Botany Publication and Papers Botany

3-1994 Morphological Diversity and Relationships in the A-Genome , arboreum and G. herbaceum Marsha A. Stanton University of Arkansas - Main Campus

J. McD. Stewart University of Arkansas - Main Campus

A. Edward Percival United States Department of Agriculture

Jonathan F. Wendel Iowa State University, [email protected]

Follow this and additional works at: http://lib.dr.iastate.edu/bot_pubs Part of the Botany Commons, Genetics Commons, and the Breeding and Genetics Commons

Recommended Citation Stanton, Marsha A.; Stewart, J. McD.; Percival, A. Edward; and Wendel, Jonathan F., "Morphological Diversity and Relationships in the A-Genome Cottons, Gossypium arboreum and G. herbaceum" (1994). Botany Publication and Papers. 7. http://lib.dr.iastate.edu/bot_pubs/7

This Article is brought to you for free and open access by the Botany at Iowa State University Digital Repository. It has been accepted for inclusion in Botany Publication and Papers by an authorized administrator of Iowa State University Digital Repository. For more information, please contact [email protected]. Morphological Diversity and Relationships in the A-Genome Cottons, Gossypium arboreum and G. herbaceum

Abstract The Asiatic or A-genome cottons, Gossypium arboreum L. and G. herbaceum L., are potentially important genetic resources for breeding programs. The aN tional Plant Germplasm System (NPGS) contains approximately 400 accessions of these species, but little information is available on the diversity within the collection or on characteristics of individual accessions. This investigation was initiated to provide a morphological description of each accession. These data were used to evaluate the range of diversity represented within the collection and to address the questions of species and race distinctions. Multivariate techniques were used to assess similarities among accessions and to evaluate morphological parameters contributing to the variation in each species. Means for 41 of 53 characters were significantly different between species, although high intraspecific av riability resulted in range overlap for all characters. Principal component analysis separated the two species. Accessions from southern Africa and racial designations of africanum and wightianum formed clusters within G. herbaceum based on the first two principal components; no clusters were noted within G. arboreum. Accordingly, the validity of intraspecific or ar cial classifications for most of the accessions of this latter species in the current NPGS collection is questionable. Additional germplasm acquisitions from under- or non-represented areas could expand genetic diversity in the collection.

Keywords Gossypium herbaceum, G. arboreum, tetraploid cotton species, National Plant Germplasm System, Germplasm Resources Information Network, principal component analysis

Disciplines Botany | Genetics | Plant Breeding and Genetics

Comments This article is from Crop Science 34 (1994): 519, doi:10.2135/cropsci1994.0011183X003400020039x.

Rights Works produced by employees of the U.S. Government as part of their official duties are not copyrighted within the U.S. The onc tent of this document is not copyrighted.

This article is available at Iowa State University Digital Repository: http://lib.dr.iastate.edu/bot_pubs/7 Published March, 1994

PLANT GENETIC RESOURCES

Morphological Diversity and Relationships in the A-Genome Cottons, Gossypium arboreum and G. herbaceum Marsha A. Stanton, J. McD. Stewart,* A. Edward Percival, and Jonathan F. Wendel

ABSTRACT HE GENUS Gossypium L. contains four cultivated The Asiatic or A-genome cottons, Gossypium arboreum L. and G. T species: the New World allotetraploids (2n = 4x herbaceum L., are potentially important genetic resources for cotton = 52), G. barbadense L. and G. hirsutum L., and the breeding programs. The National Plant Germplasm System (NPGS) Asian-African diploids (2n = 2x = 26), G. arboreum contains approximately 400 accessions of these species, but little infor­ L. and G. herbaceum L. These latter two species are mation is available on the diversity within the collection or on character­ collectively called a variety of terms including Old istics of individual accessions. This investigation was initiated to pro­ World, Asiatic, and A-genome cottons. The last designa­ vide a morphological description of each accession. These data were tion will be used herein because it most accurately defines used to evaluate the range of diversity represented within the collection the two species. Allotetraploids (AD genome) currently and to address the questions of species and race distinctions. Multivari­ ate techniques were used to assess similarities among accessions and are responsible for 95% of world cotton production, to evaluate morphological parameters contributing to the variation in having nearly displaced A-genome species from the range each species. Means for 41 of 53 characters were significantly different in which they were previously cultivated (Meredith, between species, although high intraspecific variability resulted in 1991). Gossypium arboreum (Az) is still cultivated on range overlap for all characters. Principal component analysis sepa­ marginal land in and , with active breeding rated the two species. Accessions from southern Africa and racial programs emphasizing its use in non-woven materials designations of africanum and wightianum formed clusters within G. (M. Akhtar, 1992, personal communication). Limited herbaceum based on the first two principal components; no clusters cultivation of G. herbaceum (At) occurs in Africa (Lee, were noted within G. arboreum. Accordingly, the validity of intraspe­ 1984). Today, the primary value of the A-genome cottons cific or racial classifications for most of the accessions of this latter is as a source of genetic diversity for the tetraploid species in the current NPGS collection is questionable. Additional germplasm acquisitions from under- or non-represented areas could cottons. With the development of techniques for intro­ expand genetic diversity in the collection. gression of A-genome germplasm into AD-genome culti­ vars (Stewart, 1992; Stewart and Hsu, 1978), these cot­ tons have become a more readily usable germplasm M.A. Stanton and J.McD. Stewart, Dep. of Agronomy, 115 Plant Science resource for tetraploid cotton breeding programs. Bldg., Univ. of Arkansas, Fayetteville, AR 72701; A.E. Percival, USDA, ARS, Route 5, Box 805, College Station, TX 77845; and J.F. Wendel, In past breeding efforts, A-genome cottons have con- Dep. of Botany, Iowa State Univ., Ames, lA 50011. Published with the approval of the Director of the Arkansas Agric. Exp. Stn. Received 14 Abbreviations: A~o Gossypium herbaceum; A2, G. arboreum, AD, tetra­ Dec. 1992. *Corresponding author. ploid cotton species; NPGS, National Plant Gerrnplasm System; GRIN, Gerrnplasm Resources Information Network; PC, principal component; Published in Crop Sci. 34:519-527 (1994). PCA, principal component analysis. 520 CROP SCIENCE, VOL. 34, MARCH-APRIL 1994

tributed high fiber strength and resistance to Pectinophera Table 1. Definitions of morphological characters for Gossypium gossypiella Saunders, Puccinia cacabata Arth. and arboreum and G. herbaceum. Holw., and some races of Xanthomonas campestris pv. malvacearum to AD-genome cultivars (Fryxell, 1984; color: 1 = white, 2 = cream, 3 = white wired edge, 4 = It. yellow, 5 = yellow, 6 = yellow wired edge, 7 = red. Meredith, 1991). Additionally, resistance to nematodes Petal spot: 0 = none, 1 = white, 2 = cream, 3 = pink, 4 = red. (Carter, 1981; Yik and Birchfield, 1984), fungal and Pollen color: 1 = yellow, 2 = cream. Petal length: distance from attachment to corolla edge at anthesis bacterial pathogens (Mathre and Otta, 1967; Bollen­ (em). bacher and Fulton, 1971), and insects (Benedict et al., Leaf 1987; Stanton et al., 1992) has been found in A-genome Color: 1 = yellow-green, 2 = It. green, 3 = green, 4 = dk. green, cottons. These sources of genetic resistance take on added 5 = green wired veins, 6 = red. Number of lobes: average number of lobes on 3 fully expanded leaves. importance since modem upland cultivars have a rela­ Lm length: distance from to tip of central lobe (em). tively narrow genetic base (Wendel et al., 1992) in spite Ls length: distance from petiole to tip of first lateral lobe (em). of documented interspecific introgression (Endrizzi et Lm width: width of central lobe at widest point (em). Lobe shape: Lm width!Lm length(%). al., 1985; Meredith, 1991). Lobe length: Ls length!Lm lobe length(%). Morphological resemblance between G. arboreum and Si: distance from base of sinus to the mid point of an imaginary line connecting the tips of Lm and Ls (em). G. herbaceum makes individual specimens difficult to Sinus depth: Si/LM length(%). identify, especially if fruiting material is not available. Glands: number of gossypol glands per 5 mm on main leaf vein counted Wendel et al. (1989) indicated that the mean values of between nectary and point of attachment of petiole. 25 morphological characters differed significantly be­ Stipule: average length of stipules on a fruiting branch (em). Bract tween the two species; but, infraspecific variability was Sunred: degree of light induced anthocyanin in ; 0 = none to sufficiently high that ranges of all morphological charac­ 5 =high. ters overlapped between species. Length: distance from base of bract to tip of central tooth (em). Width: distance at widest point (em). At the chromosomal level, the species differ by a Number of teeth: average number of teeth per bract. reciprocal translocation (Gerstel, 1953), and in chromo­ Teeth length: average length of three central teeth (mm). some length, symmetry, and number of secondary con­ Nectaries strictions and satellites (Gennur et al., 1988). Measures Leaf and flower nectaries: presence = 1, absence = 0. Leaf nectary area: nectary length x width (mm2). of allozyme variability indicate that G. arboreum contains Pubescence greater variability than G. herbaceum, and both species Stem rating: 0 = glabrous to 5 = very dense pubescence. exhibit diversity roughly equivalent to that found among Stem, 2 layers: 1 = present, 0 = absent. Main stem: length of most prevalent layer of pubescence, measured above upland G. hirsutum cultivars (Wendel etal., 1989; 1992). highest unrolled leaf. This is only about half of the diversity normally found Minor stem: length of minor layer of pubescence, measured above highest among cultivars of other domesticated plant species (Doe­ unrolled leaf. Simple hairs: length of simple hairs on lower surface of mature leaf (mm). bley, 1989). In addition to the decline in genetic diversity Stellate hairs: maximum length of stellate hair branches on lower surface normally associated with domestication (Doebley, 1989), of mature leaf (mm). the dramatic decline in cultivation of A-genome cottons Boll Length: perpendicular distance from apex to base (em). likely has resulted in additional erosion of genetic diver­ Width: diameter at point of maximum circumference (em). sity. and fiber Variability within G. arboreum and G. herbaceum Fuzz density: degree of fuzz fibers on seed; 0 = none to 5 = very dense. has been partitioned into infraspecific races based on a Lint color: 1 = white, 2 = off-white, 3 = tan. Seed index: weight of 100 fuzzy seed (g). combination of geographical and morphological factors Naked seed weight: weight of 100 acid-delinted seed (g). (Hutchinson and Ghose, 1937; Silow, 1944; Hutchinson, Lint percent: ginned lint wt./wt. of seed cotton(%). 1950). These racial concepts were not intended to provide E1 t: percent elongation at point of break in strength determination. Micronairet: fiber fineness (micronaire units). distinct morphological delimitations among the races, Tlt: fiber strength when clamped at 3.2 mm (kN m kg"'). and assignment to a particular race is difficult for acces­ 50% span Iengtht: length spanned by 50% of the fibers scanned at the initial starting point (mm). sions of unknown origin. One of the described races, 2.5% span Iengtht: length spanned by 2.5% of the fibers scanned at the africanum, has been assigned to various rankings below initial starting point (mm). the species level and is currently given sub-species rank At: external surface area of fibers measured at standard pressure (mm21 mm'). (Vollesen, 1987). AHt: external surface area of fibers measured at high pressure (mm21 3 The Asiatic Cotton Germplasm Collection of the Na­ mm ). tional Plant Germplasm System (NPGS) contains approx­ Difference (D)t: AH - A, approximation of flatness of fiber (mm21mm'). Maturityt: caustic soda percent maturity. 112 imately 400 accessions, few of which have been charac­ Immaturity ratio (Ot: (0.07*0 + 1) • terized or evaluated (Percival, 1987). This study was Perimetert: (12566*01A; (!JDl). 2 Weight finenesst: (485*10'*01A ; (pg). initiated to provide morphological descriptions of the 112 Wall thicknesst: 20001[A(1 + (1 - 110 )); (!JDl). accessions and determine the range of diversity repre­ t From standard breeder's test, by Starlab, Knoxville, TN; for more complete sented within the collection. The questions of species descriptions of measurements see Regional Cotton Variety Tests (1991). and race distinctions are also addressed.

MATERIALS AND METHODS Station, TX. Additional materials were obtained from working Plant Materials collections at the University of Arkansas, Fayetteville, AR, and Iowa State University, Ames, lA. All accessions were of most of the accessions in the NPGS Asiatic Cotton field grown at Fayetteville, AR, in one or more of the years Germplasm Collection were obtained from the USDA, College 1987 through 1992. Many of the accessions were also grown STANTON ET AL.: MORPHOLOGICAL DNERSITY IN THE A-GENOME COTTONS 521 at College Station, TX, in 1991 for verification of observations. accessions originated from East and Southeast Asia while For the accessions grown at Fayetteville, one or more herbar­ only 10% came from Central Asia. No accessions origi­ ium specimens were prepared as vouchers for each accession. nated from Africa. The origins of 30% of the G. arbor­ This consisted of a main stem leaf and a flowering branch eum accessions are unknown. Thirty-two percent of the taken from a typical plant within a 3.3-m row. A complete G. herbaceum accessions came from Central Asia, 18% voucher specimen was not obtained for 42 accessions due to low plant number (low seed viability) or failure to flower from Southeast Asia, 12% from the Middle East, and under field conditions. Additional accessions were not included 7% from East Asia. Only 11% originated in Southern for reasons given later, thus, detailed analyses of diversity Africa where the only wild populations of this species involved 272 G. arboreum and 84 G. herbaceum accessions. are extant. The origins of 19% of the accessions are A list of included accessions identified by their NPGS numbers unknown. may be obtained through GRIN or from the corresponding Because of the lack of basic characterization data for author. the accessions in the NPGS Asiatic Cotton Germplasm Collection, preliminary observations were made on all Passport Data accessions for which viable seed could be obtained. When available, passport data for each accession was used Twenty-three accessions in the collection were deter­ to assign a geographic origin to the accession. For purposes mined to be G. hirsutum or G. barbadense, and 17 of analysis the countries of origin were grouped as follows. appeared to be segregating populations of interspecific Middle East: Iraq, Sudan, and Turkey; Central Asia: Afghani­ crosses between G. arboreum and G. herbaceum. Fifteen stan, Iran, Pakistan, Russia, and Uzbekistan; East Asia: China, accessions recorded as G. herbaceum were identified as Japan, and Korea; Southeast Asia: Burma, India, Indonesia, G. arboreum, and 11 accessions recorded as G. arboreum and Philippines; Southern Africa: Africa, Botswana, South were identified as G. herbaceum. These accessions will Africa, and Zimbabwe; unknown: no origin listed or the origin listed is outside the range of A-genome cotton cultivation. be assigned new numbers in the appropriate species collection (GRIN, 1989). Sixty-seven accessions exhib­ Morphological Analysis ited morphological variants that we fixed by self­ pollination of individual . These variants within Fifty-three characters comprising 41 morphological traits an accession will retain the original accession number and 12 proportional characters expressed as percentages or plus a subclassification in the Germplasm Resources ratios were evaluated. Acronyms and descriptions of the char­ acters are defined in Table 1. Color and boll measurements Information Network (GRIN). Passport data, as far as were taken on live plants; other measurements were made on known, and our characterizations for each accession herbarium specimens or harvested fiber and seed samples. in the Asiatic collection are available through GRIN Characters were selected from an assessment of potentially (Percival, 1987; GRIN, 1989). This base information important variation based on initial observations of the range should provide for a more orderly use of the collection. within the collection and on published taxonomic treatments For purposes of subsequent analyses, accessions we (Hutchinson et al., 1947; Fryxell, 1979). Ratios of selected identified as G. hirsutum or hybrid swarms were ex­ primary characters were used to describe relative shape of cluded. Mislabeled accessions were included after being plant parts. Fiber characteristics were measured by STAR appropriately labeled, and the stable variants were in­ LAB, Knoxville, TN, on samples obtained from field-grown cluded as separate accessions. This resulted in a total of plants in 1988, 1990, and 1991. At this initial level of character­ ization of the many accessions, it was not practical to test for 356 entries being analyzed for morphological diversity. genotype-environment interactions. Summary statistics for the 356 accessions are shown in Table 2. Since the ranges for every character overlap Data Analysis between species, no single character serves to delineate these species. However, means for 41 of the characters Multivariate techniques were used to assess the similarities differed significantly when compared by a t-test (P ~ among accessions and to evaluate morphological parameters contributing to the variation in each species. Quantitative mor­ 0.05) (SAS, 1989). The range extremes of morphological phological characters that were not invariant or highly corre­ characters were greater for G. arboreum in 29 cases, lated to another character were evaluated using principal com­ for G. herbaceum in 16 cases, and equivalent in eight ponent analysis (PCA). Data were standardized by character cases. Two characters, petal color and petal spot, were (mean = 0, standard deviation = 1). Principal components invariant in G. herbaceum. In spite of the two invariant were calculated from a matrix based on correlations between characters and the range differences, the mean coefficient the characters. A cophenetic correlation value was used to of variation for G. herbaceum (35.6%) was not signifi­ evaluate results from the principal component analyses. This cantly different from G. arboreum (37.6%). value measures the correspondence between a similarity matrix Twenty-six quantitative characters were selected for based on the original data and a similarity matrix generated use in the principal component analyses. In most cases, based on the models developed (Wiley, 1981). Specific proce­ dures for PCA were executed with the microcomputer program characters that were highly correlated were not both NTSYS (Version 1.7) (Rohlf, 1992). used. Four characters that were used to compute ratios, and, therefore, obviously correlated to the ratio com­ RESULTS puted, were retained to capture both the size and shape components of variation. A few accessions never flow­ The available passport data for the A-genome acces­ ered and, therefore, have missing flower and fiber data. If sions give an indication of the geographic regions repre­ patterns of missing data corresponded to any relationships sented in the collection. Nearly 60% of the G. arboreum among the accessions (e.g., earliness or photoperiod 522 CROP SCIENCE, VOL. 34, MARCH-APRIL 1994

Table 2. Summary statistics for measured and derived cbaracterst for Gossypium arboreum and G. herbaceum. G. arboreum G. herbaceum Structure/Parameter N Range Mean so N Range Mean so Flower Petal color 259 1.0-7.0 4.0• 1.8 78 1.0-8.0 s.o• 0.7 Petal spot 260 0.0-4.0 3.9 0.7 79 4.0-4.0 4.0 0.0 PoUen color 260 1.0-2.0 1.0 0.1 78 1.0-1.0 1.0 0.0 Petal length (em) 210 2.1-6.0 3.9 0.8 53 2.0-5.0 3.6• 0.7 Petal/bract(%) 210 45.0-145.0 83.2• 20.5 53 38.0-167.0 73.3• 21.3 Leaf Number of lobes 238 3.0-7.0 s.s• 0.9 71 3.0-7.0 s.8• 1.0 Color 261 1.0-6.0 3.2 0.9 72 1.0-6.0 3.0 0.6 Lm length (em) 238 5.2-16.0 9.2 1.7 72 4.5-12.5 8.8 1.9 Lm width (em) 238 0.8-4.2 2.7• 0.8 72 2.3-5.8 4.0• 0.8 Ls length (em) 238 4.4-13.0 7.7• 1.5 72 4.0-11.0 7.1• 1.6 Lobe shape(%) 238 11.0-53.0 29.7• 8.5 72 21.0-58.0 46.1• 6.4 Lobe length(%) 238 52.0-104.0 84.4• 6.7 72 37.0-96.0 81.2* 10.1 Sinus depth(%) 233 15.0-93.0 51.1* 13.0 72 26.0-66.0 41.8* 7.1 Glands 232 0.5-28.7 10.1* 4.9 71 1.1-38.2 12.0* 6.9 Stipule (em) 241 2.0-22.0 9.7 2.9 68 6.0-23.0 9.9 3.1 Bract Sunred 266 0.0-5.0 2.3• 1.1 81 0.0-5.0 1.6* 1.1 Length (em) 229 1.7-5.0 3.1• 0.6 63 1.3-3.5 2.5• 0.5 Width (em) 228 1.3-4.0 2.3 0.5 63 1.0-3.6 2.2 0.5 Number teeth 221 0.0-10.0 4.5• 1.9 61 3.0-15.0 8.4* 2.0 Teeth length (mm) 212 1.0-12.0 4.1• 2.2 58 2.0-11.0 5.7• 2.2 Teeth/bract (%) 211 2.0-34.0 13.2• 6.7 ss 6.0-41.0 23.3• 7.2 Width/length (%) 228 45.0-127.0 76.1* 12.0 63 63.0-127.0 91.6* 14.9 Nectary Leaf nectary 242 0.0-1.0 0.7• 0.4 72 20.0-1.0 o.8• 0.3 Flower nectary 246 0.0-1.0 0.7• 0.4 67 70.0-1.0 0.9• 0.2 2 Nectary area (mm ) 183 0.1-3.4 0.9• 0.6 67 70.1-2.5 o.8• 0.6 Pubescence Leaf 235 0.0-1.0 1.0* 0.1 69 90.0-1.0 o.8• 0.4 Stem, 2 layers 239 0.0-1.0 0.4* 0.5 70 00.0-1.0 0.2• 0.4 Stem rating 256 0.0-5.0 2.8 1.2 82 1.0-5.0 2.8 1.0 Main stem (mm) 237 0.0-2.1 0.7 0.6 70 0.1-2.1 0.7 0.7 Minor stem (mm) 142 0.1-2.0 o.8• 0.5 53 0.2-2.4 1.3* 0.6 Simple hairs (mm) 231 0.1-1.9 0.7• 0.3 58 0.1-1.8 0.6• 0.3 SteUate hairs (mm) 240 0.1-1.0 0.3• 0.1 70 0.2-1.0 0.4• 0.1 BoD Length (em) 148 2.9-5.9 3.7• 0.4 27 2.4-4.0 3.2• 0.4 Width (em) 148 1.9-6.6 2.6 0.4 27 2.0-3.5 2.7 0.4 Width/length (%) 148 50.0-180.0 70.0* 10.4 27 66.0-98.7 8s.o• 8.3 Seed/Fiber Fuzz density 186 0.0-3.0 2.0• 0.7 37 1.0-5.0 2.5• 0.7 Lint color 196 1.0-3.0 1.1 0.3 43 1.0-3.0 1.2 0.6 Seed index (g) 158 4.7-12.8 7.0• 1.1 30 4.6-12.0 8.4* 2.0 Naked seed wt. (g) 194 3.7-9.2 6.8• 1.0 34 5.4-11.1 7.8• 1.5 Lint percent 170 0.0-48.8 32.9* 6.1 29 13.9-40.3 26.8* 6.4 E1(%) 163 2.0-10.5 7.3 1.6 36 4.0-9.5 6.8 1.6 Micronaire 216 4.1-8.0 6.7• 1.0 44 2.5-7.5 s.o• 1.1 T1 (kN mlkg) 190 100.9-283.0 187.1* 31.0 39 104.0-236.0 174.9* 30.3 SO% SL (mm) 223 8.5-24.1 12.5* 3.1 46 9.1-16.7 11.6* 1.8 2.5% SL (mm) 223 13.7-30.2 20.4 3.0 46 17.3-24.5 20.9 1.8 2 3 A (mm /mm ) 171 244.5-488.5 318.4* 44.5 46 253.5-572.0 401.1* 74.3 Au (mm2/mm3) 171 254.5-510.0 328.7• 49.1 46 260.0-660.0 436.8* 96.4 Au-A (mm'/mm3) 164 0.5-33.5 10.8* 6.7 46 4.0-92.5 35.7• 23.0 Maturity(%) 171 80.5-112.0 100.3• 6.8 46 46.0-107.0 80.6• 15.8 Immaturity ratio 171 1.0-1.8 1.3* 0.2 46 1.1-2.7 1.8* 0.4 Perimeter (pm) 171 38.3-72.9 51.7• 6.4 46 42.7-64.9 56.7• 4.9 Wt. fineness (pg) 171 3.2-11.5 6.4• 1.4 46 4.0-9.1 5.6• 1.0 wan thickness (pm) 171 2.5-7.7 4.6• 1.0 46 2.0-5.6 3.2• 0.8 • Means between species are significantly different (P = 0.05). t See Table 1 for definitions of characters. requirements imposed by similar origins), the results Principal component analysis was performed on the would be biased (Jackson, 1991). The cophenetic correla- combined data set and individually on each species in tion between PCA with and without fiber, seed, and boll order to examine infraspecific diversity and to extract data was 0.94, indicative of a very good fit between important components of variation within each species the two models. We chose to use the fiber data, even (Table 3). For the G. arboreum and combined data sets, though they were absent for approximately 27% of the cophenetic correlations (0.86 and 0.87, respectively) accessions, because human-mediated selection based on indicated a good fit between a distance matrix based on fiber traits probably contributed to differences within the original data and a distance matrix based on 25 each species. principal components, the number of principal compo- STANTON ET AL.: MORPHOLOGICAL DIVERSITY IN THE A-GENOME COTTONS 523

Table 3. Characters used in the principal component analyses of G. arboreum and G. herbaceum, the loading of each character on principal components I, ll, and ill, and the variance explained by each component. Combined data sett G. arboreum G. herbaceum Charactert II m II m II m Bract length 0.51 0.46 0.03 -0.55 -0.18 -0.34 0.66 -0.33 -0.12 Bract shape -0.57 0.06 -0.16 0.10 -0.09 0.21 -0.08 0.67 0.23 Bract tooth length -0.34 0.62 0.11 -0.46 -0.04 -0.64 0.79 -0.15 0.04 BoU length 0.46 0.12 -0.40 -0.09 -0.49 -0.05 0.71 0.22 -0.46 BoU shape -0.53 0.34 0.14 -0.25 0.27 -0.05 -0.10 0.64 0.38 BoU width -0.17 0.48 -0.17 -0.31 -0.07 -0.10 0.62 0.71 -0.18 Bt length/bract length -0.57 0.43 0.09 -0.28 0.01 -0.55 0.57 -0.06 0.23 E1 0.05 -0.03 -0.18 -0.02 -0.06 -0.03 -0.05 0.34 -0.70 Glands -0.12 0.28 0.09 -0.20 0.13 0.00 0.43 0.13 0.13 Leaf color 0.19 -0.21 0.04 0.05 0.13 0.42 -0.39 0.04 0.17 Lint percent 0.61 0.36 -0.03 -0.59 0.06 0.39 0.63 -0.43 -0.45 Lm length 0.22 0.50 0.03 -0.46 0.04 0.02 0.47 -0.05 0.10 Lobe length 0.23 -0.04 -0.21 -0.15 -0.36 -0.05 -0.26 -0.05 -0.06 Lobe shape -0.77 -0.18 -0.33 0.54 -0.52 -0.10 -0.19 0.66 -0.34 Micronaire 0.59 0.18 -0.38 -0.43 -0.47 0.43 0.28 -0.31 0.15 Naked seed weight -0.40 0.22 -0.45 0.07 -0.61 -0.14 0.53 0.70 0.05 Nectary area 0.21 0.21 -0.08 -0.22 -0.09 0.24 0.16 -0.23 0.35 Number of teeth -0.64 0.39 -0.16 -0.29 -0.38 -0.44 0.21 0.27 0.31 Petal length/bract length 0.36 0.41 0.15 -0.53 0.12 -0.02 0.35 -0.38 -0.44 Sinus depth 0.49 0.31 0.14 -0.53 0.20 0.13 0.43 0.25 0.38 SteUate pubescence -0.13 0.18 0.31 -0.18 0.36 0.17 0.19 0.05 0.11 Stem pubescence, main 0.06 0.19 0.65 -0.16 0.68 0.05 0.39 -0.20 0.48 Stipule 0.02 0.06 0.36 -0.03 0.27 -0.15 0.19 -0.10 0.54 Sunred 0.21 -0.42 -0.16 0.31 -0.10 0.27 -0.47 -0.03 -0.01 T1 0.02 -0.24 0.54 0.30 0.41 -0.38 -0.17 -0.36 0.21 2.5% span length -0.25 -0.26 0.46 0.46 0.37 -0.45 0.27 0.08 0.26

Variation explained (%) 16 10 8 12 10 8 18 13 10 Cumulative variation (%) 16 26 34 12 22 30 18 31 41 t G. arboreum and G. herbaceum data combined before standardization. t See Table 1 for descriptions.

nents (PCs) necessary to capture 100% of the variation model accounted for 34% of the variance (Table 3), and in each data set. In contrast, the cophenetic correlation the differences between the two A-genome species are value for G. herbaceum (0.90) indicated a very good apparent in plots of the first two PCs (Fig. 1). An fit, and fewer (19) PCs than in the combined or G. examination of characters with the highest factor loading arboreum models were required to obtain 100% of the show the first component to be primarily based on charac- variation in the data set. ters that have been used to differentiate the species (no. The first three components in the combined species teeth, tooth ratio, bract length, bract width/length, leaf

1.5 • G. herbaceum o G. arboreum 0

0

0 0

-1.0~------~------~------~------r -1.0 -0.5 0.0 0.5 1.0 Principal Component 1 Fig. 1. Principle component analysis (PCA) of 314 A-genome cottons; plot of individual accessions on first two PCA axes. PCA data shown in Table 3. 524 CROP SCIENCE, VOL. 34, MARCH-APRIL 1994

1·0 _,------;A~Ce_n_tra_I_As_la---,<:------, 6 Eastern Asia + Mid East C\1 <\? Southeasterp Asia .... 0.5 c: <:- ~ Q) <:- 8. E 8 ..... a; a. 0 ~ -0.5 Q.

-1.0 -0.5 0.0 0.5 1.0 Principal Component 1 Fig. 2. Principle component analysis (PCA) of 169 G. arboreum accessions; plot of individual accessions on first two PCA axes. PCA data shown in Table 3. Accessions are designated by region of origin. Seventy-four accessions with unknown origins are not shown.

sinus, lobe shape, boll shape) (Fryxell, 1979; Hutchinson Africa reflect this same mid-range (Fig. 3b). One of the and Ghose, 1937). Other characters contributing to outlier points was an accession listed as being from a differences between the species were rnicronaire and lint research group in Zimbabwe, but it is also identified in percent. the collection as kuljianum, a western China race. Middle For G. arboreum, the first three components explained Eastern accessions are clustered above the means for 30% of the variation (Table 3). Characters that appeared PC 1 and PC 2. Only one of these accessions had a important to the first component were fiber (rnicronaire, racial designation (persicum). Most of the other persicum lint percent), leaf (sinus, lobe shape), and bract (length, accessions were scattered around the mean of PC 2 rather petal length/bract length) traits. Plots of PC 1 and PC 2 than above it (Fig. 3a). Accessions of race wightianum revealed little geographical alliance, except accessions clustered in the lower right. from eastern Asia tended to fall below the mean for PC 2 (Fig. 2). Accessions from southeastern Asia were DISCUSSION distributed throughout the plot, perhaps reflecting the variation in this group of accessions from the purported At the initiation of this study, the NPGS Asiatic Cotton geographic center of diversity (Wendel et al., 1989). Germplasm collection contained 254 accessions of G. Too few of the accessions had racial designations for a arboreum and 138 accessions of G. herbaceum. The robust test of racial validity, and natural infraspecific majority of these were originally obtained from other groups were not suggested by the data; rather, accessions germplasm banks. For most accessions, only some of of G. arboreum are represented by a fairly even disper­ the following information was known, originating germ­ sion of points. plasm bank, collector and collection number, country The first three PCs explained 41 % of the variation of origin, year of collection, racial classification, and among the G. herbaceum accessions (Table 3). In the common name. Errors in documentation were evident first component, the factors with the highest loading in the original passport data of some of the accessions (coefficients >0.50) were all positive, indicating that size donated to the collection. We have noted these errors was an important factor (Pimental, 1979). For example, and, to the extent possible, made corrections in the boll length and width, but not the ratio of boll width to collection documentation in GRIN. boll length, had high factor loadings; thus, absolute Morphological analyses demonstrated that G. arbor­ size but not shape was a major variable. In the second eum and G. herbaceum differ significantly in 41 of the component, most factors with the high loadings were 53 traits measured, although not a single character serves ratios of characters, indicating the importance of shape. to completely distinguish the two. Several of these char­ Variability within G. herbaceum appears to be more acters have previously been used to distinguish between organized within the collection. Of the races identified the two species (bract: width/length, tooth length, and among the accessions examined, or judged to fit racial #teeth) (Hutchinson et al., 1947; Fryxell, 1979). Other descriptions, G. herbaceum subsp. africanum accessions characters, including leaf sinus, number of gossypol were near the means for PC 1 and PC 2 (Fig. 3a). With glands, and sunred, which varied significantly between one exception, the accessions obtained from southern species, have been mentioned as differences between the STANTON ET AL.: MORPHOLOGICAL DNERSITY IN THE A-GENOME COTTONS 525

1·0 .r-=.~a_fri_c_a_n_u_m______A____, A kuljlanum 6 perslcum C\1 + wlghtlanum ....c: 0.5 4) c: 8. + u\ + 0.0 +------+.,----':•0+-A--::;;++:------f'..------j ~ 6 as .t. 0. + + 0 + ~ -0.5 + D.. +

-1.0

1.0 o Southern Africa B A Central Asia 6 Eastern Asia C\1 +Mid East .... 0.5 + + c: "(} Southeastern AsiB.t + 4) ol}o + $ c: 6 0 .t. 8. .t. 0 .t. + " 6 0.0 - "~\+ "' " " .t.+ -- .t. .t. ~ 6.t. •• .t. .t. .t. + as 6 0. .t. 0 t:. •" " " c: " 't: -0.5 D.. ol}o "

-1.0 -1.0 -0.5 0.0 0.5 1.0 Principal Component 1 Fig. 3. Plot of individual accessions of G. herbaceum on first two principle component axes. Principle component analysis (PCA) data shown in Table 3. (A) Plot of 26 accessions designated by racial classification. Forty-four accessions without racial designations are not shown. (B) Plot of 58 accessions designated by origin. Fourteen accessions with unknown region of origin are not shown. species (Harland, 1932; Hutchinson and Ghose, 1937; Both high levels of infraspecific variability and the Silow, 1944; Hutchinson, 1954). Their separation into overall resemblance of the two species contributed to two groups by Zaitzev (in Fryxell, 1979) has been sup­ the absence of complete morphological distinction; thus, ported by other studies showing genetic and cytogenetic a combination of characters is necessary to distinguish differentiation (Gerstel, 1953; Phillips, 1961; Silow, between G. arboreum and G. herbaceum. Though taxo­ 1944; Stephens, 1950; Gennuretal., 1988; Wendeletal., nomic keys have not used foliar traits to distinguish these 1989). In this evaluation, significantly different means for species, two leaf characters (lobe shape, sinus depth) over three-quarters of the evaluated characters and the strongly influenced the separation of the two species in multivariate separation of accessions into two overlap­ this analysis. Hutchinson and Ghose (1937) maintained ping clusters (Fig. 1) further support the separation of that, because a Mendelian multiple-allelic series affected these species. The overlap of species distribution in the leaf laciniation, leaf morphology was not a useful taxo­ PC plot may be due to overlap in character ranges nomic tool in A-genome cottons; yet in this study several and probable introgression between the species during foliar traits appeared to exhibit consistent, though subtle, centuries of sympatric cultivation (Silow, 1944; Wendel differences between species. In general, G. arboreum et al., 1989). has longer leaves with narrower lobes and deeper sinuses 526 CROP SCIENCE, VOL. 34, MARCH-APRIL 1994

than G. herbaceum. Other statistically significant traits, is reasonably well represented; however, accessions from not mentioned in any keys, that distinguish G. arboreum this region were distributed throughout the PC plot. This from G. herbaceum include the ratio of petal length to may reflect the variation expected from the purported bract length, number of glands, nectary size, and seed geographic center of diversity (Wendel et al., 1989). size (naked seed wt. and seed index). Species means Knight (1954, in Endrizzi et al., 1985) published a indicate that G. herbaceum tends to have less anthocya­ list of gene symbols and linkage groups for the A-genome nin, finer lint, and smaller than G. arboreum. cottons representing both common and rare alleles (Si­ Though both species have nectariless accessions, acces­ low, 1941). Characteristics not present in the NPGS col­ sions with leaf nectaries but without involucellar nectaries lection include lintless (AI), virescent (AI and A2), single were found only in G. herbaceum. lobed leaf (A2), brown lint (AI), and some anthocyanin In contrast to the separation of G. arboreum and G. forms (AI and A2). In addition to qualitative traits, we herbaceum based on morphology, analysis of infraspe­ found also that some morphological types have limited cific variation patterns led to few recognizable groupings representation. These observations indicate that the col­ within either species (Fig. 2 and 3). These results contra­ lection has an incomplete representation of the diversity dicted expectations based on descriptions of geographic that was once available. races in the literature (Hutchinson, 1950; Hutchinson Efforts should be made to collect remnant indigenous and Ghose, 1937; Silow, 1944). Most of the difficulty seed from portions of the ranges that once were described with the original racial classifications stems from their as comprising races and to obtain better representation definitions. Rather than being based on overall patterns within the collection of G. herbaceum subsp. africanum of morphological resemblance, racial classifications in and races acerifolium and kuljianum, and of G. arboreum both species were primarily rooted in geographical from Africa and the Middle East region. Additional ranges. To some extent, the artificiality of these systems accessions from these geographic regions and races could was apparent to the original authors (Hutchinson, 1959; contribute additional diversity to the germplasm collec­ Hutchinson and Ghose, 1937; Silow, 1944) who noted tion. that many geographic regions contained extensive varia­ In spite of the limitations of the A-genome cotton tion. Some accessions within each species could be as­ collection, it is a potential source of genetic diversity signed to a specific race, especially within G. herbaceum for G. hirsutum. Introgression from several species into (wightianum, africanum, and kuljianum). However, G. hirsutum have successfully contributed genetic vari­ most accessions in G. arboreum exhibited characters ance for quantitative traits such as stress tolerance, fiber that could result in assignment to more than one race. quality, and yield (Meredith, 1991). Hybrids of G. hirsu­ Classification difficulties of the accessions may be com­ tum x G. arboreum have led to selections with earlier pounded by inadequate or misleading passport data for maturity and an increased range of fiber traits (Cooper, geographic origin and race, lack of samples representing 1969; Wange et al., 1989). With modern techniques all races, and genetic shifts within the collection associ­ interspecific introgression of A-genome traits into AD­ ated with storage and seed increases. genome cottons may be facilitated by constructing syn­ The study, maintenance, and use of the germplasm thetic AD allotetraploids or by crossing with ADD hexa­ has essentially ceased as the A-genome cottons have ploids (Stewart, 1992). been replaced by G. hirsutum in most of their former Morphological characteristics of accessions evaluated cultivated range. It is of interest to ascertain what propor­ in this study will be incorporated into the GRIN data­ tion of this formerly wide-ranging and morphologically base. Additionally, we have evaluated much of the NPGS diverse genepool has been captured in the NPGS system. Asiatic cotton collection for resistance to a number of While much variability exists within the collection, some pests (Stanton et al., 1987, 1988, 1990, 1992, 1993). morphological types and geographic races are repre­ Results from these evaluations, as well as observations sented by only a few accessions. While the race designa­ on complex characters, such as earliness, and simply tions in G. herbaceum appear to have some merit, the inherited mutants, such as fiberless, glabrous, and nec­ diversity or validity of race acerifolium cannot be ad­ tariless, will also be documented in GRIN. dressed because this race is not represented in the collec­ tion. Fewer than 10 accessions of G. herbaceum subsp. ACKNOWLEDGMENTS africanum are in the collection. Isozymic and morpholog­ We thank as a group the many students who contributed ical evidence indicates that subsp. africanum bears the technical assistance to this project. Jeff Velie is acknowledged most resemblance to a theoretical wild ancestor of G. for generation of the graphics. Parts of the research were herbaceum (Hutchinson, 1959; Wendel et al., 1989). Its supported by USDA grants 86-CSRS-2-2879 and 90-34103- position in the morphological middle of the plot lends 4985 and grant 88-B-0032 from the Arkansas Science and further support to this notion. As africanum is the most Technology Authority, all of which we gratefully acknowledge. morphologically primitive and isozymically diverse form of G. herbaceum, further collections are warranted. The REFERENCES western Chinese race kuljianum, an unusually early­ Benedict, J.H., M.F. Treacy, D.W. Altman, and K.M. Schmidt. 1987. Preference of boll weevils and tobacco budworms for five fruiting G. herbaceum, is represented by a single acces­ species of Gossypium. p. 92. In J.M. Brown (ed.) Proc. Beltwide sion with two morphological types. No representatives Cotton Prod. Res. Conf., Dallas, TX. 4-8 Jan. 1987. Natl. Cotton of G. arboreum are listed for Africa, and few are from Council of America, Memphis, TN. the Middle East. On the other hand, southeastern Asia Bollenbacher, K., and N.D. Fulton. 1971. Susceptibilily of Gossypium STANTON ET AL.: MORPHOLOGICAL DNERSITY IN THE A-GENOME COTTONS 527

species and varieties to seedling anthracnose. Plant Dis. Rptr. 55: biological data. Kendal/Hunt Pub., Dubuque, lA. 879-882. Regional Cotton Variety Tests, 1991. Regional Cotton Variety Tests, Carter, W.W. 1981. Resistance and resistant reaction of Gossypium 1990, yield, boll, seed, spinning, and fiber data. National Cotton arboreum to the reniform nematode, Rotylenchulus reniformis. J. Variety Testing Program, USDA-ARS, Stoneville, MS. Nematol. 13:368-374. Rohlf, F.J. 1992. NTSYS-pc: Numerical and multivariate Cooper, D. T. 1969. Utilization of germplasm for Gossypium arboreum analysis system. Version 1. 7. Applied Biostatistics, Inc., Setauket, L. x G. raimondii Ulb. to supplement the genetic variability in NY. upland cotton. N. Carolina State Univ., Raleigh (Diss. Abstr. 69- SAS. 1989. SAS/STAT User's Guide, Version 6, Fourth Ed, Vol. 16829). 2. SAS Institute Inc. 1989. Cary, NC. Doebley, J. 1989. Isozymic evidence and the evolution of crop plants. Silow, R.A. 1941. The comparative genetics of Gossypium anomalum p. 165-191. In D.E. Soltis and P.S. Soltis (ed.) lsozymes in plant and the cultivated Asiatic cottons. J. Genet. 42:259-358. biology. Dioscorides Press, Portland, OR. Silow, R.A. 1944. The genetics of species development in the Old Endrizzi, J.E., E.L. Turcotte, and R.J. Kobel. 1985. Genetics cytol­ World cottons. J. Genet. 46:62-77. ogy and evolution of Gossypium. Adv. Genet. 23:271-375. Stanton, M.A., J.R. Phillips, and J.McD. Stewart. 1990. Evaluation Fryxell, P.A. 1979. The Natural History of the Cotton Tribe. Texas of the Asiatic cottons, Gossypium arboreum L. and G. herbaceum A&M Univ. Press, College Station, TX. L., to Heliothis virescens F. p. 206. In J.M. Brown (ed.) Beltwide Fryxell, P.A. 1984. Taxonomy and germplasm resources. p. 27-58. Cotton Prod. Res. Conf., Las Vegas, NV. 9-14 Jan. 1990. Natl. In R.J. Kobel and C.F. Lewis (ed.) Cotton. Agron. Monogr. 24. Cotton Council of America, Memphis, TN. ASA, Madison, WI. Stanton, M.A., R.D. Riggs, and J.McD. Stewart. 1987. Resistance Gennur, M.N., A.F. Habib, S.N. Kadapa, and J.V. Goud. 1988. of Asiatic cottons to the root-knot nematode, Meloidogyne incog­ Karyomorphological studies in Asiatic cottons. ll. Karyotypic anal­ nita. J. Nematol. 19:558 (Abstr). ysis of species and races of Asiatic cottons based on nucleolar Stanton, M.A., J.McD. Stewart, and J.R. Phillips. 1988. Resistance chromosomes and symmetry of karyotype. Cytologia 53: 107-114. of Asiatic cottons to bollworm. p. 100. In J.M. Brown (ed.) Proc. Gerstel, D.O. 1953. Chromosome translocations in interspecific hy­ Beltwide Cotton Prod. Res. Conf., New Orleans, LA. 3-8 Jan. brids of the genus Gossypium. Evolution 7:234-244. 1988. Natl. Cotton Council of America, Memphis, TN. GRIN. 1989. Genetic resources information network users manual. Stanton, M.A., C.S. Rothrock, and J.McD. Stewart. 1993. Response USDA ARS PSI NGRL, Beltsville, MD. of A-genome cotton germplasm to the seedling disease pathogens, Harland, S.C. 1932. The genetics of Gossypium. Bibliographia Genet­ Rhizoctonia solani and Pythium ultimum. Genet. Resour. Crop ica 9:107-182. Evol. 40: (in press) Hutchinson, J.B. 1950. A note on some geographical races of Asiatic Stanton, M.A., J.McD. Stewart, and N.P. Tugwell. 1992. Evaluation cottons. Emp. Cotton Grow. Rev. 27:123-127. of the Asiatic cottons for resistance to thrips. Genet. Resour. Crop Hutchinson, J.B. 1954. New evidence on the origin of the Old World Evol. 39:89-95. cottons. Heredity 8:225-241. Stephens, S.G. 1950. The internal mechanism of speciation in Gossyp­ Hutchinson, J.B. 1959. The application of genetics to cotton improve­ ium. Bot. Rev. 16:115-149. ment. Cambridge Univ. Press, London. Stewart, J .McD. 1992. Germplasm resources and enhancement strate­ Hutchinson, J.B., and R.L.M. Ghose. 1937. The classification of the gies for disease resistance. p. 1323-1325. In D.J. Herber and cottons of Africa and Asia. Ind. J. Agric. Sci. 7:233. D.A. Richter (ed.) Proc. Beltwide Cotton Conf., Nashville, TN. Hutchinson, J.B., R.A. Silow, and S.G. Stephens. 1947. The Evolu­ 6-10 Jan. 1992. Natl. Cotton Council of America, Memphis, TN. tion of Gossypium. Oxford Univ. Press, London. Stewart, J.McD., and C.L. Hsu. 1978. Hybridization of diploid and Jackson, J.E. 1991. A user's guide to principal components. John tetraploid cottons through in-ovulo embryo culture. J. Heredity Wiley and Sons, Inc., New York. 69:404-408. Lee, J.A. 1984. Cotton as a world crop. p. 1-27. In R.J. Kobel and Vollesen, K. 1987. The native species of Gossypium () in C.F. Lewis (ed.) Cotton. Agron. Monogr. 24. ASA, Madison, Africa, Arabia, and Pakistan. Kew Bull. 42:337-349. WI. Wange, D.J., G.P. Li, and B.L. Li. 1989. Study on distant hybridiza­ Mathre, D.E., and J.D. Otta. 1967. Sources of resistance in the genus tion of cotton and character transfer through backcrossing. China Gossypium to several soilborne pathogens. Plant Dis. Rptr. 51: Cottons 3:6-7. 864-867. Wendel, J.F., C.L. Brubaker, and A.E. Percival. 1992. Genetic Meredith, W.R. Jr. 1991. Contributions of introductions to cotton diversity in and the origin of upland cotton. improvement. p. 127-146. In H.L. Shands and L.E. Wiesner (ed.) Am. J. Bot. 79:1291-1310. Use of plant introductions in cultivar development, Part I. CSSA Wendel, J.F., P.O. Olson, and J.McD. Stewart. 1989. Genetic diver­ Spec. Publ. 17. CSSA, Madison, WI. sity, introgression, and independent domestication of Old World Percival, A. E. 1987. The national collection of Gossypium germplasm. cultivated cottons. Am. J. Bot. 76:1795-1806. USDA Southern Crops Ser. Bull. 321. Dept. Agric. Commun., Wiley, E.O. 1981. Phylogenetics: The theory and practice of phyloge­ College Station, TX. netic systematics. John Wiley and Sons, New York. Phillips, L.L. 1961. The cytogenetics of speciation in Asiatic cotton. Yik, C.P., and W. Birchfield. 1984. Resistant germplasm in Gossyp­ Genetics 46:77-83. ium species and related plants to Rotylenchulus reniformis. J. Pimental, R.A. 1979. Morphometries: The multivariate analysis of Nematol. 16:146-153.