few of these are: Calycanthus, Perideridia, Desmodium sect. Prodo- carpium, Illicium, Shortia, and Gelsemium. In the Ericaceae Tripetalea and Elliottia, Pieris, Epigaea, 1Wolfe (1966), however, believes that the concept of the Arcto-Tertiary Geoflora is not valid. One of the basic tenents of the Geoflora concept is that the plant association has maintained itself with only minor changes in composition for several epochs or periods of earth history. According to Wolfe "it is extremely improbable from the genetic and physiological viewpoint that many lineages could have remained in association throughout the Tertiary.” He states that "The floristic changes that have occurred in the vegetation of the Pacific North- west during the Neogene indicate strongly that the present associations are also most probably transitory." 25 & ‘ .4 . ‘ V.. W .... \u . t. .\ n‘ t o g Q K's. u I N u A. . I a 9’5 a . " ... n. u v s . a ., . .. n ., , . x . C . w . u ‘ .. . MHCONA mo coflusnfluumflp UHHOB .H musmflm 26 252333355 Show i:s::ibution patte ___a_ (Li, 19 Host of t. 555:9:3 North kte 31-? are also phy $83165. W39“ t1" EISiiered. men 533.9 is easily 1 :".:~ . ‘ .U- ~91.“ to faml L % 1C activl .ertiar' perioq wulff ( QEMOPe :5.“ Seem Mary flora. 55 leave 0‘ 27 (and Chiogenes show similar disjunct distributions. The (distribution pattern of Pieris is very similar to that of Lyonia (Li, 1952; Wood, 1972). Most of the genera showing the eastern Asian- eastern North American disjunction pattern are woody; :many are also phylogenetically isolated with only a few species. When the taxonomic status of these genera is (considered, their relative antiquity in the phylogenetic scale is easily seen. A large proportion of the genera jpertain to families that are considered structurally primitive, such as the Ranunculaceae, Santalaceae, Mag- noliaceae, Berberidaceae, Hamamelidaceae, Theaceae, and Ericaceae (Li, 1952). Over the years paleobotanical findings have shown that these disjunct genera, espe- cially the woody ones, once occurred over wide areas from which they are now missing, and it is clear that the extant representatives of these genera are the sur- 'vivors of the gradual climatic deterioration, volcanism, «orogenic activity, and the glaciers that followed the {Tertiary period (Li, 1952; Wood, 1972). 'Wulff (1943) also points out that the Ericaceae in Europe seem to be a remnant of a more widespread frertiary flora. He points out that the xerophytic ever- (green.leaves of many of the species and the forced winter dormancy in floral development in £29.51 carnea indicate 'that these species were originally adapted to less agorsus clmat Exaceae are i gestures to HZ". pried from st; Bell and B; :3 find that the: '36 i‘lnter sta: V3 ‘~ N ‘e‘aill at,"- - z 1'“ . p 31519 for alias“ any ‘ “tied the; 535.: t0 be its. "‘1 _ s 1n: “=5 28 rigorous climatic conditions. Wulff's theory that the Ericaceae are in "disharmony" with the low winter tem- jperatures to which many are now exposed seems to be sup- jported from studies of floral and shoot development (Bell and Burchill, 1955; Lems, 1962). Bell and Burchill found that there is a continuous developmental series in the winter stages of the floral buds in the genera they studied, and with unusual weather conditions, some member of the group may be found in bloom during any one of the 12 months of the year! Active mitosis was observed in representatives of three genera even in the "depth of ‘winter” when the temperature was below freezing. They found that "Vaccinium pennsylvanicum [which they studied in detail] apparently cannot endure exposure to such ‘winter temperatures as commonly prevail in the regions to which it is native." These facts seem to indicate that the Ericaceae were evolved in a region where it was possible for the plants to flower for most of the year ‘without any definite resting stage. Lems (1962) also concluded that most species in eastern North America .appear to be imperfectly timed to the alteration of summer and winter. They have slow morphogenetic cycles smith only Oxydendrum arboreum leaving no exposed flower <1r inflorescence buds during the winter. Thus a tropical “| b~ “_ .iej \a which «“e N\ qeglon w 1‘22“. ‘ s Dis] 3». “-W4 «“4 sever n; A. ‘ ‘ i‘ahta . a .‘ :vc'fi,. Uri b \L‘tta,‘ . ...:‘aY ‘10 ‘ one '7‘" ' was ‘1 17:1“. . V 3311 30 It is evident that the geographic range of Lyonia has not always been as it is today. There is now considerable evidence that significant vegetational dis- placement took place during times of glacial advance (Frey, 1953; Whitehead and Barghoorn, 1962; Watts, 1969; ‘Whitehead, 1972).2 The full-glacial pollen assemblages in these studies conducted in the southeastern United States bear little resemblance to modern pollen assem- Iblages of the same regions. Components of the modern flora are not represented, and pollen, spores, seeds and other remains of northern species are well represented. Whitehead (1972) in his study of the developmental and environmental history of the Dismal Swamp states that Taxodium-Nyssa forests have only characterized the swamp for the past 3,500 years; before that time the region *was occupied by more northern forest types. Lyonia lucida which today has the farthest northward extension (of its range in the Dismal Swamp, probably migrated into 'the region with many other southern species about 3,500 years ago or slightly earlier. Disjunct populations of E. lucida are found around several ponds in Bartow Co., Georgia, northwest of Atlanta. Watts (1969) found that familiar southern 2Braun (1947), however, felt that major changes (If vegetation were limited to the Coastal Plain with the floral composition of the Appalachian highlands being only insignificantly affected. :ee genera such a :ez'lciaity of t'." E1511: species suc‘r 51:3: appear in t? ::::l:ies that the c 1“:u..“A§ I “V ‘M . , inclui: 1:129 the Postgl. fate had a more SI gazlal period an -. tie glaciers. u‘hc \' . ~~ Jed Jersey pi ;: c-L . we sandy pm ...:J. Lyonia r: 255 A ' - :5 a , e.rly Colon; “E? 1 L is to th A ‘h‘ ‘t I the 31 tree genera such as Liguidambar and Nyssa migrated into the vicinity of these ponds in postglacial times. Coastal Plain species such as Itea virginica and Psilocarya nitens first appear in the fossil record in the Postglacial. He concludes that they and probably other "Coastal Plain disjuncts” including Lyonia lucida reached these ponds during the Postglacial. It thus appears that Lyonia must have had a more southern distribution during the full- glacial period and has migrated northward with the retreat of the glaciers. Thus L. mariana, which is abundant in the New Jersey pine barrens, is also common to the north in the sandy pine barrens of Long Island (Harshberger, 1916). Lyonia mariana is a characteristic species of the sandhills in the Carolinas and probably was one of the early colonizers of the sandy coastal plains when they were uplifted after the Cretaceous inundation (Duke, 1961), Duke describes plants such as these as being indigenous to the sandhills, but not endemic there; rather, they are able to become established in ecologi- cally similar habitats. Thus E. mariana may have origi- nated in this region, migrating southward and northward in.its colonization of the ecologically similar Coastal ‘Plain. Fernald proposed a Virginia-centered classifi- cation of floral elements that later became distributed throughout the sandhills area of the fall-line. Duke (1961) noted that a northern Florida center is more sizingful. Jar. aata refugium 3223:9035; they :fiispersal. v resurarized 1 Tertiary uplif - lastal Plain s leeward across Eli‘s 3ast soc? Q ShO’u'S é 1152'xzct occur} 1:" v 1 ' , "'941 neighbe l . .2 32 Iueaningful. James (1961) and Woodson (1947) suggested that a refugium existed in northern Florida during the Cretaceous; they also favored a northern Florida center of dispersal. When the ideas of Duke, James and Woodson are summarized the following pattern emerges. After late Tertiary uplift there was a dichotomous radiation of Coastal Plain species to the middle Atlantic states and westward across the Gulf Coastal Plain. Lyonia lucida shows just such a dichotomous distribution! Lyonia mariana shows a clear northeast pattern of migration with disjunct occurrences in eastern Texas, Arkansas, and several neighboring states. Possibly at one time it 'was also found on the Gulf Coastal Plain, but for some unexplained reason became extinct in that region. As the seeds are light and easily wind blown, long distance dis- persal is also a distinct possibility. There are a number of plants which show this dichotomous distribution, a few are: Pinus taeda, Prunus angustifolia, Desmodium strictum, Toxicodendron radicans, Leiophyllum buxifolium, jBreweria pickeringii, Ruellia carolinensis, Smilax glauca and Croton glandulosus. These and many more are listed Iby Duke (1961) and Gillis (1971). The distribution of L. ligustrina var. foliosi- flora seems to show a similar pattern, but with addi- 'tional complications (Figure 2). It appears that radi- ation has occurred in several directions from a point Figure 2. 33 Possible distribution patterns in Lyonia ligustrina. Note: light arrows represent var. foliosiflora; dark arrows represent var. ligustrina. C 34 J.-- ----------------o o -— ----------o--—o---o-4 “‘1' C 0‘ awrls on lohs. 42 Flower The flowers of Lyonia are very similar to those of other genera in the Andromedeae. Palser (1951) made a detailed study of the organography and vascular anatomy of this group in which she gives a tentative classification of the tribe on the basis of floral anatomy. In this clas- sification the first separation is based on the nature of the sepal supply. Other important characters used were the presence or absence of fibers with the septal traces, position and character of the placenta, position of ventral carpel bundles, the reflection of the floral parts, the joint origin of the vascular supply to different organs, and staminal appendages. She puts all the species of Lyonia together in one group which she subdivides into two smaller groups, the first containing L. lucida and .23 mariana, and the second 2. ligustrina and L. fruticosa. This further points out the distinctness of section Maria, and the similarities in floral morphology between sections Arsenococca and Lyonia. She points out that all these species show a considerable nondivergence of vascular strands to different organs, have horizontally "S”- shaped filaments, and spurs which show disintegrating ‘tissue. Her statement that they all have a very short 3pair'of awns on each anther is not borne out by my observations . Tale C eight fused s itsed to its Ltses can be Lie Calyx 10': «395, While 4575 Short tr 43 Calyx The calyx is composed of five or rarely four to eight fused sepals. Each lobe is nearly free, being fused to its neighbors only at the very base. The calyx lobes can be glabrous or densely pubescent. In section Lyonia they are covered with peltate scales. The calyx is usually persistent, but it falls with the leaves in 2, mariana. There is marked variation in the length of the calyx lobes; section Maria has long and lanceolate lobes, while both section Arsenococca and section Lyonia have short triangular lobes. Corolla The corolla varies from cylindric to urceolate or globose-urceolate, with five (rarely four to eight) short recurved lobes (Figures 3, 4, 5, 6, 7). It is 'most often white, but occasionally plants with pinkish flowers are found. It is pink or sometimes even reddish in L. lucida. In section Arsenococca the corolla is pilose, with club-shaped glandular hairs very similar to those found on the leaves. In section Maria the corolla is usually glabrous, although in L, mariana there are occasionally a few nonglandular hairs, usually following the veins. Section Lyonia is unique in that the corolla is covered with peltate scales. The corolla is urceolate in sections Arsenococca and Lyonia, but is «cylindric to globose-urceolate in section Maria. Figure 3. 44 Drawing of flower of Lyonia ligustrina. a. Stamen, distal View, X 50; b. Stamen, proximal view, X 50; c. Flower, X 9; d. Longi-section of immature capsule, X 20; e. Cut-away View of flower, X 30. u f t . O I Figure 4. 46 Drawing of flower of Lyonia ferruginea. a. Stamen, proximal view, X 43; b. Stamen, distal view, X 43; c. Flower, X 10; d. Longi-section of immature capsule, X 20; e. Cut-away view of flower, X 30. —- eat (3 I. 4e \ ( V?{.‘?A z. Figure 5. 48 Drawing of flower of Lyonia fruticosa. a. Anther and portion of filament, distal view, X 73; b. Anther and portion of filament, proximal view, X 73; c. Flower, X 13; d. Longi-section of flower, X 30. 50 " 7?: ire; . y 1...“, “Li-F 4'”, E) f ‘ __ q‘ r:- {ff-fiv‘ l . ,..._I . . ,4. - '4;." 1 . . " ‘ '0 ‘44‘3 '. i 2; ‘4. I’.—— ‘le‘lfimrg‘— .' n.1‘4‘w‘ 1:» .. a...“ \WW ‘ .- ‘ +2 Figure 6. Drawing of flower of Lyonia mariana. if i ) Ir i- a. Anther and portion of filament, proxi- mal view, X 34; b. Anther and portion of filament, distal view, X 34; c. Flower, X 4.5; d. Longi-section of immature cap- sule, X 13; e. Cut-away view of flower, X 9.5. .Io MI. .0. I. I 92.310.» «EV/7.... . a so. us.l\l.u~...¢.~ - .. o l 51 n! .u-fp\0.nv.-o - . n L... ‘ fiat. \u a .. ...” HIV 0 I... u . .. . Ira ... .... . pills. a I . \ .Ho- noun! ..- a . u . ..\ ... usual .... ~..\ h ‘ ... 0-. ~ . n- ... . ..ua slsn~ . .4 . v .. f . .. I . u . . . . - . ... . o . a I . . V .I \\ o I ’0.“ o a a a v o . a ‘nl a u f . 06M.- . at uo~l h.» <- \\.- Vs . s I ... u 6.. I .fl" m *9) a Ice-cu. . . 1v . h n .n. a» s Figure 7. 52 Drawing of flower of Lyonia lucida. a. Anther and portion of filament, distal view, X 30; b. Anther and portion of filament, proximal view, X 30; c. Flower, X 4.5; d. Longi-section of immature capsule, X 17; e. Cut-away view of flower, X 11. corolla and if the Eric $631.19 by Peres, and ‘35.“! “90' :1“..\v-.S ha 54 Androecium The ten stamens are borne at the base of the corolla and show most of the peculiar characteristics of the Ericaceous stamen. They have inverted anthers, opening by terminal (morphologically basal) elongate pores, and lack apical awns. The filaments are flattened, slightly to moderately expanded near the base, "S"-shaped, glabrous to hairy or roughened, and with or without a pair of short, spur-like appendages on the back near the apex (Matthews and Knox, 1926; Anthony, 1927; Palser, 1951; WOod, 1961; and personal observations). The stamens always have a white line of disintegration tissue on the back of each lobe extending at least along the apex of the filament and along the upper edge of the spurs when present, thus forming an inverted "V" shaped pattern. (This strange tissue is also sometimes seen on the inside of the corolla.) The included stamens are arranged in a tight ring with their pores facing inward and pressed against the fusiform style (Figures 3, 4, 5, 6, and 7). In L. ligustrina the lower portion of the filament is jpubescent, and only small inconspicuous spurs are present. CThese spurs often disappear with age.) The filaments of _I_._. mariana are pubescent throughout their entire length; L_. lucida has only roughened filaments, but in both noticable spurs are found. All the members of seetion Lyonia ‘- ve extremely filaments. The 0‘ large (especi enlargement c nectar secret the sumit of straight, and time is trl Pilose in 1. ...10a and L LI. section L 55 section Lyonia native to North America lack spurs (or have extremely small ones) and have short roughened filaments. Gynoecium The ovary is superior, five locular, the placentae large (especially in section Maria) and undivided. An enlargement of the wall at the base usually functions in nectar secretion. The style arises from a depression at the summit of the ovary and is columnar to fusiform, straight, and included (or rarely slightly exerted). The stigma is truncate to capitate. The ovary is densely pilose in L. ligustrina, glabrous to nearly so in L. lucida and E. mariana, and densely pubescent and lepidote in section Lyonia. The ovary is generally subglobose in L, ligustrina and section Lyonia, but is often flattened- ovoid in L, lucida and characteristically urn-shaped in L. mariana. Fruit The fruit is a five-angled, subglobose to ovoid, loculicidal capsule with prominent pale, thickened sutures, which may separate from the five valves in dehiscence (Figure 8). The placentae are persistent at the top of the columella. The capsule of L_. mariana is iniique, usually ovoid-triangular with a contracted truncate apex. Sometimes the capsule of _11. lucida shows Figure 8. 56 Variation in capsule morphology and dehiscence in Lyonia. a. L. squamulosa, partially dehisced, X 13.5; b. E. ferruginea, just beginning to dehisce, X 13.5; c. g. fruticosa, unopened, X 13.5; d. E. fruticosa, completely dehisced, X 13.5; e. E. mariana, just beginning to dehisce, X 8; f. E. lucida, unopened, X 10; g. E. lucida, completely dehisced, X 10; h. E. ligustrina var. ligus- trina, unopened, X 8; i. E. ligustrina var. foliosiflora, partially dehisced, X 8. ‘r‘y‘g. I .~ '95 ’i 0'... 57 '1’ ' ’.'..¢O . .“a'ohfi. ha. . 41.. | ararked resent is never as dis section Lvonia extremely thicj Less thickened "‘u' Ara usually Marately thi n‘ . . *1;- mariana \ '35 Sutures. is a unit is n Oi «£53 "“R‘er Species, 58 anwrked resemblance to this unusual shape, although it is never as distinct as in E. mariana. The fruit of section Lyonia is always lepidote and pubescent with extremely thickened sutures. The capsule is pilose with less thickened sutures in section Arsenococca. Section EEEEE usually has glabrous capsules with slightly to moderately thickened sutures. The western populations of E. mariana often have scattered nonglandular hairs on the sutures. The tendency for the sutures to separate as a unit is most strongly developed in species of section Lyonia although it is occasionally seen in the other species, especially E. mariana. All ’c elliptic flov tissue at the large termine :urved f ilame Style. In L. Pores open is this Spe Amity at aye by the “he 10mg ha} teams at ‘5‘L9r POLLINATION BIOLOGY All the species have pendulous urceolate or elliptic flowers, included stamens, and a rim of glandular tissue at the base of the corolla. The anthers open by large terminal pores which are directed inward, with the curved filaments holding the anthers against the fusiform style. In E. ligustrina the anthers are mature, and the pores open in buds about to Open. When the filaments are bent outward and allowed to spring back, the pollen is discharged. The stigma is at the same level as the mouth of the corolla, rendering the opening so small that bees visiting the flower cannot fail to come in contact with it. It appears to be receptive in newly opened flowers (Lovell and Lovell, 1935). The Lovells also observed bumblebees sucking nectar from the flowers of this species. The nectar accumulates in considerable quantity at the base of the flower and is probably held there by the densely pilose pubescence of the ovary and the long hairs at the base of the filaments. The bee's tongues at first passed between the curved filaments, but later, as the bees turned half to three-fourths of the 59 say around the :he filaments exsiusively bj The b1 bybr‘ashing t] he dusted on ; carried to an: IZlCh almost I ise Vibrathn to be dUStEd ( Fillination CC 1'26 anthers tc LOvel] hash fOr two i received sever Mala“. and {Bits in One Judgir 60 vmy around the flower, their tongues passed outside of the filaments. The flowers were pollinated almost exclusively by bumblebees. The bumblebees probably accomplished pollination by brushing the curved filaments, thus causing pollen to be dusted on its proboscis. The pollen could then be carried to another flower and deposited on the stigma which almost blocks the small corolla opening. Possibly the vibration of the bee's wings also causes the pollen to be dusted on the stigma and the bee itself. Self— pollination could easily occur by pollen falling from the anthers to the stigma below. Lovell and Lovell (1935) in observing a large bush for two hours noted that all the flowers probably received several visits. Bombus ternarius was quite abundant, and one speciman was seen to make sixteen visits in one minute. Judging from their similarity in floral morphology, the pollination mechanisms of the other species in the genus are probably quite similar to that of E. ligustrina. Honeybees have been observed visiting the flowers of E. ferruginea in northern Florida (Richard Harris, oral communication). The long spurs present on the filaments of E. mariana and lucida possibly also function in pollination. These spurs stick out toward the corolla and when the grcboscis of a flower it may causing the pc mechanism is c sgecies of E Late corollas, LOVel Y I - \ The netted pa 323?. as many 61 proboscis of a bee is inserted into the bell-shaped flower it may strike against one of these processes, causing the powdery pollen to fall out. A very similar mechanism is described by Knuth (1909) for several species of Vaccinium which also have pendulous, urceo- late corollas and prominent spurs. Lovell and Lovell covered several panicles of E. ligustrina with a fine netting to exclude all insects. The netted panicles produced many seed capsules, though not as many as from clusters of flowers which had not been covered. Thus both self- and cross-pollination are probably important in this Species. The E: ifications in 13.93rtance in C'I'course, tfv lzclasSifl’in f‘lLCtion in t tlll‘sate and h filth the hate :Etranspirat COWar EP I DERMAL APPENDAGES The Ericaceae as a family show many curious mod- ifications in trichome structure, which often are of importance in determining relationships among species. Of course, the trichomes are not there just to help us in classifying! Trichomes probably have an important function in the economy of the plant in its relation to climate and habitat. They may be intimately concerned with the water balance of the plant and in the regulation of transpiration. Cowan (1950) pointed out that the Rhododendrons of higher altitudes, in order to survive the rigors of winter, or periods of drought, must either severely limit or arrest transpiration for a long period. This is accomplished in some species by a dense covering of hairs, but most are equipped with peltate scales. Thus he concluded that scales when closely packed furnish the plant with a means of restricting transpiration even more efficiently than a covering of hairs. When these scales are so close together that the rims of adjacent ones overlap, a protective canopy is formed under which 62 transpiration scales in Rhoc tire function the leaf and l Iowan, 1950) correlation b :at. For exa :zdes tend to scales. He c have varied 8 Efficient the ."\v Ru 63 transpiration and gas exchange can occur. The peltate scales in Rhododendron may also provide a water absorp- tive function, by trapping the water of rain and dew on the leaf and holding it there until it can be taken in (Cowan, 1950). Cowan noted that there is a definite correlation between the type of pubescence and the habi- tat. For example, epiphytes and species from high alti— tudes tend to be more densely covered with hairs or scales. He concluded that "these diverse structures have varied and contrasting functions but none is more efficient than the well-known Rhododendron scale, a minute organ, wonderfully made." The Rhododendrons are not alone in their possession of such an adaptation. Peltate ferrugineous scales quite similar to those of Rhododendron are present in E. £25: ruginea, E. fruticosa and E. squamulosa. In fact, the presence of such scales is a diagnostic feature of sect. Lyonia. Their function is probably very similar to that of the peltate scales of Rhododendron described above. These three species often occur in xerophytic conditions. Eyonia fruticosa, especially, is abundant on the sandy plains of southeast Georgia and Florida where almost every plant shows at least some xerophytic adaptations. Probably one of the reasons why E. fruticosa can prosper under such harsh conditions is because of the possession of this protective covering of scales. Lyonia ferruginea has the additii Leaves. Em :‘tese plains, restricted to srons or the « The c :‘rc‘Dibly also interest that are generalh NlEd1 tense of the aClassifica artificial. 'a‘it'n M scales. He 64 has the additional protection provided by its revolute leaves. Lyonia lucida, another evergreen Lyonia on these plains, lacks peltate scales and is usually restricted to more moist sites, such as slight depres- sions or the edge of ponds or streams. The club-shaped glandular hairs of E. ligustrina probably also function in water regulation. It is of interest that the mountain populations of this species are generally quite pubescent. Niedenzu (1890) used the leaf anatomy and pubes- cence of the Arbutoideae and Vaccinioideae to construct a classification for these groups. The result is rather artificial. For example, he classified E. ferruginea xvith Cassandra calyculata because they both have peltate scales. He gives illustrations of the peltate scales of Ii' ferruginea and the stalked glands of E. ligustrina. I found each section of Lyonia as represented in herth America to have a distinctive pubescence type. Itikey is given below which brings out the distinctive features of each group. Illustrations of each trichome type are given in Figures 9-12. Key to sections and species of Lyonia based on trichome type 1x. Peltate scales present on leaves (Section Lyonia) . . . . . . . . . . . . . . . 65 Papillae-like hairs abundant or moderate, pel- tate scales usually not dark red . . . . . . C C. Papillae-like hairs abundant (rarely absent), scales mostly deciduous thus giving older leaves a whitish appearance beneath . . . . . . . . . E. fruticosa C. Papillae-like hairs moderate, scales not" deciduous, thus older leaves rusty colored beneath . . . . . . . . . L. ferruginea Papillae-like hairs absent (rarely present), pel- tate scales dark red to almost transparent . . . . . . . . . . . . . E. squamulosa Peltate scales absent from leaves . . . . . . . D D. Leaves with multicellular, clear or white, elongate, club-shaped hairs, sometimes also with filiform hairs intermixed; corolla pilose (Section Arsenococca). . . . . . E. ligustrina Leaves with small, rounded, red-tipped, stalked glandular hairs, glandular tip sometimes drawn out into a definite point, filiform hairs often along major veins; corolla nearly glabrous (Section M)...............E E. Glandular hairs all with small rounded tips, reddish, usually with filiform hairs along major veins beneath . . . . . . E. mariana E. Gla gre elc ha: SU 66 Glandular hairs fewer, usually deciduous with greater age, of various types with rounded to elongate tips, reddish to white; filiform hairs only on midvein of upper leaf surface . . . . . . . . . . E. lucida Figure 9. 67 Variation in glandular hairs in Lyonia; darker shading indicates a reddish color. a-b. E. mariana, ca 160 u long; ligustrina, ca 525 u long; d-i. 100-400 u long. c. L. E. lucida, Figure 10. 69 Variation in size and shape of fer- rugineous scales of Eyonia ferruginea (ca 0.20 to 0.75 mm wide). O . . on .. .Ii .‘. .- ..u. ......r....:. .. i . .. . . I . .. \ ...\. o a u n f“ on-aooao I a . .91. u. . \ .\\ \o '5‘ c u . . . . o I I . \l.\ u ‘ O ‘ C ...alu..lu-|\ ‘IO'II‘OO u o . u ., 0. u.... at... .Illv..00n..oucf t b O'...’ I . 0 .ID 0 l to. I O ‘ u \ o o n- o o e o o u. a. fell. a (-0 9 0 I I 71 A.n CH wmcflmsuumm .m m0 mafia: Euomwafim umHfiEwm mnu muozv .mcoa 2 omH mo .mmoowusum am no momwnsm mama Hmwxmnm so muwmn EHOMfiHflm .o unmfln n oma mo .mmOOfiusum .m mo mamom msosowomn mo uncenomuum Human mo mchEmm .0 unwound 1 00¢ mu .mmcfimsunmm .m mo mommusm mama Hafixmnm c0 mmamom msomcwmsuumm m0 mGOHuomm Iflmcoq .Qnm .masomq coauomm ca momma Hams mo m3mw> Hmcwosuwmcoq .HH ousmflm Figure 12. 73 Hair types in Sections Maria and Arseno- cocca. a. External view of filiform hair of Lyonia lucida, ca 200 u long; b. Longi-section of filiform hair of E. ligustrina, ca 300 u long; c. Longi- section of filiform hair of E. mariana, ca 400 u long; d. Longi-section of glandular hair of E. lucida, ca 150 u long. Pro) eaten. The PUison andr :0 kill She Stagger-bu5 leaves on in Old Pas Scraping I“ \.S\Qc no , x “This“ ECONOMIC IMPORTANCE Probably all species of Lyonia are poisonous when eaten. The leaves of E. mariana contain the narcotic poison andromedotoxin, C31H51010, and have been known to kill sheep. In fact it received its common name, stagger-bush, because of the intoxicating effect of its leaves on sheep and cattle (Pammell, 1911; Muenscher, 1939)! In Arkansas where E. mariana is locally abundant in old pastures, cut-over woods, sandhills, and along road sides, farmers have tried to eradicate it by scraping it up, etc. All such attempts have been unsuccessful (Delzie Demaree, oral communication). The extensive underground rhizome network produced by E. mariana probably accounts for the difficulty in its removal. Lyonia ligustrina is also said to cause livestock Ix:isoning, producing staggering and disorientation in goats. Usually the major problem is in early spring when it is one of the few things in leaf (Correll and .Johnston, 1970; Muenscher, 1945; Delzie Demaree, oral communication) . 75 Pannel? species‘ of EL these, but sir bush probably of L. ovalifo - \ ”ElaY Peninsu Scats (Sarge: An i a“'*'910Y‘3<1 ext diseases! in insects (Sat have also b states; up bdhdsgme f( in amlDle c recogniZed with its 1 has a tent 1V0: ' Geekiful 76 Pammell (1911) states that "all three [sic] species” of Lyonia are poisonous. He does not list these, but since E. lucida is also often called stagger- bush probably it, too, is poisonous. The leaves and buds of E. ovalifolia, a small tree of the Himalayas, the Malay Peninsula, China, and Japan, are poisonous to goats (Sargent, 1893). An infusion of the leaves of E. ovalifolia is employed externally in the treatment of cutaneous diseases, and the young leaves are used to destroy insects (Sargent, 1893). Several species of Lyonia have also been used as ornamental shrubs. Sargent states: ”Most of the species of Andromeda produce handsome foliage and beautiful flowers often arranged in ample clusters, and their value as garden plants is recognized in all temperate regions." Lyonia mariana ‘with its large white flowers is especially beautiful but has a tendency to have a rather straggling growth. JLyonia lucida and E. ferruginea also would make very Jbeautiful ornamentals, and more study should be directed toward their growth and culture. NUMERICAL TAXONOMIC STUDIES The use of electronic computers in the analysis of large amounts of comparative data has become a powerful and popular technique in plant systematics. The field of numerical taxonomy began in the 1950's and has gained ‘widespread attention since the publication of Sokal and Sneath's Principles of Numerical Taxonomy in 1963. It is still somewhat controversial with various people advocating one or another of the many methods that have been proposed since the 1950's. Two major viewpoints exist: the 2E3: 23523 school and the phyletic school. The followers of the phenetic school believe that numerically generated classifications should reflect relationships based on ‘as many characters as possible and should not take phylo- geny into account (Hsiao, 1973; Sokal, 1963; Sokal and Sneath, 1963; Sokal and Crovello, 1970). In contrast, the followers of the phyletic school believe that quanti- 'tative evolutionary schemes are the primary goal (Kluge ,and.Farris, 1969; Cain and Harrison, 1960; Farris, 1970; Whiffin and Bierner, 1972; Wagner, 1962, 1969; Mayr, 1965). 77 In t] approaches w Ilith the 0th Clu: malfléd ve 3330313 is a Ii.-- 78 In this study both the phenetic and the phyletic approaches were tried. The results of each are compared with the other and with previous classifications. Phenetic Studies Cluster analyses were performed using a slightly modified version of program HGROUP of Veldman (1967). HGROUP is a hierarchical grouping program which utilizes a generalized distance function based on the concept of error sum of squares or within-group variance (i.e., the sum of squared deviations from group means). The program uses the total within-groups variation as the function to be minimallyincreased at each step in the grouping pro- cess. Principal components analyses were made using program FACTOR of Veldman (1967). This program reduces a set of variables that were used to gather data (e.g., leaf length, corolla color) to a smaller set of new, uncorrelated variables which are defined solely in terms of the original dimensions, and which retain the most ”important" information contained in the original data. 'The process can be thought of as the construction of a :neW'space which is maximally representative of the space «defined by the original variables (Veldman, 1967). All programs were run on the CDC 6500 computer at Michigan State University. Taxa for numerical analysis were first selected ‘by conventional herbarium methods (i.e., species were sorted “to 1 within and d variable E. three 9609“ an Operatic northern va western POE an indepenc considered seven taxa A from the I examinatiI acters th specimens measured. character such as 1 presence be invar fliese We “3 (Tab Chained m' "& tation -L 79 sorted into sets showing continuous patterns of variation within and discontinuous patterns among sets). The widely variable E. ligustrina var. foliosiflora was divided into three geographical regions, each of which was treated as an Operational Taxonomic Unit (OTU). The more homogeneous northern variety was treated as a single OTU; the disjunct western populations of E. mariana were also considered an independent OTU. In all other cases each species was considered to be an OTU. In all, ten OTU's representing seven taxa were selected (Table 3). A tentative list of characters was abstracted from the literature, and many others were added after examination of specimens from all taxa. The 144 char- acters thus obtained were measured on each of 10 to 84 specimens of each OTU. In all, 366 specimens were :measured. The measurements were continuous in such characters as length or width, multistate in features such as color or degree, and dichotomous in cases of ;presence or absence. Several characters were found to Ibe invariant, or were difficult to measure with accuracy; these were eliminated, leaving 128 characters for analy- sis (Table 4). The specimens from which the data were obtained are indicated by an asterisk (*) in the specimen citations of the taxonomic treatment. Cluster analyses of the OTU's were carried out using all 128 characters and also the 79 vegetative LIGM LIGE LIG] MAR LUQ 80 TABLE 3. List of OTU's of Lyonia, giving the code name of each. Code Name Taxon FRUT FERR SQUA LIGW LIGM LIGE LIGL MARW LUCI onia fruticosa )3 . ferruginea k* lb . squamulosa L. ligustrigg var. foliosiflora Tbklahoma, Texas, Afkansas, Louisiana) L. ligustrina var. foliosiflora YTennessee, Mississippi, Alabama) L. ligustrina var. foliosiflora TFlorida, and northward on coastal plain) E. ligustrina var. ligustrina L. mariana Twestern disjunct populations) L. mariana Teastern Coastal Plain) E. lucida *2an 4. en (E % C} x 1. VEGETAT 1. Hab 2. Nun one [rei— {Characters "4. 3. Lu: 4 Lo 5. De la me 6. Le ir 7, 0‘ m i] 8' DI d1 h. ( 9. L. i: 81 TABLE 4. Characters used in numerical taxonomic study (phenetic method). Character Character States I. VEGETATIVE CHARACTERS 1. Habit 1. Small shrub (1' to 4') 2. Large shrub (5' to 10') 3. Small tree (over 10') 2. Number of nodes in * one year's growth As actual measurement (Characters 3 through 17 refer to twig characters) 3. Luster l. Dull 2. Somewhat shiny 3. Very shiny 4. Longitudinal ridges 1. Absent 2. Slightly angular 3. Very angular 5. Density of nonglandu- 1. Glabrous lar hairs at inter- 2. Sparse nodes (mature) 3. Moderate 4. Dense 6. Length of hairs at 1. Short internodes (mature) 2. Medium 3. Long 7. Density of ferrugi- 1. No scales neous scales at 2. Sparse internodes (mature) 3. Moderate 4. Dense 8. Density of glandular l. Glabrous and nonglandular 2. Sparse hairs at internodes 3. Moderate (immature) 4. Dense 9. Length of hairs at 1. Short internodes (immature) 2. Medium 3. Long TABLE 4. Cc Cl “ 10. Den: neo int (in 11- Col int 12. Ra 13- Si 14. c 15, lg. 17 I [3. 3“: at is. 82 TABLE 4. Continued Character Character States 10. Density of ferrugi- 1. No scales neous scales at 2. Sparse internodes 3. Moderate (immature) 4. Dense 11. Color of scales at 1. All or most dark orange internodes (mature) 2. All brown or black 3. Half orange, half white or clear 4. All white or clear 12. Rato of small to 1. All small large scales 2. Some large, mostly small (mature) 3. Half and half 4. Some small, mostly large 5. A11 large 13. Shape of scales 1. Small and regular (mature) 2. Large and regular 3. Small and irregular 4. Large and irregular 14. Color of scales at 1. All dark orange internodes (immature) 2. All brown or black 3. Half and half 4. All white or clear 15. Shape of scales 1. Small regular (immature) 2. Large regular 3. Small irregular 4. Large irregular l6. Persistence of 1. Persistent scales 2. Deciduous 17. Texture of older 1. Smooth twigs 2. Somewhat rough 3. Shredding bark (Characters 18 through 25 concern 18. Shape l. 2. the winter buds) Round Elongate it ll 22 2 83 TABLE 4. Continued Character Character States 19. Apex 1. Rounded 2. Sub-acute 3. Acute 20. Length *Measured in mm 21. Diameter at widest *Measured in mm 22. 23. 24. 25. 26. (Characters 27 through 76 refer to 27. 28. 29. point Rato of length to diameter Density of hairs on bud Density of scales Color of scales Variance in leaf size Base Angle made by base Serration type (distal portion) *Calculated from 20 and 21 PI sawed .boawrd obbJNFH O O O O O 0 O O O O I 0 LAN 0 I'U'IthNH ooooo Glabrous Sparse Moderate Dense None Sparse Moderate Dense Dark orange Light orange Clear Leaves becoming very reduced Moderately reduced Not reduced at top of branch the leaves) Attenuate Cuneate Obtuse Truncate Cordate Measured in degrees 1. 2. 3. Entire Serrulate Serrate TABLE 4. Ce % Ch K 30. 31. 32. 33. 34. 35. 35, 37 38. 39 40. 41 42. Serr (prc Revc (prc Revc (die Per: Nunu per (prc per (dis Len: (prc Len: (di: Hei. Thi. 84 TABLE 4. Continued Character Character States 30. Serration type 1. Entire (proximal portion) 2. Serrulate 3. Serrate 31. Revolute margin 1. Not revolute (proximal) 2. Slightly revolute 3. Very revolute 32. Revolute margin 1. Not revolute (distal) 2. Slightly revolute 3. Very revolute 33. Persistence 1. Deciduous 2. Evergreen 34. Number of serrations *Actual number counted per unit length (proximal) * 35. Number of serrations Actual number counted per unit length (distal) * 36. Length of serrations Measured in mm (proximal) 37. Length of serrations *Measured in mm (distal) 38. Height of teeth 1. Very short 2. Medium 3. Long 39. Thickness of blade 1. Thin 2. Moderately coriaceous 3. Very coriaceous 40. Apex 1. Obtuse 2. Mucronate 3. Acuminate 4. Acute 41. Angle made by apex *Measured in degrees * 42. Length Measured in mm «)6. 47. 48, 49, so, 51. Rat Rat ler Co Lu 85 TABLE 4. Continued Character Character States 43. Length from base to *Measured in mm 44. 45. 46. 47. 48. 49. 50. 51, widest point Width at widest point Ratio of length to width Ratio of length to length at widest point Color above Luster above Density of non- glandular hair on veins below Density of non- glandular hairs on veins above (mature) Length of non- glandular hairs on veins below (mature) * . Measured in mm *Calculated from 42 and 44 *Calculated from 42 and 43 WM 000 o WNH WNH Light green Medium green Dark green Dull Moderately shiny Very shiny No pubescence Only hairs on mid vein Slight pubescence on mid vein and lateral veins Moderate or quite pubescent on mid vein and lateral veins Mid vein glabrous but pubescent on lateral veins No pubescence Slight pubescence on mid vein Moderate pubescence on mid vein Very pubescent on mid vein with some pubes- cence on laterals Short Medium Long TABLE 4. 52. S3. 54. 55. se. 59. 86 TABLE 4. Continued Character Character States 52. Length of nonglandular 1. Short hairs on veins above 2. Medium (mature) 3. Long 53. Density of glandular l. Glabrous and nonglandular hairs 2. Sparse on blade below 3. Moderate (mature) 4. Dense 54. Length of glandular 1. Short and nonglandular 2. Medium hairs on blade below 3. Long (mature) 55. Density of scales on 1. None blade above (mature) 2. Sparse 3. Moderate 4. Dense 56. Density of scales on 1. None blade below 2. Sparse 3. Moderate 4. Dense 57. Color of scales on 1. All brown or black blade below 2. All or most dark orange 3. Half orange half white 4. All white or clear 58. Ratio of small to 1. All small large scales on 2. Some large, mostly small blade below (mature) 3. Half and half 4. Some small, mostly large 5. All large 59. Ratio of small to 1. All small large scales 2. Some large, mostly (immature) small 3. Half and half 4. Some small, mostly large 5. All large 87 TABLE 4. Continued Character Character States 60. Density of scales on 1. None leaf blade above 2. Sparse (immature) 3. Moderate 4. Dense 61. Shape of scales 1. None (mature) 2. Regular 3. Irregular 4. Both 62. Shape of scales 1. None (immature) 2. Regular 3. .Irregular 4. Both 63. Persistence of small 1. Persistent scales on lower sur- 2. Partly deciduous face 3. Deciduous 64. Persistence of large 1. Persistent scales on lower sur- 2. Partly deciduous face 3. Deciduous 65. Glandular hairs on 1. Yes blade above 2. No 66. Average size of *Measured in mm scales 67. Prominence of major 1. Not prominent veins below 2. Moderately so 3. Very prominent 68. Curvature of blade 1. Curved upward 2. Flat 3. Curved downward 69. Prominence of major I. Not prominent veins above 2. Moderately so 3. Very prominent 70. Depression of veins 1. Depressed above 2. Level with surface 3. Raised TABLE 4. C 71. Ma] 72. Le: 73. Ra 74. De 75. Le 76- De 77. Pc 88 TABLE 4. Continued Character Character States 71. Marginal vein present 1. Yes 2. No 72. Length of petiole *Measured in mm 73. Ratio of length of *Calculated blade to length of petiole 74. Density of non- l. Glabrous glandular hairs 2. Sparse on petiole 3. Moderate 4. Dense 75. Length of hairs on 1. Short petiole 2. Medium 3. Long 76. Density of scales on 1. Absent petiole 2. Sparse 3. Moderate 4. Dense (Miscellaneous vegetative characters) 77. 78. 79. II. 80. Position of branches Water relations Location Inflorescence type 1. 2. 3. 4. 1. 2. 3. Rigidly ascending Arching Only somewhat arching No particular orien- tation Growing in standing water Moist soil Dry soil Coastal Plain, Pied- mont or mountains Coastal Plain Tropical mountains FLOWER AND INFLORESCENCE CHARACTERS Reduced raceme (fascicle) Raceme Panicle TABLE 4 . 31. Nu ir 32. Re ta 83. r;- m m 84. N h e 85. 1 1 86, 5 87, < 88, 1 89. 90. 91. 92. 91 89 TABLE 4. Continued Character Character States 81. Number of flowers per *Actual number counted inflorescence 82. Relationship to vege- 1. Flower buds always tative growth above 2. Some mixing 3. Flower buds below vegetative buds 83. Number of panicles 1. Does not apply making up pseudoter- 2. Actual number counted minal inflorescence 84. Number of leafy *Actual number counted bracts per inflor- escence * 85. Total length of Measured in mm floral bracts 86. Shape of corolla l. Urceolate 2. Elongate 3. Ovoid-elongate 87. Color of corolla 1. White 2. Pink 3. Red 88. Pubescence of 1a Glabrous corolla 2. Sparse hairs 3. Dense hairs 4. Scales only 5. Scales and hairs 39. Length of corolla *Measured in mm * 9C’. Width of corolla Measured in mm 91. Ratio of length to *Calculated width 92. Length from base to *Measured in mm widest point 93. Length of calyx *Measured in mm lobes T‘LE 4. 4 % Cl ‘——_ 94. Ra 95. Ni 96. Ra 97. PU 98, Pt 99- Pe no. L. 101. 102, 103. 104 90 TABLE 4. Continued Character Character States 94. Ratio of length of *Calculated corolla to length of calyx lobes 95. Width of calyx *Measured in mm lobes * 96. Ratio of length to Calculated width of calyx . lobes __ 97. Pubescence of calyx l. Glabrous (abaxial) 2. Sparse hairs 3. Dense hairs 4. Scales only 5. Scales and hairs 98. Pubescence of calyx 1. Glabrous (adaxial) 2. Sparse hairs 3. Dense hairs 4. Scales only , 5. Scales and hairs 99. Persistence of l. Deciduous calyx 2. Partly deciduous 3. Persistent 100. Length of filaments *Measured in mm 101. Curvature of fila- 1. Straight ments 2. S-shaped 3. Several curves 102. Pubescence of fila- 1. Smooth ments 2. Spines 3. Spines and long hairs on lower portion 4. Long hairs covering entire filament 103. Length of anther 1"Measured in mm 104. Spur length 1. Spurs absent 2. Short spurs 3. Long spurs NUS 4. C 107. 5‘: 112. T or, PM 114. ‘ 115 115 91 TABLE 4. Continued Character Character States 105. Length of style *Measured in mm 106. Pubescence of ovary l. Glabrous 2. Sparse hairs 3. Dense hairs 4. Scales only 5. Scales and hairs 107. Shape of ovary l. Subglobose 2. Ovoid 3. Elongate-ovoid ‘ 108. Length of pedicel *Measured in mm I * O 109. Length of bractioles Measured in mm 110. Pubescence of pedicel l. Glabrous 2. Sparse hairs 3. Dense hairs 4. Scales only 5. Hairs and scales 111. Width of bractioles *Measured in mm 112. Time of flowering l. March-April 2. May-June 3. July-August 4. September-October 5. November-February III. FRUIT AND SEED CHARACTERS * . J~13. Length of capsule Measured in mm 114. Width of capsule *Measured in mm 115. Ratio of length and *Calculated width 1. Oval to elongate-oval 2. Subglobose 3. Ovoid with a flattened top 4. Urn-shaped 116. Shape of capsule TABLE 4. 119. 120. 126, 127. 92 TABLE 4. Continued Character Character States 117. Thickening of sutures 1. Slightly thickened 2. Moderately thickened 3. Very thickened 118. Pubescence of capsule l. Glabrous 2. Very slight pubescence 3. Moderate pubescence 4. Very pubescent (hairs) 5. Scales only--sparse 6. Scales only--dense 7. Scales and hairs 119. Color of capsule l. Yellow-tan 2. Light brown-orange (or greenish) 3. Dark brown-brown 120. Separation of 1. Not separating sutures from valves 2. Partially separating of capsule 3. Sutures completely separate 121. Number of capsules *Actual number counted per inflorescence 122. Number of seeds per *Actual number counted capsule 123. Length of seeds *Measured in mm 124. Width of seeds *Measured in mm 125. Ratio of length to 1”Calculated width 126. Color of seed 1. Yellowish tan 2. Brown 3. Dark brown 127. Depression of apex 1. Slightly depressed of capsule 2. Moderately depressed 3. Very depressed * 128 Length of pedicel Measured in mm \ (on mature capsule) ‘I'l characters 33 floral . characters loo-charac acters we: Vere found Here elim; was also Finjures 1 tea OTU‘ s the 108-c 9139110ng tive, fr] Table 5 . based up respondi Tables E Pornon Show“ is 93 characters, the 49 fruit and floral characters, and the 33 floral characters. A cluster analysis of the 128 characters themselves and a factor analysis of a similar loo-character set were also made to determine which char- acters were most closely correlated. Several characters were found to be highly linked, a good portion of these were eliminated, resulting in a 108-character set, which was also analyzed by cluster analysis. The phenograms in Figures 13 and 14 represent the cluster analysis of the ten OTU's for the complete (128-character) set, and for the 108-character set. Figures 15, 16 and 17 show Phenograms produced in cluster analyses of the vegeta- tive, fruit and floral, and floral data sets, respectively. Table 5 gives the distance between each pair of OTU's based upon analysis of the 108-character set. Cor- responding values for the other data sets are found in Tables 6, 7, 8, and 9. A graphic representation of a Portion of the information presented in these tables is Shovm in Figures 18, 19, 20, and 21. The phenograms resulting from the 128 and 108 chali‘acter sets are very similar. The phenogram based on vegetative characters is also quite close to these two. All the phenograms indicate that the genus consists of three distinct groups. Lyonia ferruginea, E. squamulosa and E. fruticosa are closely clustered in the first group (section Lyonia), E. ligustrina is by itself in Figure 13. 94 HGROUP phenogram of Lyonia based on 128 character set. Units on axis indicate the sum of squared deviations from group means or error index. ‘ 39 95 FERR SQUA FRUT LIGM LIGE L__J LIGW LIGL MARE MARW 100 200 ’_.___-._—— i 300 360 480 540 —————-—-—’—-——-—'—-—-—— E—JL , —.__.___._._ ...—_— _-—_— LUCI 96 Figure 14. HGROUP phenogram of onnia based on 108 character set. I \n ..Ié'j 1330 ‘Q‘ “ 100 I: 200 400 460 FERRSQJAFRUT if 97 MARW LUCI L194 LIGE LIGW LIGL MARE 98 i A. Figure 15. HGROUP phenogram based on vegetative characters of Lzonia. 99 FERR SQUA FRUT LIGW LIGE LIGW LIGL MARE MARW LUCI 50 100 140 380 1000 100 Figure 16. HGROUP phenogram based on flower- fruit characters of onnia. 101 LIGL LIGW LIGM LIGE FRUT SQUA FEPR MARW MARE LUCI o L.__J LT—J 1 __J |__1__j L___ 50 100 150 200 250 —— 102 Figure 17. HGROUP phenogram based on floral char- acters of Lzonia. 103 FERR SQUA FRUT LIGM LIGE LIGW LIGL MARE MARW LUCI ° L4 Le L___. LJ 20 40 60 80 1130 I T J 104 oo.o HUDA mm.mHH oo.o mam: vo.vma mm.vm 00.0 3nd: Ho.ava ma.nma mm.mma oo.o AUHA nw.m~a ma.~ma mm.mma an.om oo.o mqu ¢>.Hma mm.o¢a mm.mma mm.mm ov.ma oo.o ZUHA om.nma m¢.mna m~.mma Hm.ma ma.ma em.m~ oo.o smug mm.mva mm.mma mm.wma mm.mma mm.NHH nn.mma mv.hma oo.o 490m mm.oma mv.mma mo.mna ah.¢ma mm.maa o¢.mma HH.o¢H mv.am oo.o mmmm ma.mma mo.ao~ mm.oom mo.hva m~.mma mn.m¢a mv.o>a m~.vm mm.mn oo.o 99mm HUDA mmdz zmdz QGHQ MUHA EUHA SUHA dDOm mmmm 82mm .umm HmuOMHM£0ImOH .mamwucmuom Houuw mo xflnumz umHmMHmcm mbomwm .m Manda 105 00.0 HUDA 00.0HH 00.0 mud: mh.mma 00.00 00.0 3m¢z 00.vva 0v.HOH V0.NOH 00.0 AUHA 00.00H 00.00H 0H.0va 00.HN 00.0 muHA 00.00H 00.0va 00.00H 00.5N hm.va 00.0 ZUHA N5.H¢H ~0.mva 0h.0va 00.0H eh.0H H0.0N 00.0 3UHA Nh.00H 05.50H «0.50N 05.00H Hm.0mH 0N.50H 05.00H 00.0 4000 mN.0mH 00.00N N0.0HN 0H.00H 00.00H 0v.HOH HN.h0H 00.05 00.0 mmmm vv.mm~ ma.0v~ 0N.Nvm 00.00H m0.v0a 00.HOH Hm.mHN vv.vh 00.00 00.0 90mm HUDA mmdz 3m¢2 AUHA MUHA ZUHQ 3UHA dDOm mmmh BDmm .umm “muomumsou0ma .mamwucmuom uouum mo xwuumz ”mammamcm macaw: .0 wands 106 00.0 HUDA 00.00 00.0 mad: 0H.00 00.00 00.0 3m¢2 H0.H0 v0.vv 00.0w 00.0 AUHA 00.00 00.H¢ 00.00 00.vH 00.0 HGHA 00.00 00.00 0¢.00 00.0H 00.0 00.0 ZUHA 00.00 00.00 00.00 00.0H 00.0H H0.0H 00.0 zUHA 00.NNH v0.0HH 00.0NH 00.0NH 00.HHH 00.0NH 00.00H 00.0 . £500 N0.0HH 00.0HH 00.00H 00.50H 0H.0HH HN.00H 00.00H 00.00 00.0 mmmm 00.00H 0H.00H 00.00H HN.0¢H 00.0NH H0.¢¢H 0N.v0H H0.00 00.Hh 00.0 BDmm HUDQ mmdz EEG: AUHA meA ZUHA BUHA €300 “mum BDMh .umm Hmuomumso m>flumum0m> .mamwunwuom Houum mo anumz ”mmeHmcm macaw: .5 Handy 107 00.0 HUSH 00.0v 00.0 mud! 00.vv 00.0 00.0 Bud: «0.00 00.00 00.00 00.0 AUHA 0v.ah 00.00 00.00 00.0 00.0 MUHA 00.00 00.00 00.00 H~.0 00.0 00.0 szq 05.H0 50.00 0H.¢0 00.H 00.0 00.0 00.0 stA 00.00 00.00 00.00 00.V0 00.0v 00.00 00.00 00.0 4500 00.00 00.N0 00.00 H¢.Nv 00.00 00.00 H¢.0v HN.¢H 00.0 mmmm va.00 0N.HOH v0.H0 00.0w NH.00 00.00 00.00 H0.0 v0.0 00.0 Roam HUDA mud: Bud: AGHA meA ZUHA stA 0500 mmmm 85mm .mnwuomumso u05HMIHm3OHm .mamwuawuom Honnm mo xwuumz «mamaauqm 000000 .0 00009 I‘ll J‘I‘ 108 00.0 HUDA 00.00 00.0 mmmz 0H.00 00.0 00.0 3m¢2 0~.H0 00.00 00.00 00.0 AOHA 00.00 av.0h 0H.h0 00.0 00.0 HUHA H0.Nm 00.00 N0.N0 v0.0 0h.N 00.0 SUHA 0H.H0 b0.v0 00.00 00.H 00.N H0.N 00.0 3UHA 00.00 00.00 00.00 00.HN 00.00 00.00 00.00 00.0 £000 V0.0v 00.00 00.00 H0.0N 00.0N 00.00 N0.0N 00.0 00.0 mmmm v0.0v 00.00 00.v0 v0.HN 00.0w 0v.v~ 00.00 00.0 00.0 00.0 90mm HUDA mmaz Bums quq meq EOHQ szq 0000 mmmm 83mm .mumuomumno HmHon .mamwucmuom uouum mo xwuumz ummeHmcm 000000 .0 00.9% 109 Figure 18. HGROUP relationships between OTU's of onnia based on 108 character set; solid lines indicate closest related OTU. dotted lines indicate second closest OTU. 110 lll Figure 19. HGROUP relationships between OTU's based on vegetative characters of onnia. 112 Figure 20. 113 HGROUP relationships between OTU's based on flower-fruit characters of onnia. 114 46.4 115 I‘ll .Aaamn Hm3oHv mmcwmsuumm am and Amancwev mmoofiusum .A .Aaamn umosummmsv mmoHsEmnvm .m 00 Ummomeoo ma unmfiu Ham man no msoum mna .mcfluumsmwa .um> mucmmmummu Hana unnumflv muoe map can .muOHMflmoflHom .Hm> mo mmcmu may no mcofluuom cumummm 02m mHvUHE .cumumms ms» CH 0:500 mucmam 0n» ucwmmumwu maamn ammummc mmunu one .MCMHumsmfia .m Mano mmGSHo:« msoum mavwfifi 0:9 .Amaamn ucmumflv muoe 03» man an nmucmmmummuv mcmwnma xm 0cm Aum3mfl> may on ummmoHo damn 0:» ma vmucmmmummuv mmmmmw am mmcsaocfl puma man may no msoum one .m.DBo mo mmcflmsoum Momma mmunu may muoz A.m.Dao Umumawu hammoao umoE mmunu mufl can 080 comm cmw3umn mocmumfic 0:030:00 .mwmmamcm umumaao Eoum mama so 0mmmn .macowq mo m.090 macaw mmflnmcowumamu 0:0300m Hmwoa HMGOflmcmfiwcummuss .HN ousmflm 116 the (sec IESI 117 the second group (section Arsenococca), and E° lucida and g, mariana are loosely clustered in the third group (section Earls). Using just flower and fruit characters results in a much closer clustering of E. mariana with E, lucida, indicating the many similarities of these two taxa in reproductive structures. The grouping of l E. mariana and g. lucida is much less strong when only " vegetative characters are used. The three species, 1 ‘ £° fruticosa, £0 ferruginea and 2° squamulosa all seem ‘ to be equidistant from each other. gyonia squamu1osa and E, fruticosa seem to be slightly closer in flower- fruit characters, but E. ferruginea and E. squamulosa are closer when only vegetative characters are considered. When a.line is drawn at a error index level of 240 on the phenogram based on 128 characters, the genus is delimited into three groups, each corresponding to one of the sections. Six groups are present when a line is drawn at the 60 error index level. This lower error level indicates that much more homogenous groups are involved. These six groups correspond to each of the species in the genus native to North America. SeCtion Maria is clustered first with section Arsenococca When Vagetative characters are used, indicating the Vegetative similarities between these two groups. On the other hand, section Arsenococca is much more similar t o . o seat-1°11 Lyonia in floral morPhOJ-OQY: and clusters th se Va in la 118 first with this group when just flower-fruit characters are considered. Over all, section Arsenococca seems to be most closely related to section Maria. Lyonia ligustrina var. ligustrina is most closely clustered with the western part of the range of E. ligustrina var. foliosiflora, while both the middle and eastern parts of the distribution of var. foliosiflora cluster very closely. This indicates the similarities between the Arkansas populations of var. foliosiflora and the typical variety. It is sometimes hard to identify these Arkansas populations, especially late in the season, but they clearly belong with the southern variety, having large coriaceous leaves, rather lax inflorescences and reddish twigs. The western popu- lations of E. mariana also appear to be rather distinct. They have been recognized as forma vestita by Rehder (1924) and tend to be more pubescent than the eastern populations. The clustering of the 128 characters themselves also indicated a division of the genus into three major groups; Table 10 lists the characters in each group. Group I includes characters distinctive of section Elfiflfiav'groups II and III.indicate characters important In sections Arsenococca and Maria, respectively. This clUSterj-ng also indicated that all the characters 1 . nv°lV1n9 ferrugineous scales are highly correlated. 1Jl9 TABLE 10. HGROUP clustering of characters of Lygnia. GROUP I GROUP II GROUP III 7 Density of ferr. scales 18 Shape of bud 86 Shape of corolla at internodes, mature 10 Density of ferr. scales 20 Length of bud 89 Length of corolla at internodes, immature 11 Color of ferr. scales 22 L/w of bud 90 Width of corolla at internodes, mature 12 Ratio of small to large 29 Serration type 92 Length of corolla ferr. scales, mature from base to widest pt. 13 Shape of ferr. scales, 30 Serration type 93 Length of calyx lobes mature 14 Color of ferr. scales 34 No. of serrations/ 95 Width of calyx lobes at internodes, immature unit distance 15 Shape of ferr. scales, 35 No. of serrations/ 96 L/w of corolla immature unit distance 16 Persistence of small 36 Length between 100 Length of filaments scales serrations 24 Density of scales on bud 38 Length of teeth 103 Length of anthers 25 Color of scales on bud 54 Length of hairs on 105 Length of style lower leaf surface 55 Density of scales on 80 Inflorescence type 107 Shape of ovary upper leaf surface, mature 56 Density of scales on 81 No. of flowers/infl. 114 Width of capsule lower leaf surface, mature 57 Color of scales on lower 83 No. of panicles 122 No. of seeds/capsule leaf surface, mature 58 Ratio of small to large 84 No. of bracts/infl. 123 Length of seeds scales on lower 1f. surface 60 Density of scales on 85 Total length of bracts upper 1f. surface, immature 61 Shape of scales on leaf 121 No. of capsules/fruit mature cluster 64 Persistence of large scales 66 Size of scales 76 Density of scales on petiole 82 Relation of flowers 1:0 veg. growth 97 Pubescence of calyx 1 10 Pubescence of pedicle 11 7 Thickening of sutures 120 This probably indicates that all these characters are controlled by one complex of closely linked genes. Other characters such as length of corolla and length of fila- ments, or number of flowers per inflorescence and inflor- escence type seem to be logically correlated. The results of factor analyses based on vegetative characters and flower-fruit characters appear in Figures 22, 23 and 24. It is easily seen that this study also indicates a clear separation of the genus Lyonia into three groups. Figure 22 points out the distinctness of £3 mariana and L. lucida in vegetative morphology, while Figure 20 shows their similarities in flower and fruit characters. All three sections seem to be equally distant from each other, with L. lucida showing a slight tendency to group with section Lyonia (note Figure 22). Figure 23 and to a lesser extent Figure 22 illustrate the distinct- ness of E. ligustrina var. ligustrina from E. ligustrina var. foliosiflora. Principle factor III seems only to separate L. lucida from all the other species. Factor analysis of a loo-character set provided further evidence of the three strong subgroups within Lyonia. Characters highly correlated with Principal Factor I are important in distinguishing section Lyonia; Principal Factor II clearly represents section §£§227 ococca, while characters most closely associated with Principal Factor III are those important for section 121 Figure 22. Factor analysis: Principal Factors I and II based on vegetative characters of Lyonia. 122 II 11.0 . MARE ‘ O MARW 1 0.5 d '1 -o.s Q‘ LUCI 0.5 1.0 FERR ‘ 0.10 I I f T I I g SQUA .LIGL O FRUT LIGW . LIGM 4 LIGE d-0.5 . «-100 Figure 23. 123 Factor analysis: Principal Factors I and III based on vegetative characters of Lyonia. ~1.o FERR r .FRUT LUCI 124 ‘ 0.5 $1.0 LIGM .MARE LI LIGL Figure 24. 125 Factor analysis: Principal Factors I and II based on flower-fruit characters of Lyonia. 126 II FRUT ‘ 1.0 FERR ‘1 SQUA ‘ j 6.5 ., , . —0.5 0.0 .5 1.0 I . LIGM Q . LIGL .’ LIGE _ .LIGW LUCI ‘ ' O . --o.5 0m: 4 .4 4-1.o 127 $2533. (A list of the characters most highly correlated with the first four principal factors is given in Table 11.) Principal Factor IV seems to be closely correlated with characters that are important in distinguishing E. gaggif cosa from L. ferruginea. This points out one of the advantages of a numerical taxonomic investigation. Not only are major groups clearly indicated, but also char- acters important in delimiting these groups are indicated! For example, from the intercorrelation analysis of these 100 characters it is seen that the amount of curvature of the leaf blade is highly correlated with: (1) the length-width ratio of the leaves, (2) the depression of the veins above, (3) the growth form, (4) the amount of revolution of the leaf margins, (5) the type of leaf apex, (6) the time of flowering, and (7) the prominence of the major veins on the lower leaf surface. This combination of characters makes little sense until it is realized that it is precisely these characters which separate L. fruticosa from L. ferruginea. They are cor- related because they occur in definite combinations in these species. Lyonia ferruginea has extremely curved leaves with depressed veins vs. the flattened, reticulate- veined leaves of L. fruticosa. It is a large shrub or tree, while 2, fruticosa is a small shrub, often not over knee high. Lyonia ferruginea flowers in the early spring, but 2. fruticosa does not usually flower until TABLE 11. 128 Factor analysis of loo-character set. Charac- ters most highly correlated with first four principal factors. Principal Factor Correlated Characters II \lmmbw so... 0000 o 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 10. 11. Density of scales on twigs (mature) Density of scales on twigs (immature) Texture of bark Apex of bud Density of scales on bud Length of leaves Length from base of leaf to widest point Width of leaves Color of leaves Density of scales on upper leaf surface Density of scales on lower leaf surface Density of scales on petiole Relation of vegetative to floral buds Pubescence of corolla Pubescence of calyx Spur length Pubescence of pedicel Shape of capsule Pubescence of capsule Width of seeds Thickness of sutures of fruit Shape of bud Length of bud Length to width ratio of bud Serration type (distal) Serration type (proximal) Number of serrations per unit length (distal) Number of serrations per unit length (proximal) Length of pubescence Inflorescence type Number of flowers per inflor- escence Number of bracts per inflor- escence 129 TABLE 11. Continued Principal Factor Correlated Characters II (cont.) 12. Total length of bracts per inflorescence 13. Length of peduncle 14. Length of bracteoles 15. Width of bracteoles 16. Length of capsule 17. Width of capsule 18. Shape of corolla III 1. Density of nonglandular hairs on bud 2. Density of nonglandular hairs on veins of upper leaf surface 3. Length of hairs on veins of lower leaf surface 4. Presence of marginal vein 5. Density of nonglandular hairs on petiole 6. Color of corolla 7. Curvature of filaments 8. Thickness of blade 9. Longitudinal ridges on stem 10. Presence of revolute margin 11. Angle made by apex of leaf 12. Luster of leaves 13. Presence of glandular hairs on lower leaf surface 14. Position or growth form of branches IV 1. Curvature of leaf blade 2. Depression of veins on upper leaf surface 3. Habit (tree or shrub) 4. Type of leaf apex 5. Prominence of major veins on upper leaf surface 6. Prominence of major veins on lower leaf surface 7. Length of petiole 8. Length-width ratio of leaf 130 late spring or early summer. Many other distinctions are listed and are discussed in the taxonomic treatment. Thus the knowledge of one good distinguishing character can lead to the discovery of many more! Of course, the human mind is the best "computer" and perceptive botanists have long recognized the distinctive characters separating these two taxa. Phyletic Studies Twenty-four characters were selected and assigned generalized (primitive) and Specialized (advanced) states using the method of Wagner (1962, 1969). Table 12 lists these characters. Then each taxon was scored Q'if gen- eralized and l’if specialized for each of the characters. All the taxa are listed in Table 13 with the character divergence values for each. The total divergence index of each taxon was determined by adding the individual divergence values of each character together. Then mutual groupings of characters between taxa were deter- mined, and the taxa were arranged in sequence according to these groupings. They were plotted on a concentric graph (Figure 25), the radii being determined by the mutual groupings of characters and the distance being determined by the divergence of each taxon. Several instances of parallel derivation of Characters are evident. These include the deciduous V 131 TABLE 12. Characters used in phyletic study (Wagner method). Code Generalized Specialized Letter State State A Leaves evergreen Leaves deciduous Bl Flowers in racemes Flowers in corymbose fascicles BZ Flowers in racemes Flowers in panicles C Flowers white Flowers pink or red D Corolla cylindric Corolla strongly urceo- late E Calyx persistent Calyx deciduous F Plant not lepidote Plant lepidote G Small spurs or none on Long spurs on filament filament H Capsule ovoid or sub- Capsule urn-shaped globose I Sutures of capsule less Sutures of capsule thickened strongly thickened J Thickened sutures not Thickened sutures separ- separating from valves ating from valves as a unit K Tall shrub or small tree Small shrub L Floral buds not always Floral buds usually above vegetative buds always above vegetative buds M Leaves coriaceous Leaves thinner N Branches terete or Branches round angled O Calyx lobes broad and Calyx lobes elongate short 132 TABLE 12. Continued Code Generalized Specialized Letter State State P Ovary with massive p1a~ Placentae less massive-- T1 T2 centae--many small ovules Buds globose, usually small Buds with many outer scales Inflorescence with leafy bracts Filaments not pubescent Filaments not pubescent Filaments elongate Capsule almost com- pletely glabrous Lower leaf surface without dense covering of hairs Glandular hairs small or absent fewer ovules Buds elongate, larger Buds with only two outer scales Inflorescence lacking leafy bracts Filaments pubescent on lower half only Filaments pubescent for their entire length Filaments short Capsule pubescent, at least at base Lower leaf surface with a uniform covering of densely packed hairs Glandular hairs large, club-shaped 133 Character divergence values for each taxon, used to construct Wagner tree. TABLE 13. FRUT SQUA LUCI MARI LIGL LIGF FERR Character 1* 1010000000111011101* 1101.. 1* 1010000000111011111* 1101 11* 00101111111110010* 10000 01* 10001000010010010000000 01* 01010011000001010001100 01* 01010011100001010001110 01* 01010011000001010001110 * lad * 112 AnbnuCnuqununuIwuxaanmuonrnuRnamlTHUvaua 13 12 13 10 Total Character does not apply. Character R was not included in the calculations. ** 134 .mumum @wNflHMHommm ca mumuomumao mo mumuuma moon nufl3 aflcomq mo mucmmum>HU mo mecH .mm musmflm 135 «_zax_ mo wwfiumdm z_a mo xmaz~ Augguwundav Sudan-I Agasgnwflv 333333 .ug a .n_. Aguhmog BNQV A f f 35658. ”cue-2.3! m 8.5:: £652: a... ‘“ A383 .303 . ‘_- 55255 333.3 . .4 .4 $3.53“: 3: t3 ... \IIIII|I AEZBAB 383?: J / nanny-mu: 51> aqua-alw- .m Ill . . flu Hi t Qv 136 habit and round branches in Lyonia mariana and in a. ligustrina; and small urceolate flowers with short fila- ments and pubescent capsules in E. ligustrina and in the E. ferruginea group. Lyonia mariana and the E. ferruginea group show parallel development of the reduced racemose inflorescences and strongly thickened sutures of the cap- sule. Lyonia fruticosa and E. mariana have both evolved as dwarf rhizomatous shrubs of sandy soils. Lyonia lucida is the most primitive species studied; E. mariana and E. ligustrina are the most advanced. The E. ferruginea group (section Lyonia) is moderately advanced with E3 squamulosa occupying the most primitive position and E. fruticosa the most advanced. The results also strongly indicate that the genus is divided into three major parts, corresponding to the three sections recognized in the taxonomic treat- ment. Both E. mariana and E. lucida are grouped in section flagia in spite of their obvious vegetative dif- ferences. The evolutionary development of the characters in each section can easily be traced in Figure 25. The Wagner method requires numerous assumptions regarding which represents the primitive and which repre- sents the advanced state of each character. These decisions, although made after an extensive study of the genus and a thorough overview of related genera, are somewhat subjective and introduce a major source of error. 137 Thus a second method was used for comparison which does not require that any assumptions be made as to which state of each character is primitive and which is advanced. This method (Whiffin and Bierner, 1972) requires only that one taxon be chosen as representing the most primi- tive (ancestral) taxon. Twenty-three characters (essentially similar to those used in the Wagner method above) were used and arbi- trarily one state was designated as primitive (O) and the other as advanced (1). Then a table of designated char- acter states of each character for each taxon was con- structed. From this a matrix (Table 14) of differences was formed which records the number of characters for which each pair of taxa exhibit different states. By using this matrix and following the steps outlined by Whiffin and Bierner (1972) a phyletic tree can be con- structed. The resulting trees will be different if a different taxon is selected as the most primitive, although I found that in every case the major grouping of species was similar (Figure 26). These phyletic trees clearly indicate that the genus is composed of three distinct groups. Again these three groups correspond to the three sections of the genus present in North America. In every case E. lucida and E. mariana are clearly linked, with E, mariana as the more advanced, and E. lucida pro- viding a connection to the other members of the genus. 138 TABLE 14. Matrix of character divergence used in Whiffin- Bierner method. LUCI MARI FERR FRUT SQUA LI GL LIGF LUCI 2 3 MARI 15 23 FERR ll 7 23 FRUT lO 8 22 23 SQUA 12 8 20 19 23 LIGL 8 10 10 9 ll 23 LIGF 9 11 ll 10 12 22 23 Note: Numbers indicate number of characters in common. Symbols for taxa are the same as given in Table 3 except that MARI indicates E. mariana and LIGF indicates E. ligustrina var. foliosiflora. 139 Figure 26. Phyloqenetic trees of Lyonia produced by Whiffin—Bierner phyletic method. A. Tree assuming that E. lucida is most primitive. B. Tree assuming that E. squamulosa is most primitive. 140 LIGL LIGF MARI FRUT SQUA ? LUCI A FRUT FERR LIGF LUCI ’ SQUA 141 Lyonia ligustrina var. ligustrina is always more advanced than var. foliosiflora, which links this group to either section Maria or section Lyonia depending upon which species is taken as the most primitive. Lyonia ferruginea, E. fruticosa and E. squamulosa are always closely con— nected with E. squamulosa in the most primitive position unless either E. fruticosa or E. ferruginea is specifi- cally designated as primitive in constructing the network. A high degree of similarity is evident between the methods of Wagner (1962, 1969) and the "quick method" of Whiffin and Bierner (1972). Both indicate that Lyonia is divided into three clearly demarked groups. Both also indicate that E. mariana is more advanced than E. lucida and that E. ferruginea and E. fruticosa are probably more advanced than E. squamulosa. The relationships of the three sections to each other are not as clear. The Wagner method seems to indicate a closer tie between Arsenococca and Maria, while the modified method indicates that Arsenococca is closer to Lyonia. Comparison of the Phyletic and Phenetic Methods Both the phyletic and phenetic methods lead to the same conclusions. They both show that Lyonia is divided into three clearly demarked groups which I have recognized as the sections Lyonia, Arsenococca, and .Maria. It is interesting that these three subgroups were 142 very early recognized by botanists using only "classical" taxonomic methods. These studies also indicate that section Maria has the greatest amount of intrasectional variation, especially when only vegetative characters are considered. They also indicate which characters are important in differentiating each section from the others. A more detailed discussion of these characters can be found in the Systematic Treatment. Both studies indicate the distinctness of each species recognized in the taxonomic treatment. CHEMOSYSTEMATIC STUDIES With the advent of modern methods of analysis, comparative studies of the chemistry of plants have become much more widespread. Such progress has been par- ticularly valuable to the systematist because it has opened up a new dimension for study and has provided evidence confirming interpretations established by more traditional methods (Brehm, 1966). Sometimes the com- pounds themselves are identified and the pathways of their synthesis determined (Belzer and Ownbey, 1971); in other studies such as those of Alston and Turner (1959) or Furlow (1974), the chromatographic patterns are compared without determining the actual chemicals involved. The compounds are treated as ordinary taxo- nomic characters and, as with any taxonomic characters, mmltiple correlations, rather than single differences, are considered the most useful in determining relation- ships. Giannasi and Rogers (1970) list three reasons for using chemical data in taxonomic studies: (1) to test classifications based on morphology, (2) to determine species specific patterns, and (3) to detect possible 143 144 phyletic relationships. Just like any other taxonomic character, chemical data can give us insights into intra- specific variation and speciation (Turner, 1970). In this study phenolic chemical characters were used to provide a comparison with the classifications based upon morphology and to help clarify taxonomic relationships which could not be perceived from morphology alone. As pointed out by Bate-Smith (1963) and Heywood (1966) the results of chemosystematic studies must be used along with other kinds of data, not by themselves. Care must also be taken in evaluating the results! Numerical analysis of unidentified compounds can sometimes be mis- leading especially if the sample size is small, as is often the case in such studies (Weimarch, 1972). Methods and Materials Samples of leaves of five species of Lyonia were obtained from widely separated parts of the geographical range of each taxon (Table 15). These were collected from early to late summer from one- or two-year-old branches and were dried for approximately 96 hours in plant presses (natural drying). Two to four populations of each species were sampled. Within each population specimens were taken from five individuals. The indi- ‘Vidual samples from each population were combined. A 145 TABLE 15. Geographic location of specimens from which foliage samples for analysis of phenolic com- pounds were obtained. Collection . N ler Taxon Location 18, 19, 20, E. mariana Near Wading River, Bur- 21, 22 lington Co., N.J. 319 E. mariana Near Pritchardville, Beaufort Co., S.C. 277 E. mariana Near Morven, Brooks Co., Ga. 62, 65, 66, E. lucida Near New Bern, Craven 67, 70 Co., N.C. 100, 101, 102, ‘E. lucida Near Jamestown, Berkeley 103, 105 Co., S.C. 326 E. lucida Along 0.8. 82 near Liberty-Long Co. line, Long Co., Ga. 267 E. fruticosa North of Sparks, Cook Co., Ga. 314 L. fruticosa Amelia Island, Nassau — Co., Fla. 264 L. ferruginea Southwest of Tifton, — Tift Co., Ga. 300 E. ferruginea Jekyll Island, Glynn Co., Ga. 33, 34, 36, E. ligustrina North of Pocomoke City, 38, 40 var. ligustrina Worchester Co., Md. 239 E. ligustrina Near Highlands, Macon var.IIgustrina Co., N.C. 42, 43, 44, E. ligustrina Near South Mills, Camden 50, 52 var. foliosiflora Co., N.C. 252 L. ligustrina Northeast of Ashburn, var. fdlioSIflora Turner Co., Ga. 146 total of 15 samples was analyzed. Voucher specimens are depOsited in the Beal-Darlington Herbarium of Michigan State University (MSC). The phenolic compounds were extracted from dried leaves following, in general, the method of Hanover and Hoff (1966), Hanover and Wilkinson (1970) and Furlow (1974). Four-tenths of a gram of leaf material from each of the five collections at a particular site were combined into a single sample weighing 2.0 grams. Phenolic sub- stances were extracted by homogenizing the samples in a blender with 100 m1. of boiling water for 2 min., filter- ing, washing with 50 ml. of boiling water and then repeat- ing the entire process once more except with two 50 m1. boiling water washes. This method was found by Hanover and Hoff (1966) to effectively eliminate such interfering substances as tannins and chlorophyll from the extract. The water extract was washed five times with 50 ml. portions of ethyl ether in a separatory funnel to remove waxes and fats, followed by 50 ml. portions of normal butyl alcohol. Compounds were separated by two-dimensional descending paper chromatography. Fifty microliters of extract were applied with a micropipet to a point on the upper surface of 46 by 57 cm. Whatman 3MM chromato- graphic filter paper 8 cm. from each edge in the upper left-hand corner. The papers were folded on a line 147 5.5 cm. from the upper edge and placed, in groups of eight, in a chromatographic chamber, which was equil- ibrated for 2 hrs. with the lower portion of a mixture of n-butanol, acetic acid, and water (4:1:5). The sheets were then irrigated with the upper portion of the solvent and allowed to develop until the solvent front neared the bottom of the sheets (about 16 hrs.). After drying, the chromatograms were irrigated in the second direction with a mixture of acetic acid and water (3:17), and allowed to develop until the solvent again had nearly reached the bottom of the sheets (about 5 hrs.). They were then thoroughly dried at room temperature. Completed chromatograms were examined without chemical treatment and in daylight and under both long- and short-wave ultraviolet illumination after a 30 min. exposure to fumes of ammonium hydroxide. Spots were out- lined, their R values calculated, and their color f reactions to each treatment noted (Table 16). Spots produced by the same compound on various sheets were identified, numbered, and tallied, but no attempt was made to determine the chemical identity. Results and Discussion Seventy-four compounds were found in the butanol fractions of the five species studied; of these, 8 (numbers 1, 2, 12, 17, 18, 21, 22, and 28) occurred on every sheet and four (numbers 3, 16, 20, and 23) were TABLE 1 6 . 148 Rf values and color reactions of the phenolics in L onia. Color abbreviations: L = light; B1 = blue; R = red; G = green; Y = yellow; P = purple; Br = brown; V = violet; O = orange. Rf n-Butanol, Rf Acetic Com— acetic acid, acid, Ordinary UV Light pggnd water water Light Untreated NH (4:1:5) (3:17) 3 1. 0.00 0.06 0 LBr BrR 2. 0.88 0.00 LY V V 3. 0.37 0.06 —— LY Y 4. 0.33 0.12 -- LB Y 5. 0.42 0.14 -- P B1P 6. 0.26 0.08 -- P LG 7. 0.28 0.14 -- LP P 8. 0.64 0.13 -- P PBl 9. 0.56 0.01 -- LBl LP 10. 0.44 0.30 -- LBr Br 11. 0.67 0.39 Y BrR CY 12. 0.55 0.44 Y BrR Y 13. 0.35 0.46 LY -- LPR 14. 0.30 0.43 -- -- L0 15. 0.23 0.44 -- -- L0 16. 0.31 0.56 -- LBrR LR 17. 0.47 0.58 LY BrR Y 18. 0.67 0.58 YO BrR BrRO 19. 0.52 0.57 LO Br BrR 20. 0.77 0.63 -- P Bl 149 TABLE 16. Continued Rf n-Butanol, Rf Acetic Com- acetic acid, acid, Ordinary UV Light pggnd water water Light Untreated NH (4:1:5) (3:17) 3 21. 0.55 0.73 -- P 81G 22. 0.60 0.86 -- P PBl 23. 0.49 0.85 -- P P 24. 0.51 0.81 -- LGBl G 25. 0.35 0.82 -- -- G 26. 0.34 0.82 -- -- PBl 28. 0.98 0.90 LY LG LP 29. 0.56 0.26 -- -- G 30. 0.19 0.29 -- -- LB 31. 0.16 0.39 -- -- L0 32. 0.51 0.15 -- -- LB 33. 0.58 0.33 LY LBr Y 34. 0.54 0.50 YO Br CY 35. 0.67 0.67 -- Bl G 36. 0.64 0.64 LY RBr O 37. 0.40 0.68 -- P PBl 38. 0.38 0.78 LY LOBr Br 39. 0.29 0.79 -- LBrP LBrP 40. 0.62 0.59 -- -- Y 41. 0.24 0.19 -- -- G 42. 0.27 0.07 -- -- CY 43. 0.56 0.09 -- -- P 150 TABLE 16. Continued Rf n-Butanol, Rf Acetic Com- acetic acid, acid, Ordinary UV Light P§:?d water water Light Untreated NH (4:1:5) (3:17) 3 44. 0.21 0.44 -- -- LY 45. 0.36 0.32 -- LBr LY 46. 0.30 0.62 -- -- LY 47. 0.30 0.50 -- LBr LR 48. 0.41 0.42 -- GBl GBl 49. 0.46 0.38 -- -- L0 50. 0.80 0.70 -- G B1 51. 0.61 0.79 -- RBr PBl 52. 0.75 0.88 -- B1 B1 53. 0.59 0.90 LY RBr RBr 55. 0.81 0.51 -- B1 B1 56. 0.43 0.79 -- BlG G 58. 0.43 0.71 -- -- B1 59. 0.76 0.81 -- LBlY LBl 60. 0.75 0.42 -- LBl LBl 61. 0.98 0.44 -- L81 L31 62. 0.70 0.90 LY RBr PBl 63. 0.55 0.66 -- P BlP 64. 0.51 0.90 -- LY YG 65. 0.66 0.71 -- LBl -- 66. 0.48 0.76 LY O O 67. 0.40 0.85 LY LO 0 151 TABLE 16. Continued Rf n-Butanol, Rf Acetic Com- acetic acid, acid, Ordinary UV Light p§:?d water water Light Untreated NH (4:1:5) (3:17) 3 68. 0.48 0.76 -- L81 B1 69. 0.67 0.76 -- LBl B1 70. 0.67 0.85 -- LP P 71. 0.57 0.64 -- -- G 75. 0.53 0.68 LY Br RBr 76. 0.49 0.86 -— Br Br 77. 0.65 0.81 -- LBl -- 78. 0.63 0.90 -- LBr LBr 79. 0.59 0.89 -- LBr LBr 152 present in every species but not on every sheet. The number of spots on individual sheets ranged from 18 to 27; from 25 to 35 compounds were found in the species studied (see Table 17). At least one unique compound appeared in every taxon (Table 18). A representative chromatogram for each species is given in Figures 27 through 32. Chemical similarities among the taxa were calcu- lated using the paired affinity index of Ellison, Alston and Turner (1962); the resulting values are shown in Table 20 (see also Table 19). The results of this study indicate that each taxon has a definite pattern of phenolic compounds and can be recognized readily on this basis. Some taxa such as E. mariana and E. lucida seem to be uniform chemically. The population of E. mariana sampled in Burlington Co., N.J. is extremely similar to the population sampled in Beaufort Co., S.C.! Similarly, the population of E. lucida sampled in Craven Co., N.C. is quite similar to the population in Long Co., Ga. Thus the chemical characters seem to compare well with the morphological characters which are also uniform in these two species. On the other hand, E. ligustrina, a very polymorphic species, had the most different phenolic compounds of the Species studied, and showed the greatest intra- specific variation. The northern E. ligustrina var. 153 TABLE 17. Total listing of phenolic compounds present in each population of each species studied. —— -::=_:9 may? Arm—1:."— Spot MARI MARI MARI FRUT FRUT FERR FERR LUCI LUCI LUCI LIGF LIGF LIGL LIGL No. 18 319 277 314 267 300 264 326 62 101 252 42 33 239 ><>< XXXXX XXXXXX XXX XXX XXX XXX >< XXX xxx >< XXXX X X xxx xxx X X X ...; oomqamnwww Ht—o NH O xxx xx XXX >< XX X >< >< >< ><>< >< XX 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 28. X 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. 45. 46. 47. X 48. 49. 50. 51. 52. 53. 55. 56. 58. 59. 60. 61. 62. 63. X X 64. 65. 66. 67. 68. 69. 70. 71. 75. 76. 77. 78. 79. XXX xxx xxx xxxx XX XX XXXXX ><><><><><><><>< XXX XXX xxxxxxxx ><><><>< XXXXXXXXXXXXXXX XXX ><><><><§< XXX XXXXXXXXXX xxxxxxxxx XX ><><><>< ><><><><><><><><><><><>< XXXXXX ><><><><>< ><><><>< X ><>< XX XX XX X X >< XXX X ><><><><>< XXX XXX XXX ><><><>< X XX X ><><>< X XX ><><>< X XX X ><><><$<>< >< ><>< XXX XXXXXX >< XXX ><><><><><><><><>< xxxxxxxx xxxxxxxx XXXXXXXXXXXXX ><>< XX Note: Numbers skipped were rejected as not representing distinct compounds, in most cases they were included with another spot. 154 TABLE 18. Characteristic phenolic compounds for each Species of Lyonia. Compound Number Frequency (in percent of sheets) I. E. mariana 5 100 7 33 8 66 14 66 25 33 26 33 63 66 II. E. lucida 64 100 65 100 66 A 100 67 100 68 100 69 100 70 100 71 100 75 50 76 50 77 50 78 100 79 100 III. E. fruticosa 49 50 IV. E. ferruginea 58 50 59 50 60 50 61 50 62 100 V. E. ligustrina 30 75 31 25 32 25 37 75 38 25 39 50 40 25 41 25 42 25 43 25 44 25 45 25 46 25 155 .mmoofiusum macomq mo Emumoumeouso m>flumucmmmummm .nm mummflm 156 l-BUIAKL ACHIC ACID, HATER (4:1:5) A ' .7” —; mncmn.mm (3:17) 157 .mmcflmsuumm choNq mo Emumopmeouno m>Humucmmmuomm .mm ousmflm 158 I-BUMNOL, ACETIC ACID. HATER (4:1:5) j: 3 v N" #: ACETIC ACID. HTER (3.17) 159 .mcflnuwsmfla .um> mcfluumsmfla macomq mo Emumoumfiouco m>wunucmmmummm .mm musmflm 160 mm 34.3 5:: 53¢ :53 £153.: Inca N'ETK AUE, IATEP (3.175 A 161 .muoHMAmOMHom .um> mcfluumsmfia macomq mo Emumoumfiouso m>flumusmmmumom F .om wusmflm 162 I 8.. :3 E i 55 3 S i i A KEIIC ACID. UTE! (3:17) 163 .mowosa MHCQNM mo Emumoumeouno m>flumucmmmumwm .Hm mesons 164 C HATER (4:1:5) ACETIC ACID. l-BUTANOL, \ % ACETIC ACID. HATER (3:17) ' 165 .mcmwuna second mo EmumoumEonso m>wumucmmmummm .Nm musoflm 166 33.: a; .23 ...—E g ACETIC ACID. HATER (3:17) ' 167 TABLE 19. Number of compounds in common among the taxa of Lyonia studied. I MARI FRUT FERR LUCI LIGF LIGL L_fi MARI 28 FRUT 15 29 FERR 12 25 29 LUCI 16 15 13 31 LIGF 16 17 16 14 30 LIGL 16 18 16 15 20 25 168 TABLE 20. Chemical similarity between species of L onia in North America. (Numbers represent pEIEEd— affinity values.) MARI FRUT FERR LUCI LI GF LI GL MARI 100 FRUT 35.7 100 FERR 26.7 75.8 100 LUCI 37.1 33.3 27.8 100 LIGF 38.1 40.5 37.2 29.8 100 LIGL 43.2 50.0 44.4 36.6 57.1 100 169 ligustrina is more homogenous in phenolic compounds than the southern var. foliosiflora. Again the phenolics seem to accurately reflect the total variability of each taxon, in that the southern variety is the more diverse morpho- logically. The two widely separated populations of E. ligustrina var. ligustrina both have very similar chemical profiles, while a greater difference is shown between the geographically closer populations sampled of var. Egiigr siflora. Definite differences in chromatographic patterns occur between E. ligustrina var. ligustrina and E. Eigggf trina var. foliosiflora. The phenolic compounds separating these two varieties are listed in Table 21 below. Lyonia and E. ferruginea are both moderately variable and quite distinct in their phenolic compounds. Lyonia fruticosa possesses the following compounds not found in E. ferruginea: ll (50% of sheets of L. fruti- cosa), 24 (100%), 49 (50%), 51 (100%), 58 (50%), 59 (50%), 60 (50%), 61 (50%), and 62 (100%). Thus the evidence from phenolic compounds further supports the recognition of 'Uhese two taxa as distinct species. The p0pulation of E. fruticosa from.Nassau Co., Fla. which showed some morphological similarity to E. ferruginea is much more similar to E. fruticosa in its phenolic pattern than it is to»E. ferruginea, and it does not appear to be a hybrid between these two species. 170 TABLE 21. Compounds separating L. ligustrina var. ligus- trina from E. ligustrIna var. foIIOSiflora. Frequency (in percent Compound Number of sheets) I. E. ligustrina var. ligustrina f 9 100 24 50 _ 31 50 32 50 38 50 II. L. ligustrina var. foliosiflora 4 100 6 100 39 100 40 50 41 50 42 50 43 50 44 50 45 50 46 50 171 From Tables 18, 19 and 20 it is seen that E. fruticosa and E. ferruginea are probably very closely related and quite distant from the other taxa studied. The most closely related from outside section Lyonia seems to be E. ligustrina. Eyonia ferruginea and E. fru- ticosa show the following group of compounds which are absent from all other taxa. TABLE 22. Diagnostic compounds for section Lyonia. Frequency (in percent Number of Compound of sheets) 52 100 53 100 56 100 35 75 47 75 48 50 50 50 55 50 62 50 49 25 58 25 59 25 60 25 61 25 .A close relationship is also indicated for E. ligustrina ‘var. ligustria and E. ligustrina var. foliosiflora. These tw0>taxa have a number of compounds in common and form a <1istinct group separate from E. mariana, E. lucida or the IL. ferruginea group. Lyonia lucida does not seem to be closely related to any other taxon. Its closest ties seem to be with E. mariana in that they both possess 172 compounds 10 and 15 which occur in no other taxon. Lyonia mariana seems to show relationships to both E. lucida and E. ligustrina. It shares compound 19 with E. ligustrina. The above information is summarized in the phylogenetic tree showing the compounds held in common by each succes- sive group of taxa (Figure 33). This tree gives a clear total picture of the possible interrelationships within the genus as a whole and seems to support the division into the same three sections as recognized on the basis of morphological characters alone., The only difference is that the chemical data seem to indicate that section Arsenococca is slightly closer to section Lyonia whereas the total morphological classification puts it closer to section EEEEE. In floral morphology section Arsenococca and section Lyonia do show several similarities (see also Figure 34). The variability of phenolic compounds in higher plants has been discussed by Alston (1966), Alston and Turner (1963), Hanover and Hoff (1966), Taylor (1971) and others, who point out that chemical composition can vary with season, age, tissue, growth site, individual, etc. These factors were controlled as much as possible in this study. Fresh material of E. lucida (no. 326) was chromatographed to compare with the phenolic pattern obtained from dried material from the same population. 173 Figure 33. Phylogenetic tree showing the phenolic compounds distinctive for each taxon of Lyonia. Figure 34. 175 Phylogenetic trees based on chemical data (Table 20), using Whiffin-Bierner method. a. Tree with Lyonia ferruginea taken as the most primitive. b. Tree with E. lucida taken as the most primitive. c. Tree with E. fruticosa taken as the most primitive. 176 MARI 1|) LUCI LIGL FRUT LIGF * 9 (JD FERR A FERR FRUT LIGL T‘P MARI LIGF d' LUCI B MARI LUCI q. FERR LIGL LIGF .1 \ FRUT 177 Although some differences were noted, the resulting patterns were quite similar. Thus extracts prepared from dried material provide a reasonable indication of the phenolics present in the living plant. A 1. i TAXONOMIC TREATMENT Lyonia Nuttall Lyonia Nutt. Gen. N. Am. Pl. 1: 266. 1818, nom. cons., non Eyonia Elliott. Sketch Bot. S.C. 1: 316. 1817 (= Cynanchum L., Asclepiadaceae) nec Lyonia Rafinesque. Med. Repos. N.Y. ser 2, 5: 353. 1803 (= Polygonella Michx., Polygonaceae). Xolisma Raf. Amer. Monthly Mag. & Crit. Rev. 4: 193. 1819. Neopieris Britton and Brown. Ill. Fl. ed 2, 2: 690. 1913. Desmothamnus Small, Shrubs Fla. 96. 1913. Arsenococcus Small E3 Small and Carter. F1. Lan- caster Co. 218. 1913. Evergreen or deciduous shrubs, rarely small trees, 'usually from an underground rhizome, with round to terete or angled branches. Leaves alternate, short-petioled, entire or shallowly toothed or serrulate, nearly glabrous, or vwith hairs or peltate scales; winter buds ovoid, with two outer scales. Inflorescence axillary (sometimes appearing terminal but falsely so), corymbose, racemose or paniculate, the racemes sometimes contracted into 178 179 axillary fascicles; each flower in the axil of a small bract and with two lateral bracteoles at the base of the pedicel, these sometimes quickly deciduous. Calyx 5-, rarely 4- to 8-lobed, the lobes valvate or reduplicate in bud, persistent or deciduous in fruit. Corolla cylindric-campanulate to urceolate or globose-urceolate, 1 with 5 (4 - 8) short lobes, white to red, glabrous to pilose or lepidote. Stamens 10, rarely 8 - 16; filaments flattened, S-shaped, expanded near base, glabrous to hairy or roughened, with or without a pair of short spurlike appendages in the back near the apex; anthers obtuse, the lobes parallel, lacking apical awns, dehiscent by large terminal pores, always with a white line of disintegration tissue on the back of each lobe extending at least along the apex of the filament and along the upper edge of the appendages when present; pollen tricolporate, surface psilate, indistinctly tectate, in tetrads without viscin strands. Stigma truncate to capitate; style columnar to fusiform, straight, not exerted; ovary superior, five- locular, the placentae large, undivided; disc an enlarge- ment of the ovary wall, variously develOped. Capsule subglobose to ovoid, five-angled, loculicidal, with paler, thickened sutures which may separate from the five valves in dehiscence; placentae persistent at the top of the columella; seeds scobiform, with a loose thin testa. 180 Asia (Kashmir to Japan, south to Malaya), Greater Antilles, and North America (eastern United States and Mexico). Lectotype species: Lyonia ferruginea (Walt.) Nutt.: see I.C.B.N. 261. 1956, and Rickett and Stafleu, Taxon 9: 75. 1960. Key to Sections, Species and Varieties ofIEyonia A. Lower surface of leaves, pedicels, and calyx lepidote (Section Lyonia) . . . . . . . . . . . . B. B. Capsule subglobose, 2.5-4.0 mm. long; leaves only slightly pubescent or glabrous beneath. . . . . . . . . . . . l. E. squamulosa B. Capsule ovoid, oblong-ovoid or oval, 3.0- 6.0 mm. long; leaves usually moderately pubescent beneath . . . . . . . . . . C. C. Leaves conspicuously reduced toward the ends of the branches, major veins not depressed, margin not revolute or only slightly so; ultimate branchlets rigidly ascending . . . 2. E. fruticosa C. Leaves not conspicuously reduced toward the branches, major veins usually depressed, margin usually revolute; ultimate branchlets not rigidly ascending. . . . . . 3. E. ferruginea A 181 A. Lower surface of leaves, pedicels, and calyx glabrous or pubescent but not lepidote . . . . . . . . D. D. Inflorescence paniculate; corolla urceolate, 4 mm. long or less; calyx lobes less than 2 mm. long; leaves pilose (Section Arsenococca, E. ligustrina the only species) . . . . . . . E. E. Inflorescence naked or with only a few leafy bracts (chiefly mountains and Pied- mont south of Va. and mountains, Piedmont and Coastal Plain north of Va.). . . . . . . . . . . . . .4a. var. ligustrina E. Inflorescence conspicuously leafy bracted or at least the lower panicles with large bracts (chiefly Coastal Plain from SE Va. to W to Tex. and Ark.) .4b. var. foliosiflora Inflorescence corymbose (fasciculate); corolla elongate, longer than 4 mm; calyx lobes 2 mm or longer; leaves glabrous or with slight to moderate pubescence on veins, sometimes glandu- lar dotted (Section EEEEE) . . . . . . . F. F. Leaves evergreen, rigidly coriaceous, with intramarginal vein; corolla less than 5 mm wide, usually pink; flowers in leaf axils . . . . . . 5. L. lucida 182 F. Leaves thinner, deciduous, lacking intra- marginal vein; corolla more than 5 mm wide, usually white; flowers borne on leafless branches . . 6. E. mariana Specimens Examined In the citation of herbarium specimens, abbrevi- ations of institutions follow the fifth edition of Index Herbariorum (Lanjouw and Stafleu, 1964) and Taxon 15: 335, 1966. The herbaria and their abbreviations are: Arnold Arboretum (A); Duke University (DUKE); Escuela Nacional de Ciencias Biolégicas, Mexico (ENCB); Florida State University (FSU); University of Georgia (GA); Gray Herbarium (GH); Mississippi State College (MISSA); Uni- versity of Michigan (MICH); Missouri Botanical Garden (MO); Michigan State University (MSC); University of North Carolina, Chapel Hill (NCU); New York Botanical Garden (NY); University of Texas (TEX); National Museum, Smithsonian Institution (US); Vanderbilt University (VDB); University of Wisconsin (WIS). The specimens from which data were obtained for the numerical taxonomic study are indicated by an asterisk (*). Ordinarily one specimen per county has been cited for species in the United States. A: I | 183 Lyonia Nutt. sect. Lyonia Andromeda L. sect. Lyonia Gray, Syn. F1. N. Am, 2(1): 32. 1878, in part. Xolisma Raf. sect. Lyonia Rehd., J. Arnold Arb. 5: 54. 1924. Lyonia Nutt. sect. Eulyonia K. Koch, Dendr. 2(1): 119. 1872, in part. Evergreen shrubs or small trees with terete or angled branches. Leaves entire or shallowly toothed and lepidote. Flowers in dense axillary fascicles; corolla urceolate; filaments unappendaged or with only very small spurs. Capsule elpidote with pale much thickened sutures which separate as a unit from the five valves in dehiscence. Coastal Plain from Florida to South Carolina, montane areas of eastern and southern Mexico, and the Greater Antilles. l. Lyonia squamulosa Mart. & Gal. Lyonia sqgamulosa Mart. & Gal., Bull. Acad. Brux. 9: 542. 1842. Xolisma squamulosa (Mart. & Gal.) Small, F1. N. Am. 29(1): 66. 1914. Type: MEXICO: "NE. du coffre du Perote, surout pres de la Venta del Esquilon, a 4,500 pieds de hauteur absolute." E. Galeotti 1797 (BR?, not seen). A widely branched shrub usually under 3 m (occasionally to 5 m). Young branches covered with ndJnIte ferrugineous scales, becoming glabrous or with 184 only sparse covering of scales with greater age, also often with short nonglandular hairs. Buds lepidote, ovoid, 1.0-2.0 mm long. Leaves evergreen, moderately coriaceous, alternate, short petioled, entire or rarely shallowly toothed, sometimes with revolute margins, elliptic, oval, ovate, or obovate-elliptic, 1.0-6.0 cm long, 0.5-3.0 cm wide, with tips acute, acuminate or rounded and bases attenuate or cuneate, adaxial surface lusterous, glabrous to slightly lepidote, with slight pubescence on mid-vein, abaxial surface densely lepidote, rarely with nonglandular hairs intermixed. Flowers in axillary corymbose fascicles; pedicels lepidote, each with a pair of bracteoles at the base. Calyx S-lobed, persistent, lobes triangular, 1.0- 2.0 mm long, 0.5-2.0 mm wide, abaxial surface lepidote and sometimes also nonglandular pubescent, adaxial sur- face sparsely to moderately nonglandular pubescent. Corolla urceolate, with 5 short lobes, white, lepidote, 2.5-4.0 mm long, 2.5-3.5 mm wide. Stamens 10, inserted at the base of the corolla; filaments flattened, expanded :near the base, S-shaped, roughened, unappendaged or with 'very short spur-like appendages on the back near the apex; anthers obtuse, the lobes parallel, lacking apical awns, dehiscent by large terminal pores, always with a white line «of disintegration tissue in the back of each lobe extend- ing at least along the apex of the filament and into the short appendages if present. Style columnar or fusiform, 185 not exerted; ovary superior, 5-1ocular, tomentose and lepidote. Capsule subglobose, S-angled, 2.5-4.0 mm long, 2.5-4.0 mm wide, loculicidal, with paler thickened sutures which separate as a unit from the five valves in dehis- cence, lepidote and often nonglandular pubescent, espe- cially at the base. Seeds ca 80 per capsule, very narrow, averaging 1.5 mm long (Figure 35). Distribution: Mexico, in temperate montane regions along the eastern Sierra Madre from Nuevo Leon southeast to Veracruz, Puebla, and Oaxaca, Chiapas and Guerrero, mostly between 1000 and 2400 meters (Figure 36). Flowering: throughout the year but especially March-June. Representative specimens: MEXICO. Chiapas. 4 miles north of Jitotol on rd. to Pueblo Nuevo Solista- huacan, municipio of Jitotol, 5500 ft, Breedlove £222 (ENCB, US); Lago de Monte Bello, 25 miles east of La inrinitaria, Municipio of La Trinitatia, 5200 ft, Breedlove 714959 (FSU*); near the Lago de Monte Bello, southeast of Comitan, Miranda 2__6_§_4_ (US); 5 miles south of Pueblo Nuevo Solistahuacan along rd. to Jitotol, Municipio of Jitotol, 5300 ft, 39133 & Breedlove E2233 (ENCB*); 3 kilometers northwest of Pueblo Nuevo Solistahuacan, Municipio of Pueblo Nuevo, 5400 ft, Zuill 126 (ENCB). Guerrero. 10 kilometers southwest of Amojileca, Municipio of Chilpancingo, 2050 meters, Rzedowski 23567 (ENCB, use”. v Figure 35. 186 Representative specimen of Lyonia squamu- losa (Ventura 9205, ENCB). 187 Fan. 10c I‘Jtodc ' nt HMMMIU us LA -‘UIIA MACDOIAL as M“ ”MARIA! [WI W‘NK‘O IAOIOIAL “UFO. I) ' PLANT“ WAS rochn AH 188 .mmoHsEdmwm nanomm mo coausnwuumwo . mm 85m: 189 no. 94. 30.55.3404 (a: 4(30- n .288 g.0>30.v >00. 000. «...-l... I.- in IL I a CI '1 . D n.4x ‘ ' x f , I ; I .4 o n . I . . . r ~I I, I \ a a . I . \ r ‘ I I I f I a O. In. Cu. .... .‘fl 190 Hidalgo. Jacala, Kenoyer 312'! 16 November 1937 (M0); El Estribo, Carretera Tulancingo-Tenango, 44 kilometers, 2000 meters, Leyva 536 (ENCB, TEX*); Tenango de Doria, 2100 meters, Puig 2100 (ENCB); alrededores de Zacualtipan, Quintero 808 (ENCB MSC*); 6 kilometers southeast of Zacualtipan, 2150 meters, Rzedowski & Madrigal 29457 »(ENCB*); Cerro de las Cruces, Aqua Blanca, Vela 586 (ENCB*). Nuevo Leon. Cerro del Viejo, 15 miles west of Dulces Nombres, Municipo of Zaragoza, Meyers & Rogers £323! 20 August 1948 (M0); Dulces Nombres, 24N, 99, 5-100.5° W, Meyers & Rogers EZEE (MO); Potrero Redondo west to Puerto. a Laguna Sanchez, municipio of Villa Santiago, Mueller .EEEZ (TEX*). Oaxaca. Cumbre de Huehuetlan near Teotitlan, Conzatti 4111 (US); without definite locality, 6000 ft, Galeotti 1849 (US); near Cerro Zempoaltepetl, southwest slopes of mountain near Tlahuitoltepec, 2400 meters, Hallberg 953 (ENCB*, US); Patio de Arena, ca 9500 ft, Ihigh point of trail from Zacatepec to Tamazulapan, :Leonard 3931 (NCU*); Villa Alta, Eg£§y_32§ (ENCB). :Puebla. Serras Decoration archeveché, near Puebla, Arsene 1221!: (MO, US); Chinantla, Liebmann 8629 (MSC*); £1350 meters, Municipio of Pahuatlan, Egig_§22§ (ENCB); near Honey Station, 5800 ft, Pringle E993 (US); Zaca- poaxtla, Salazar _s_.3_1_., 15 April 1913 (US). Egg Eu_i_s_ Potosi. 4 kilometers northeast of Guadalcazar, 1800 meters, Dias 3739 (ENCB); Sierra de San Miguelito, near 191 the Cueva del Mezquite, 2200 meters, Rzedowski 5451 (ENCB*); Cerro Grande, ca 6 kilometers northwest of Guadalcazar, 2000 meters, Rzedowski 5956 (ENCB); Mesita de Gallos, ca 20 kilometers southeast of Zaragoza, 2050 meters, Rzedowski 6100 (ENCB*). Tamaulipas. highest part of the Sierra de Tamaulipas, 900-1500 meters, Rd. from Rancho Las Yucas to Santa Maria de los Nogales, Martinez & Luyando F-l917 (TEX*). Veracruz. Santa Ana, Xalacingo, 2150 meters, Vela & Martinez 1189 (ENCB); Agua Cruz, municipio of Jalacingo, 1650 meters, Ventura EQEQ (ENCB); Paso Enriquez, municipio of Yecuautla, 1600 meters, Ventura EZEZ (ENCB*, WIS); Esquilon, municipio of Jilotepec, 1310 meters, Ventura 3331 (ENCB); La Florida, municipio of Atzalan, 1750 meters, Ventura §§§§_(ENCB); Tezchiquiapa, municipio de Jalacingo, Ventura 222§_(ENCB). Martens and Galeotti (1842) reported both E. squamulosa and E. ferruginea from Mexico. Sargent (1893) recognized only E. ferruginea as occurring in Mexico, while Standley (1924) listed only E. squamulosa, but stated: "The Mexican plant differs only in its slightly shorter capsules and it is doubtful whether itrepresents a distinct species." Small (1914) included all Mexican plants in E. gquamulosa, separating this taxon from E. fruticosa and E. ferruginea by its subglobose capsules and the width of the calyx. Rehder (1924) implied that several members 192 of the genus occur in Mexico, and Wood (1961) stated that E. ferruginea occurs in Mexico about the eastern Sierra Madre from San Luis Potosi south to Oaxaca, but leaves the question somewhat open by saying that "one or two" species occur in "eastern and southern Mexico." Lyonia §quamulosa is a variable species, thus Martens and Galeotti were only describing two extremes from the total pattern of variation. This variation is interesting in that it parallels that between E. fruticosa and E. ferruginea on the Coastal Plain in the southeastern United States. Some plants tend to have large elliptic leaves with a dense covering of dark ferrugineous scales on the abaxial surface, while others are small shrubs with ascending branches bearing small obovate leaves with acute or blunt tips and light gold or whitish scales! The former are superficially like 3E. ferruginea; the type specimen may be of this group. .Martens and Galeotti state that it blooms in May and compare it with the West Indian E. jgmaicensis. The latter taxon shows some similarities to E. fruticosa. Dhartens and Galeotti may have called such plants E. ferruginea, especially if they were following the nomen- <:1ature of Nuttall (1818). The flowering time is given .as June. I found it impossible to satisfactorily separate two groups of Mexican plants as there is a continuous range of variation between them and in many respects the 193 characters seemed to show a reticulate pattern within the taxon as a whole. Lyonia squamulosa is clearly separated from E. fruticosa by the glabrous and densely lepidote abaxial leaf surface, its subglobose angular capsules, its larger size, the more elliptic leaves, and its usual lack of very reduced leaves along the upper portions of the branches. It can be differentiated from E. ferruginea by the general lack of nonglandular pubescence on the abaxial leaf surface, the unrevolute or only slightly revolute leaf margins, the lack of noticeable depression of the major veins of the leaf, and by its subglobose angular capsules. Lyonia squamulosa also has a tendency to have darker orange or brown peltate scales, and more elliptic and ovate leaves. Lumping this taxon with either E. ferruginea or E. fruticosa would completely wreck the consistency of the characters upon which these taxa are based. A specimen collected by Rzedowski (no. EEEQ) in San Luis Potosi is very unusual in that it has a fine nonglandular pubescence on the lower leaf surfaces. Extremely revolute leaves are present on Rzedowski 5451, also collected in San Luis Potosi. A specimen from Veracruz (Ventura 5655) showing small obovate leaves with rounded tips is also unusual. More collecting is needed to determine if these characters are limited to these areas. There is still insufficient material to 194 reach a complete understanding of the variation of this taxon but it seems to be quite polymorphic, showing relationships with both E. fruticosa and E. ferruginea. Possibly this variability is due to its widely scattered distribution in temperate montane areas. The vegetation of the montane areas on the eastern escarpment of the Anahuac Plateau or Central Mesa (begin- ning about Xilitla, San Luis Potosi and extending through Hildago, Puebla, and part of Veracruz) and certain high- lands in Oaxaca and Chiapas were discussed by Miranda and Sharp (1950). Lyonia squamulosa is listed as occurring in the understory of the mixed oak forest with Gaultheria hirtiflora and Heberdenia penduliflora. Important trees are Quercus, Liquidambar, Symplocos, Prunus and Phoebe. It is also listed as an understory component of the pine forest where it occurs with the following shrubs: 922: ostegia icosandra, Zinowiewia integerrima, Ascyrum hyper- icoides, Vernonia patens, and Eyrica pringlei; trees in the overstory include Eiggg and Quercus. Lyonia squamulosa is also important in disturbed communities which follow after the original pine or mixed oak forest has been destroyed by timber cutting, agriculture or fire. It sometimes invades in almost pure stands! In such com- munities it commonly occurs with Myrica pginglei, Bac- charis conferta, Rubus spp. and Pteridium aquilinum var. feei (Miranda and Sharp, 1950). 195 Although Eyonia squamulosa has not yet been reported from Guatemala (Standley and Williams, 1966) it should be searched for there. It has been collected in Chiapas, and Steyermark (1950) suggests that the vege- tation of the Cuchumatanes and Sierra de las Minas in Guatemala may have much in common with the montane regions of Mexico. 2. Lyonia fruticosa (Michx.) G.S. Torrey E2 Robinson Lyonia fruticosa (Michx.) G.S. Torrey E3 Robinson, Prox. Am. Acad. Arts 51: 527. 1916. Andromeda ferruginea Walt. var. fruticosa Michx. Fl. Bor. Am. 1: 252. 1803. Xolisma fruticosa (Michx.) Nash, Bull. Torrey Bot. Club 22: 153. 1895. Lyonia ferruginea (Walt.) Nutt. var. fru- ticosa (Michx.) Rehder Ea Bailey, Cyclop. Am. Hort. 3: 960. 1900. Type: "in Florida and Georgia" Michaux (P?, not.seen). A small shrub, often colonial, usually rigidly ascending, up to 1.5 m tall (rarely 3 m). Young branches tfliin, rigid, coated with minute ferrugineous scales and often densely nonglandular pubescent, becoming glabrous or lightly pubescent with age. Buds ovoid, lepidote, (1.5-2.0 mm long. Leaves evergreen, moderately coriaceous, alternate, short—petioled, entire or rarely obscurely toothed, only rarely with revolute margins, obovate to oblanceolate, sometimes oval or elliptic, 0.5 to 5.5 cm 196 long, 0.3 to 2.8 cm wide, with tips acute, abruptly acuminate or obtuse, and bases cuneate, obtuse or atten- uate, adaxial surface glabrous to slightly lepidote with a slight nonglandular pubescence on mid-vein, abaxial surface lepidote and usually nonglandular pubescent, scales easily falling off thus mainly absent on older leaves, leaving their lower surface with a whitish color; leaves much reduced toward the end of the branches, flat I... or with a slight upward curve, with prominently reticulate, not depressed veins. Flowers in axillary corymbose fas- cicles, often densely clustered in the upper portions of the branches; pedicels lepidote and nonglandular pubescent, each with a pair of bracteoles at the base. Calyx 5-1obed, persistent, lobes triangular, 1.0-1.5 mm long, 0.5-1.0 mm 'wide, with abaxial surface lepidote, and adaxial surface 'with.sparse to moderate nonglandular pubescence. Corolla 'urceolate, with 5 short lobes, white, lepidote, 2.5-4.0 mm Long, 2.5-4.0 mm.wide. Stamens 10, inserted at the base <3f the corolla; filaments flattened, expanded near the base, S-shaped, roughened, unappendaged or with very small spur-like appendages on the back near the apex; anthers cflrtuse, the lobes parallel, lacking apical awns, dehiscent by large terminal pores, always with a white line of dis- ithegration tissue on the back of each lobe extending along the apex of the filament and into the short spurs if present. Style columnar or fusiform, not exerted; 197 ovary superior, 5-locular, tomentose and lepidote. Cap- sule ovoid, 5-angled, 3.0-5.0 mm long, 2.5-4.5 mm wide, loculicidal, with paler thickened sutures which separate as a unit from the five valves in dehiscence, lepidote and moderately to densely nonglandular pubescent, espe- cially at the base. Seeds ca 140 per capsule, very narrow, averaging 2.2 mm long (Figure 37). Distribution: Coastal Plain from Florida to Georgia (Figure 38). Flowering: throughout the year but mainly in May and June. Representative specimens: UNITED STATES. Florida. Alachua Co.: 9 miles south of Gainsville, 92525 s.n., 28 December 1935 (NCU). Baker Co.: 5 miles east of (Jlustee, Weigand & Manning 2389 (GH). Bradford Co.: locality unknown, Murrill £32}! (MO). Brevard Co.: 10 Iniles south of Melborne, Deam 57588 (NY). Broward Co.: 4 sales west of Deerfield, Eggg_§2§§§ (NY). Charlotte Co.: (Saloosa Forest, Parrott El (DUKE). Clay Co.: 3 miles west of Green Cove, Murrill 357 (US). Collier Co.: Marco IIsland, Lakela 22222 (FSU, GH). Columbia Co.: 24.5 miles ruxrth of Lake City, Godfrey E3112 (FSU*). Dade Co.: Buena Vista, Moldenke 326a (NY*, MO); Miami, Tracy 9210 (GB, MSC, TEX, NY, US). DeSoto Co.: south of Frostproof, Small E DeWinkeler 9588 (NY*). Duval Co.: vicinity of Mayport, Godfrey E Reinert 61200 (FSU*) . Gilchrist Co.: 198 Figure 37. Representative specimen of Lyonia fruticosa (Nash 700, MSC). 199 bani: mm (Michx.)form m‘. [.10- m PLANTS 0‘ CENTRAL PENINSULAR FLORIDA .‘ouzcuo m vu..~nv ()0 comm . nu coulnv "v I): - y ~A‘u “1.: f. I.“ 700. Xolisma fruhcou .\l . ... bar-nu ...c. 3...".... .4,” ”n.4,”, 200 Figure 38. Distribution of Lyonia fruticosa. 201 a H r . ... . I v p 4 I': V b 4 A L 1 1' . 1g}, . . J? I ‘\L V / m ,1 ‘ . - nonun- ‘ x ' -_-- t ' i x - ' \ 6' I I. n- ' ' ' ,.--/ , \ """ "" . \ . o O 202 3.8 miles west of Gilchrist-Alachua Co. line, about 6 miles east of Trenton, EE£§_EE.21° £223 (FSU, GH, NCU, US). Glades Co.: Indian Mound near Citrus Center, Small 9915 (GH). Highlands Co.: 3.5 miles south of Baias Den, 33y EE‘EE. 2E2; (NCU*). Jefferson Co.: 4 miles east of Drifton, Kral 6422 (FSU*, GA, GH). Lake Co.: vicinity of Eustis, Nash 700 (A, MSC, NY, US). Lee Co.: Pine Island, Tracy 7263 (GH, MSC, NY, US). Leon Co.: junction of US 20 and Fla. 267, Musselman 4123 (NCU). Levy Co.: 4 miles north of Cedar Key, Godfrey 56618 (DUKE, FSU, GH). Liberty Co.: between Telogia and Sumatra, Godfrey 62634 (FSU, NY, TEX). Madison Co.: 8 miles north of Greenville, Kral 3758 (FSU). Manatee Co.: near Bradentown, £353 Churchill E32}! 2 April 1923 (A, GH). Marion Co.: on Rt. 40, 2.3 miles west of junction with Rt. 19, Egg§.§,g., 16 July 1959 (MSC). Martin Co.: north of Port Salerno, Lakela 32231 (GH*). Nassau Co.: Amelia Island, Judd 314 (MSC*). Osceola Co.: marsh near Kissimmee, Hunnewell 8708 (GH). Palm Beach Co.: 3 miles east of Sunshine Parkway toward Boca Raton, McDaniel 9136 (DUKE). Pasco Co.: 1 mile east of Gowers Corner, Bay. 22 El: 2231 (GH*). Pinellas Co.: 5 miles south of Port Richey, 33y. EE.E£° 22§2_(FSU, NCU*). Polk Co.: near Lake Wales, Demaree ggggg (FSU*). Putman Co.: Welaka, University of Florida Conservation Reserve, EEE§.§rE-r 17, 18 July 1959 (FSU, MSC*, US). St. Johns Co.: without definite locality, 203 Reynolds 274 (US). St. Lucie Co.: 3 miles southwest of Fort Pierce, McCart 10485-A (NCU). Sarasota Co.: near Phillippe Creek, about 3 miles south of Sarasota, Ward 3135 (GH, NCU). Taylor Co.: at Steinhatchee, Godfrey 55953 (FSU*, GH, NCU); vicinity of Keatons Beach, Godfrey 61656 (FSU*, US). Volusia Co.: northeast edge of DeLand Air- port, Prichard 648 (NCU*). Wakulla Co.: Apalachicola National Forest, along Forest Rd. 309 between Helen and Lost Creeks, Clewell 1662 (FSU*). Georgia. Appling Co.: 4 miles northwest of Baxley, £222§.l§lgi (DUKE). Bacon Co.: 1.8 miles east of Alma, Duncan EEEEE (GA). Berrien Co.: 2 miles north-northwest of Ray City, Faircloth 1947 (NCU). Brantley Co.: just east of Brantley-Ware Co. line, Judd 291 (MSC*). Brooks Co.: 6 miles south of Morven, Bozeman 2211 (NCU*); 3.7 miles west of Lowndes-Brooks Co. line, Faircloth 3083 (NCU*). Camden Co.: along Crooked River, at the Crooked River State Park, Bozeman 22§§_(NCU*). Charlton Co.: 1.7 miles north of St. George, Bozeman 3312 (NCU). Clinch Co.: at Argyle, Clewell g§§l_(FSU*). Coffee Co.: 7.6 miles east of Douglas, Duncan EEEEE (GA). Col— quitt Co.: Indian Creek, 3.3 miles northwest of Berlin, Faircloth 2664 (NCU). Cook Co.: about 7.5 miles north of Sparks, 2293.321 (MSC*). Echols Co.: Tom Creek Swamp, 0.7 mile north of Fla. state line, Bozeman EEEQ (NCU*). Effingham Co.: 4 miles north of Stillwell, Duncan EEEEE (GA). Glynn Co.: 0.3 mile east of junction of US 84 & l7, 204 Bozeman 6440 (NCU). along Rt. 17, north of Fancy Bluff Creek, Mellinggr 232°! 11 June 1960 (GH*). Lanier Co.: along US 84, just east of the Alapaha River, Judd 281 (MSC*). Liberty Co.: 3 miles southwest of Hinesville, Wiegand E Manning 2387 (GH). Long Co.: along US 82, about 0.1 mile west of Liberty-Long Co. line, Judd 325 (MSC*). Lowndes Co.: 1.5 miles west of Dasher, Faircloth 1653 (NCU). McIntosh Co.: 1.3 miles south of north tip of Sapelo Island, Duncan EQEEE (DUKE, GH, NCU, US). Pierce Co.: about 0.5 mile south of Patterson, 22123.222l3.(NCU)' Tattnall Co.: 3.7 miles 24° southwest of Glenville, Padgett 111 (GA). Tift Co.: near Tifton, Haltern 292:! August 1926 (GA). Ware Co.: near Waycross, Godfrey §Q§Q§_ (FSU, GH, TEX); 0.3 miles west of Ware-Brantley Co. line on US 84, Judd 290 (MSC*). Wayne Co.: 5 miles east of Broad- hurst, Bozeman 6186 (NCU). This taxon is quite distinct from the closely related and sympatric Lyonia ferruginea; there is a long tradition beginning with Michaux (1803) in support of this view. Recently there has been a tendency to lump E. fruticosa with E. ferruginea (eg., Radford, Ahles, and Bell, 1968); this I consider to be a serious error in that the two taxa are separated by several good morpho- logical characters and very few plants appearing to be intermediate have been discovered. Perhaps Redford EE.El° 205 confused these two taxa because only L. ferruginea occurs in the range of the Flora 2E the Carolinas. Michaux (1803) lists E. fruticosa as a variety of E. ferruginea, giving as its distinguishing characters the growth form, the prominent reticulations of the veins of the leaf, the nonrevolute margins and the difference in flowering time (E.g., June for E. fruticosa and April for E. ferruginea). Asa Gray also considered these two taxa as distinct, pointing out that E. ferrgginea has narrower (sic) less reticulated leaves while E. fruticosa has mostly cuneate-obovate or rhomboidal leaves that are conspicuously reticulated. Elliott (1817) and Pursh (1814) called this taxon Andromeda ferruginea, and Nuttall (1818) called it Lyonia ferruginea, giving the name Andromeda (or Lyonia) rigida to the plant known as E. ferruginea today! This situation deve10ped from the ambiguous original description of Andromeda ferruginea given by Walter (1788). Pursh, Elliott and Nuttall took the description to refer to E. fruticosa, when in fact, it actually describes E. ferrgginea! In Walter's descrip- tion "follis rotundatis margine revolutis, subtus fer- rugineis” can only refer to E. ferruginea as it has markedly revolute leaves with a dense covering of fer— rugineous scales beneath. In contrast, E. fruticosa usually has nonrevolute leaves and tends to loose its ferrugineous scales as the leaves age. This confusion 206 does not alter Nuttall's transfer of Walter's Andromeda ferruginea to the genus Lyonia in 1818. Thus Lyonia ferruginea (Walt.) Nuttall is the correct name for E. ferruginea, although Nuttall would have been thinking of what is now E. fruticosa when he used this name! G. Don (1834) and DeCandolle (1839) in following Nuttall, made _ f the same error. I Small (1914, 1933), Wood (1961), Long and Lakela (1971) and Lems (1962, and unpublished notes) all con- sidered E. fruticosa to be specifically distinct from E. ferruginea. Part of the difficulty involving the identity of this species is that the characters which separate it from E. ferruginea are best seen in the field. Nash (1895) reports: [L onia fruticosa] is specifically distinct from XoIisma ferruginea, which grows from 6-12 ft. high, Has Iight green leaves which are very much rolled in on the margins, and flowers early in March. X. fruticosa never attains a height of more than 6 feet, has a much stricter habit, very dark leaves which are not revolute, and does not flower until two months later. The difference is quite marked in herbarium specimens, but in the field there can be no mistaking it. How the leaves are held on the stem, the amount of curva- ture of the leaf, the appearance and depression of the major veins, the density of ferrugineous scales on the abaxial leaf surface, and the growth form or appearance of the plant are all good characters in differentiating L. fruticosa from E. ferruginea. The importance of these vegetative characters was noted quite early, for example, 207 Elliott (1817) states that the leaves of E. ferruginea are "entire" and "convex" with "revolute" margins in contrast to the "very entire, generally flat" leaves of E. fruticosa (Figure 39). The table on page 210 summarizes the characters that Wood (1961), Long and Lakela (1971), Small (1914, 1933), Lems (unpublished notes) and I have found to be most helpful in differentiating E. fruticosa from E. fer- ruginea. Small (1914, 1933) used the corolla length as a distinguishing character, giving the values of 2.5-3 mm for E. ferruginea as compared to 3.5—5 mm for E. fruticosa. Long and Lakela (1971) also used the corolla length in differentiating these two taxa, giving a value of 2-3 mm for E. ferruginea and 4-5 mm for E. fruticosa. Actually this is not a good character for separating the two species as the range of overlap is too great. K. Lems (unpublished notes) suggested also that E. fruticosa has more densely tomentose leaves, larger flower clusters, wider and shorter leaves and shoots which continue to grow after flowering. Lyonia fruticosa does tend to have shorter leaves and thus leaves that are wider in proportion to their length than those of E. ferruginea, but there is a wide range of overlap between the two taxa in this character. Lyonia fruticosa does have a slight tendency to be more pubescent, but this character, too, breaks down frequently. I could 208 Figure 39. Leaf curvature and position in Lyonia fruticosa and E. ferruginea. a. E. fru- ticosa. b. E. ferruginea. TABLE 23. 210 Comparison of Lyonia fruticosa and E. ferruginea. E. fruticosa E. ferruginea 10. 11. Leaves much reduced 1. toward the ends of the branches Margins of leaves 2. usually not revolute Leaves tending to be 3. obovate or oblanceolate Leaves generally small 4. Small shrub with rigidly 5. ascending branches Leaves flat or arched 6. upward Leaf veins not depressed, 7. prominently reticulate Ferrugineous scales 8. usually deciduous-~thus absent from older leaves Peak of flowering in 9. March-April Slightly longer flowers, 10. 2.5 to 5.0 mm (average 3.3 mm) Very abundant in cut 11. over or disturbed areas Leaves only slightly reduced toward the end of the branches Margins revolute Leaves elliptic, ovate or obovate Leaves very small to quite large Shrub or small tree, branches not rigidly ascending Leaves with convex curvature Leaf veins depressed, not reticulate Ferrugineous scales usually not deciduous-- thus present on older leaves Peak of flowering in May-June shorter 2.0 to 4.0 mm 2.9 mm) Slightly flowers, (average Not as common in dis- turbed regions 211 not find any significant difference in the number of flowers per inflorescence. A difference in flowering time was observed by Michaux (1803), Pursh (1814) and Nash (1895), but Elliott (1817) states "the different periods of flowering, as they were determined by Michaux by cultivation, mark a dif- ference of habit; yet in the woods this difference is' not obvious, but there appears to be a constant succession of flowers from April to June." Which one of these views is correct? In examining herbarium specimens I found that both views are correct, in part (Figure 40). In Georgia there is a clear difference in flowering periods with E. ferruginea blooming in March and April, and E. fruticosa blooming in June and July! Farther south this sharp distinction breaks down and both can be found flowering at once, but even in Florida E. ferruginea blooms significantly earlier than E. fruticosa. In plants grown under uniform greenhouse conditions at East Lansing, Michigan, E. ferruginea began to flower early in March, while E. fruticosa did not begin to flower until late in April, not coming into full bloom until the end of June! Most of the specimens observed from the vicinity of Tampa, Florida, are completely lacking in nonglandular pubescence on the abaxial leaf surface. This interesting variation seems to be concentrated in just that region, 212 A.mmcflmsuumm .m n I “mmmmmmmmw .q u o "muozv .mmoflmsuumm am now mmooflusum mflcomq mo moofluwm mcflum3oam may mo comHHnmEoo d .os musmHm 213 5; I 00000 _ >42 an? i mum in Irp-Ip-r.IL .P- I» ¢¢¢¢¢¢¢¢¢I¢u0 06¢¢¢¢¢¢ .QO ¢II¢¢|I ‘ "' I, I ¢l" l¢ | 1\\ \ 9 IxerJaMcPuI On 2 214 although scattered specimens with only a very slight pubescence have been observed from throughout the range of the taxon. Lyonia fruticosa occurs in hammocks and pinelands in sandy acid soils on the Coastal Plain. It does not extend as far north as E. ferruginea; I have seen no collections from South Carolina, and have not seen it growing in either Jasper or Beaufort counties in South Carolina despite a careful search for it. Eyonia fruticosa commonly occurs with other ericaceous shrubs such as L. ligustrina var. foliosiflora, L. mariana, L. lucida, Leucothoe racemosa, Gaylussacia spp., Vaccinium spp. and Befaria racemosa. Other associated shrubs include Clethra alnifolia, Ilex glabra, Cyrilla racemiflora, Cliftonia monophylla and Myrica cerifera. Harshberger in his Vegetation of South Florida (1914) included E. fruticosa in his Sand-Pine (Rosmary Scrub), Slash—Pine and Pond—Pine formations. The pond- pine formation occurs on the ancient dunes, related in origin to the coastal dunes, which have covered an older, :flat land of limestone, which is seen in the valleys, or swales between the ridges. The dominant tree is Pinus calausa, the understory includes several species of Quercus, aIui Serenoa repens, Ceratiola ericoides, Vaccinium myrsin- :Ltes and Lyonia fruticosa. The vegetation of these £1_;‘ 215 ancient dunes is essentially xerophytic because the water, which falls as rain, rapidly percolates through the sand. The coriaceous leaves and covering of scales may be xerophytic adaptations, and indeed I observed in collecting that E. fruticosa could be kept for several hours without wilting! The slash-pine (Pinus caribaea) formation occurs on the geologically older flatlands. This forest consists of tall slash-pines of even age with an open crown admitting light to the forest floor beneath. Common understory species include Serenoa repens, Myrica cerifera, Quercus spp., Geobalanus oblongifolius, Vaccinium myrsinites, Gaylussacia dumosa, Lyonia fruticosa, Befaria racemosa, and Crossopetalum floridanum. During the dry season the plants are essentially under xeric con- ditions. The water which falls as rain is quickly lost by percolation through the sand and by limestone holes to underground channels (Harshberger, 1914). 3. Lyonia ferruginea (Walt.) Nutt. Lyonia ferruginea (Walt.) Nutt., Gen. N. Am. Pl. l: 266. 1818. Andromeda ferruginea Walt. Fl. Car. 138. 1788. Cassandra ferruginea (Walt.) Ndz., Bot. Jahrb. Syst. 11: 146, 148, 224. 1889. Xolisma ferruginea (Walt.) Heller, Cat. N. Amer. Pl. 6. 1898. Type: (BM?, not seen). Andromeda ferruginea Walt. var. arborescens Michx., Fl. Bor. Am. 1: 252. 1803. Eyonia ferruginea (Walt.) Nutt. var. arborescens (Michaux.) Rehd. $3 Bailey, Cyclop. Am. H 216 Hort. 3: 960. 1900. Type: "in Florida and Georgia" Michaux, (P?, not seen). Andromeda rigida Pursh, Fl. Am. Sept. 292. 1814. Eyonia rigida (Pursh) Nutt., Gen. N. Am. Pl. 1: 266. 1818. Type: "in barren sandy woods: Carolina to Florida" (not seen). .I A widely branching shrub or small tree, often colonial, to 6 m tall. Young branches thin, rigid, some- what angular, divergent, lepidote and nonglandular pubescent, becoming glabrous or only slightly pubescent with greater age; bark of trunk divided into long narrow ridges by shallow longitudinal furrows, reddish brown, separating into short thick plates. Buds ovoid, lepidote, 1.0-2.5 mm long. Leaves evergreen, moderately coriaceous, alternate, short petioled, entire, usually with revolute margins, elliptic to ovate or obovate, 1.0-7.5 cm long, 0.5-3.0 cm wide, with tips acute or short acuminate, and bases attenuate or cuneate, adaxial surface glabrous to slightly lepidote with slight nonglandular pubescence on midvein, abaxial surface lepidote and usually non- glandular pubescent; scales persistent, thus usually present on older leaves; leaves 222 greatly reduced in size toward the end of the branches, often with a down- ward curvature; veins obscure, quite depressed. Flowers in axillary corymbose fascicles, pedicels lepidote, some- times also nonglandular pubescent, each with a pair of 217 bracteoles at the base. Calyx S-lobed, persistent, lobes triangular, 1.0-2.0 mm long, 0.5-1.0 mm wide, with abaxial surface lepidote, and adaxial surface glabrous or with sparse nonglandular pubescence. Corolla urceolate, with 5 short lobes, white, lepidote, 2.0-4.0 mm long, 2.0-4.0 mm wide. Stamens 10, inserted at the base of the corolla; filaments flattened, expanded near the base, S-shaped, roughened, unappendaged or with very short spur-like appendages on the back near the apex; anthers obtuse, the lobes parallel, lacking apical awns, dehiscent by large terminal pores, always with a white line of disintegration tissue on the back of each lobe extending along the apex of the filament and into the short spurs if present. Style columnar or fusiform, not exerted; ovary superior, 5-locular, tomentose and lepidote. Capsule ovoid, S-angled, 3.0-6.0 mm long, 3.0-4.0 mm wide, loculicidal, with paler thickened sutures which separate as a unit from the 5 valves in dehiscence, lepidote and moderately to densely nonglandular pubescent, especially at the base. Seeds ca 90 per capsule, very narrow, averaging 2.3 mm long, Figure 41. Distribution: Coastal Plain from Florida to South Carolina (Figure 42). Flowering: mainly in March, April and May. Representative specimens: Florida. Bay Co.: 5 miles north of Lynn Haven, Lazor 5009 (FSU*); St. Andrews 218 Figure 41. Representative specimen of Lyonia fer- ruginea (Griscom 21596, GH). 219 Lu‘( ‘\ "BIA U} \URTNWL‘TTJL\ FLUuID\ L. “A... Man i m \n LI nun. “In "I 220 Figure 42. Distribution of Lyonia ferruginea. 221 222 State Park, Wooten 208 (FSU*). Brevard Co.: north of Coca, Rhodes 8305 (A). Calhoun Co.: 3 miles west of Blountstown, Coker 3.3., 26 February 1947 (NCU). Clay Co.: sandy bank of Kingsley Lake drain, Camp Blanding, Totten §.E., 2 April 1943 (NCU, NY*). Columbia Co.: on bluff of Santa Fe River, northeast of High Springs, Godfrey 52847 (DUKE, FSU, GH, MISSA, NCU, NY). Dixie Co.: 0.8 mile west of Suwannee River, about 13.9 miles north of Suwannee, Traverse 736 (GH*). Duval Co.: vicinity of Mayport, Godfrey E Reinert 61199 (FSU*). Franklin Co.: near St. Theresa, Godfrey 53100 (DUKE, FSU, NCU, NY*); St. Theresa, Griscom 21596 (GH). Gulf Co.: near Wewahitchka, Demaree E9332 (NCU). Hernando Co.: Weeki-Wachee, Cooley §11§_(FSU, NY*). Highlands Co.: near Lake Placid, Demaree E22§2_(FSU*, NCU). Hills- boro Co.: sand barrens Hillsboro River, 2.5. Churchill .§\E°' 9 April 1923 (GH). Jefferson Co.: without definite locality, Lighthipe 3.3., 15 March 1891 (NY). Lake Co.: near Astor, Clewell 2108 (FSU*); vicinity of Eustis, Nash El (A, MSC, NY, US). Lafayette Co.: 4.5 miles north of Day, Duncan 18240 (GA). Leon Co.: near Lake Bradford, Godfrey 57997 (FSU*). Levy Co.: Cedar Key, Godfrey 56607 (DUKE, FSU, GH). Liberty Co.: near Hosford, E.g. Palmer 38526 (A, US). Marion Co.: Ocala National Forest, Perry 1643 (DUKE). Orange Co.: near Orlando, §.g. Palmer 38306 (A, US). Osceola Co.: north of Lake 223 Tohopekaliga, Lemaire 362 (FSU). Pasco Co.: near Jessa- mine, Barnhart 2540 (NY). Pinellas Co.: near Dunedin, Tracy 6648 (GH, MSC). Polk Co.: about 2 miles northeast of Lakeland, Ford EE.El° 2440 (NCU). Putnam Co.: Welaka, University of Florida Conservation Reserve, Lems 3.3., 17, 18 July 1959 (MSC). St. Johns Co.: about 2.2 miles south of Crescent Beach, Wood 9329 (NCU*). Seminole Co.: 6 miles southwest of Sanford, Duncan 4607 (GA). Sumter Co.: west of Bushnell, Kral 7870 (GA, US). Taylor Co.: I between Buckville and Perry, Godfrey 52855 (DUKE, FSU, GH, MISSA, NY*). Volusia Co.: sand dunes opposite Daytona, Small E DeWinkeler 9423 (NY). Wakulla Co.: St. Marks Wildlife Refuge, Godfrey 71410 (FSU*); between Panacea and Sopshoppy, St. Marks Wildlife Refuge, Trott 181 (FSU*). Walton Co.: without definite locality, Ashe 3.3., 20 March 1927. Georgia. Appling Co.: without definite locality, Duncan 16290 (US). Bacon Co.: 7.5 miles east of Alma, Bozeman 4471 (NCU). Berrien Co.: 6.1 miles north-northeast of Nashville, Faircloth 2142 (NCU). Brantley Co.: on Ga. 259, ca 5.0 miles south of junction with US 84, Bozeman 4014 (NCU). Brooks Co.: 5.5 miles north-northwest of Quitman, Faircloth 1982 (NCU). Bryan Co.: 2 miles west of the Ogeechee River on Ga. 204, Bozeman £332 (NCU). Camden Co.: Little