The Systematic Significance of Seed Morphology in () Under Scanning Electron Microscopy Author(s): James F. Matthews and Patricia A. Levins Source: Systematic Botany, Vol. 11, No. 2 (Apr. - Jun., 1986), pp. 302-308 Published by: American Society of Taxonomists Stable URL: http://www.jstor.org/stable/2419120 . Accessed: 09/05/2014 05:55

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

.

American Society of Plant Taxonomists is collaborating with JSTOR to digitize, preserve and extend access to Systematic Botany.

http://www.jstor.org

This content downloaded from 83.204.224.234 on Fri, 9 May 2014 05:55:35 AM All use subject to JSTOR Terms and Conditions SystematicBotany (1986), 11(2): pp. 302-308 ? Copyright1986 by the American Society of Plant Taxonomists

The SystematicSignificance of Seed Morphology in Portulaca (Portulacaceae) under Scanning Electron Microscopy

JAMES F. MATTHEWS and PATRICIA A. LEVINS Department of Biology, University of North Carolina, Charlotte, North Carolina 28223

ABSTRACT. Seeds of seven southeasternand one southwesternUnited States species of Portulaca were examined with the scanning electron microscope.Some species could be characterizedby the external seed morphology while others could not. Those species with the largest geographical ranges and ecological diversities showed the greatestvariation in morphology. Comparisons be- tween species oftenshowed significantoverlap of morphological characteristics.Seed morphology in Portulacacan be used as a taxonomic trait,but should be combined with other traits for the diagnostic determinationof species.

Since Davis and Heywood (1963) proposed 2. A distinctiveseed coat morphology usually seed morphology as a potential taxonomic allowing identificationof species solely on character,an increasingnumber of papers have the basis of seed characters(Clark and Jern- reportedon the scanning electron microscopy stedt 1978). of seed surface patterns.Brisson and Peterson 3. Both inter-and intrapopulationalvariation; (1976, 1977) provide a complete bibliography some species monomorphicand some poly- through 1977. morphic with small populations tending to In general, the findingsuntil now have sup- be monomorphicand widespread taxa poly- ported the following groupingsregarding seed morphic (Hauptli et al. 1978). surfacetaxonomic implications. 4. Seed morphologybeing subject to the same uncertaintiesand ambiguities as any other 1. Intrageneric consistency complete, so that morphological attributeused in taxonomic seed surfacecharacters are predictable from studies (Newell and Hymowitz 1978). species to species (Chuan and Heckard 1972; 5. Monomorphic seeds with variation in those Hill 1976; Clark and Jernstedt1978; Crow species with the largest distribution(Canne 1979). 1979). 2. Intragenericconsistency high, some species 6. External seed morphology distinctive and variability (Newell and Hymowitz 1978; valuable for species distinctions (Wofford Canne 1979, 1980; Wofford1981; Elisens and 1981). Tomb 1983; Chance and Bacon 1984). 7. High levels of infraspecificvariation sug- 3. Intragenericvariability of some species, re- gesting caution in using SEM-level features lated to ecological conditions (Hauptli et al. to characterizespecies or taxa (Wyatt 1984). 1978). 8. Species specificseed coat morphologies that 4. Intraspecificvariability, showing different do not correspond well with the more clas- expressions over geographical range and sical treatmentsand suggest a restructuring ecological conditions (Wyatt 1984; Mat- of the intragenericand even the generic tax- thews and Levins 1985a). onomy (Chance and Bacon 1984). 5. Intraspecificconsistency, involving subspe- 9. Variability in seed surface patterns occurs cies, related to ploidy level (Danin et al. within geographic regions and within the 1978). species (Matthews and Levins 1985a). These varied findings have resulted in the A recentstudy (Matthews and Levins 1985a) following conclusions, listed in historical se- concentratedon the variation in seed surface quence. morphology of Portulacapilosa over its south- 1. SEM photographs have not provided suffi- eastern and southwesternUnited States range. cient evidence to sustain a taxonomic inno- The present report will analyze the six other vation not held before(Brisson and Peterson species occurring in the southeastern United 1976). States,which constitutethe seven southeastern 302

This content downloaded from 83.204.224.234 on Fri, 9 May 2014 05:55:35 AM All use subject to JSTOR Terms and Conditions 1986] MATTHEWS & LEVINS: PORTULACA 303

species of Portulaca (Matthews and Levins TABLE 1. Voucher data for representativespeci- 1985b), plus one species fromthe southwest. mens of Portulacaused for SEM studies.

MATERIALS AND METHODS Fig. 1. P. oleracea L. Alabama: Mobile Co., 26 Aug 1965, DeramusD732 (ALU). Seeds were examined fromdried specimens Fig. 2. P. oleraceaL. Louisiana: Rapides Parish, 19 Jul fromthe followingherbaria: ALU, ASTC, DUR, 1979, Pias 4345 (NLU). FLAS, FSU, G, GA, JSU, KNK, KY, LAF, LL, Fig. 3. P. pilosa L. Florida: Citrus Co., 13 Aug 1958, MO, NCU, NLU, NO, NY, SMS, SMU, TENN, Kral 7825 (NY). TEX, UARK, UNA, UNCC, US, USAM, USCH, Fig. 4. P. pilosa L. Arkansas: White Co., 14 Oct 1974, USF, VDB, VPI, VSC, and WILLI. Representa- Demarees.n. (GA). Fig. 5. P. pilosa L. Oklahoma: Cimarron Co., 4 Aug tive specimens from which seeds were taken 1977, Taylor25255 (NLU). for microscopy are listed in table 1. For the Fig. 6. P. pilosa L. Texas: BrewsterCo., 23 Aug 1970, electronmicroscope, seeds were sputtercoated Semple408 (MO). with gold-palladium on a Hummer V and Fig. 7. P. umbraticolaKunth. South Carolina: Lancas- viewed at 100 x with a JeolJSM-35CF scanning terCo., 25 Sep 1973, Hatleyand Sowerss.n. (UNCC). electron microscope. Fig. 8. P. rubricaulisKunth. Florida: Lee Co., 21 Jul 1954, Cooley2316 (G). RESULTS AND DISCUSSION Fig. 9. P. retusaEngelm. Texas: Williamson Co., 16 Aug 1961, Walker73 (TEX). Seeds were examined with a 10 x handlens Fig. 10. P. smalliiP. Wilson. Georgia: Hancock Co., to determinewhether the surface was smooth 22 Oct 1952, Duncan 14481 (VSC). or rough. At 25 x, tubercles could be seen on Fig. 11. P. amilisSpegaz. Georgia: Echols Co., 23 Aug the dorsal spine and the lateral surfacescould 1961, Faircloth4786 (VSC). be determined to be smooth or rough, the tu- Fig. 12. P. grandifloraHooker. South Carolina: Ker- bercles producing the rough surface.However, shaw Co., 6 Oct 1957, Radford30004 (NCU). the relation of the tubercles to the remainder of the surface cells was not easily detected. In addition, the seeds of some species were cov- ered with an iridescentcoating which oftenob- 1, 2) fromthe southeast show that the surfaces scured the surfacepatterns. To betterinterpret vary fromnearly smooth to very granulate.Tu- the primary and secondary structuresof the bercles on the dorsal spines are minute or in- seeds, they were examined at 100 x with an distinct.While Legrand (1962) questioned the SEM. use of seed surfacetexture as a characterto de- Using the classificationof Barthlott(1984) for lineate varietiesof P. oleracea,Danin et al. (1978) the microstructuralfeatures of the seed sur- used seed size and seed coat morphology to faces of Portulaca,the primarystructure of the describe nine subspecies. Cytotaxonomiccom- anticlinal walls of the isodiametric seed coat parisons of these subspecies showed that each cells is S-undulate (in earlier reports,Matthews had a fairlydistinct distribution which may in- and Levins 1985a and 1985b,the term"stellate" dicate theirspecialization to certainclimatic and was used to describe the S-undulate pattern). edaphic conditions. Although there was some The secondarystructure exhibited by the seeds subspecificvariation in seed coat morphology, is produced by channelled anticlinalwalls, with seed size and/or ploidy level were used to as- periclinal walls varying fromflat (smooth sur- sist in the final identification.In our examina- faces) to convex, to papillate, to tuberculate. tion of P. oleracea,no attemptwas made to de- . The most widespread terminehow many of theirsubspecies could be species of Portulacain the southeast,as well as identified in the southeastern United States, in the world, is P. oleracea L., a species with only thatvariation within P. oleraceawas clear- yellow flowers,flat and sparse pubes- ly evident.Our figure1 is theirP. oleraceasubsp. cence. Legrand (1962) reported that the sur- nitida,figure 2 is their subsp. granulato-stellula- faces of P. oleraceaseeds varied widely. Danin ta. et al. (1978) demonstrated a correlation be- Portulacapilosa. Portulacapilosa L., the sec- tween ploidy level and seed coat morphology ond most common species in the southeast,ex- of P. oleracea.Seed surfaces of this taxon (figs. hibits a wide variation in seed surfaceand tex-

This content downloaded from 83.204.224.234 on Fri, 9 May 2014 05:55:35 AM All use subject to JSTOR Terms and Conditions 304 SYSTEMATIC BOTANY [Volume 11

4~~~~3

-A-4~~~~~~~~~

I~~~~~~~~~~~~~~- in~~I 4;-

FIGS. 1-6. Scanning electron micrographsof seeds of Portulaca.Scale = 100 ,um. 1. S-undulate surface pattern of P. oleracea. 2. Granulate pattern of P. oleracea. 3. S-undulate pattern of P. pilosa with short tubercles. 4. S-undulate pattern of P. pilosa with rounded tubercles. 5. S-undulate-smoothpattern of P. pilosa. 6. S-undulate-tuberculatepattern of P. pilosa.

This content downloaded from 83.204.224.234 on Fri, 9 May 2014 05:55:35 AM All use subject to JSTOR Terms and Conditions 1986] MATTHEWS & LEVINS: PORTULACA 305

ture. Portulacapilosa has red to purple flowers, Kunth (fig. 7) is easily identifiedby its yellow terete leaves, and is densely pubescent. Nu- to orange flowers, flat leaves, sparse pubes- merousplants were examined fromFlorida, the cence, and a membranaceouswing on the cap- Gulf Coast and westward into Texas, Okla- sule. However, the seeds of P. umbraticolamore homa, Arkansas,and Missouri. The detailed re- closely resemble the tuberculatespecimens of sults of this study have been reported else- P. pilosa (fig. 6) although the tubercles are not where (Matthews and Levins 1985a), and are quite as stronglydeveloped. brieflysummarized here. Portulacarubricaulis. The flattenedsurfaces In Florida, P. pilosaseeds have smooth to pa- of the seeds of the yellow- to orange-flowered, pillate lateral surface patterns and secondary terete-leaved,pubescent P. rubricaulisKunth (fig. tubercleson the dorsal spine (fig. 3). The Gulf 8) approximatethose of P. pilosa fromwestern Coast examples are very similar to those of Oklahoma (fig.5). However, matureP. rubricau- Florida, but may have a reduction in tubercles lis seeds are gray or brown while those of P. on the dorsal spine. Specimens fromnorthern pilosaare black. Texas, Arkansas (fig. 4), Oklahoma (fig. 5), and Portulacaretusa. Portulacaretusa Engelm. is Missouri exhibit fewer dorsal tubercles but re- closely related to P. oleracea(e.g., yellow flow- tain the smooth to papillate lateral surface. In ers,flat leaves, and sparse pubescence), but dif- west Texas, an area of diverse habitats,surface fers fromit by retuse leaves and a highly tu- texturevaries from very tuberculate on both berculate seed coat. The distribution for P. the spine and lateral surfaces (fig. 6) to few retusa,derived from the specimens we exam- dorsal spines and lateral tubercles (figs. 4 and ined, is limited to central and west Texas, and 5). We have interpretedthese variations as re- Oklahoma. Steyermark(1963) listsP. retusafrom sponses correlated with ecological conditions, Missouri,but we saw no specimens fromthere. with a reduction in tubercles in dry habitats. It has been assigned to Arkansas (Steyermark Geesink (1969), in a study of the portulacas 1963; Correll and Johnston 1970), but Smith of Indo-Australia and the Pacific, commented (1978) questions its presence, based on lack of on the variabilityof P. pilosa,characterizing it documentation.A comparison of the seeds of as a very complex species into which he syn- P. retusa (fig. 9) with the highly tuberculate onymized about 60 names. However, he did seeds of P. pilosafrom the Big Bend and south- recognize eight subspecies which probably central Texas (fig. 6) shows a strikingsimilari- were originally geographically separate, but ty. We are not suggesting a relationship be- now through the activity of man are inter- tween two species with very differentflower mixed. For the American taxa, he stated that colors, shapes, and pubescence, but merely there were some 150 names recognized by von a similarityin seed coat morphology.Danin et Poellnitz (1934) and Legrand (1962) thatshould al. (1978) in their characterizationof nine sub- be subordinated to the P. pilosa complex. species of P. oleracea,described one, subsp. im- From these conclusions regarding P. pilosa, polita(Type: Texas: HartleyCo., sandy soil along the characterizationof seeds from specimens Punta de Agua Creek, between Romero and thatdo not representthe totalrange of a species Middle Water,9.X.1964, Correll 30330, holotype may provide misleading conceptsof the species UC; isotype, GH) that has a seed coat mor- by giving improperweight to only one type of phology very similar to that of P. retusa.The seed morphology. This is not to say that seed range of P. oleracea subsp. impolitagiven by coat morphologyis not a good taxonomicchar- Danin et al. overlaps our observed Texas dis- acter,but ratherthat it should be considered in tributionof P. retusa.A study of P. oleraceaand combination with other morphological traits. P. retusashould be made to determinewhether Unfortunately,in some families and genera, the taxonnow called P. retusais a distinctspecies particularlythe succulent ones which do not or falls within the range of P. oleracea. make good dried specimens, there is a tenden- Portulacasmallii. Portulacasmallii P. Wilson cy to rely on a single feature,such as a hard (fig. 10) is a red-flowered,terete-leaved, pubes- seed coat, as a primarycharacter for identifi- cent endemic of the granitic outcrops of the cation.The followingcomparisons illustrate the southeast. Seeds of P. smalliiand P. pilosawere potential problems by pairing similar seeds of similar in morphology (fig. 3). Portulacasmallii differentspecies of the southeasternportulacas. is separatedfrom P. pilosaby its ecological niche Portulaca umbraticola. Portulaca umbraticola and consistentlylarge (0.6-0.85 mm) elongate

This content downloaded from 83.204.224.234 on Fri, 9 May 2014 05:55:35 AM All use subject to JSTOR Terms and Conditions 306 SYSTEMATIC BOTANY [Volume 11

A9P-.

C V

4P~~~~~~~~~~~~~~4

FIGS. 7-12. Scanning electron micrographsof seeds of Portulaca.Scale = 100 Am. 7. S-undulate-tuber- culate patternof P. umbraticola. 8. LightlyS-undulate-smooth pattern of P. rubricaulis. 9. S-undulate-tuber- culate pattern of P. retusa. 10. S-undulate-tuberculatepattern of P. smallii. 11. S-undulate pattern of P. amilis. 12. LightlyS-undulate-dorsally tuberculate pattern of P. grandiflora.

This content downloaded from 83.204.224.234 on Fri, 9 May 2014 05:55:35 AM All use subject to JSTOR Terms and Conditions 1986] MATTHEWS & LEVINS: PORTULACA 307

seed. All of the graniticoutcrop specimens la- creasinglymore accurate when the data avail- beled P. smalliihad seeds largerin size than any able include the leaf morphology,vestiture of of the P. pilosaseeds that were measured. the stem and inflorescence,flower color, cap- Portulacaamilis. The small seeds of P. amilis sule structure,and seed morphology.Decisions Speg. have very smooth lateral surfaces and are less accurate as these characters are less no tubercles (fig. 11). Over its southeastern available. Workingwith difficulttaxa like Por- geographical range very little seed surface tulaca is easier when herbarium label data in- variation is found. The identificationof this clude some of the more transientfeatures, such red-flowered,flat-leaved, pubescent, newly in- as flowercolor, or commentson the ecological troduced southeasternspecies (Judd and Wun- conditions of the habitat. derlin 1981) and the hypothesis of its intro- ACKNOWLEDGMENTS. We wish to thankSandra F. duction and spread (Matthews and Levins Zane for the electron microscopy and T. L. Melli- 1985b) suggest a reason for the lack of vari- champ,T. L. Reynolds,and J.R. Massey forassistance ability.Being introducedinto the United States in the preparationof the manuscript. within the past 50 years fromSouth America, probablyfrom only a few seeds, the southeast- LITERATURE CITED ern gene pool of P. amilismay be small. Portulacagrandiflora. The final species is P. BARTHLOTT, W. 1984. Microstructuralfeatures of seed grandifloraHook., the commonrose moss of rock surfaces. Pp. 95-105 in SystematicsAssociation gardens, with variable flower colors, terete Special Vol. No. 25, Currentconcepts in planttax- leaves, and pubescence. Limited herbarium onomy,eds. V. H. Heywood and D. M. Moore. specimens of this species, as escapes fromcul- London and Orlando: Academic Press. BRISSON, J. D. and R. L. PETERSON. A tivation, were available for study. The large 1976. critical review of the use of scanning electron micros- seeds (0.75-1 mm) are smooth,except fordorsal copy in the study of seed coat. Ill. Inst. Techn. tubercles(fig. 12). Our limited sample may not Res. Inst./SEM/1976/II:477-495. representthe total range of variabilityin this and . 1977. The scanning electronmi- widely cultivated species, so comparison with croscope and X-raymicroanalysis in the studyof the native and naturalized noncultivated seeds: A bibliography of the period 1967-1976. species is not helpful. Ill. Inst. Techn. Res. Inst./SEM/1977/II:697-712. CANNE, J. M. 1979. A light and scanning electron CONCLUSION microscopestudy of seed morphologyin Agalinis (Scrophulariaceae) and its taxonomic signifi- As shown in the introductionto this study, cance. Syst. Bot. 4:281-296. there are indications that variability in seed . 1980. Seed surface features in Aureolaria, surfacemorphology may be linked to diversity Brachystigma,Tomanthera, and certain South in geography,ecology, and ploidy level. In Por- American Agalinis(Scrophulariaceae). Syst. Bot. tulaca,it may be that all three factorsaffect the 5:241-252. CHANCE, G. and BACON. expressionof the surfacemorphology. Seed coat D. J. D. 1984. Systematic implications of seed coat morphology in Nama morphologyis a usefulcharacter, one thatneeds (Hydrophyllaceae). Amer. J. Bot. 71:829-842. to be included as of the of the part description CHUAN, T. I. and L. R. HECKARD. 1972. Seed coat species. However, it does not appear to be di- morphology in Cordylanthus(Scrophulariaceae) agnostic foreach species, so must be combined and its taxonomic significance.Amer. J. Bot. 59: with other taxonomic charactersin determin- 258-265. ing the systematicrelationships within the ge- CLARK, C. and J. A. JERNSTEDT. 1978. Systematic nus. studiesof Eschscholzia(Papaveraceae). II. Seed coat It is unfortuatethat seed coat morphologyis microsculpturing.Syst. Bot. 3:386-402. not reliable at the species level, since deter- CORRELL, D. C. and M. C. JOHNSTON. 1970. Manual minations are often difficultfrom herbarium ofthe vascular of Texas.Renner, Texas: Tex- as Research Foundation. specimens. However when other features are CROW, G. E. 1979. The systematicsignificance of this has been often with mixed lacking, used, seed morphology in Sagina (Caryophyllaceae) results.With some familiarityof Portulacaand under scanningelectron microscopy. Brittonia 31: the morphological traitsthat assist in identifi- 52-63. cation of species, the seed coat can be helpful. DANIN, A., I. BAKER, and H. G. BAKER. 1978. Cyto- In Portulaca,taxonomic decisions become in- geography and of the Portulacaolera-

This content downloaded from 83.204.224.234 on Fri, 9 May 2014 05:55:35 AM All use subject to JSTOR Terms and Conditions 308 SYSTEMATIC BOTANY [Volume 11

cea L. polyploid complex. Israel J. Bot. 27:177- MATTHEWS, J. F. and P. A. LEVINS. 1985a. Portulaca 211. pilosa L., P. mundulaJohnst., and P. parvulaGray DAVIS, P. H. and V. H. HEYWOOD. 1963. Principlesof in the southwest.SIDA 11:45-61. angiospermtaxonomy. London: Oliver and Boyd. and . 1985b. The genus Portulacain ELISENS, W. J.and A. S. TOMB. 1983. Seed morphol- the southeasternUnited States. Castanea 50:96- ogy in new world Antirrhinae(Scrophulariaceae): 104. Systematic and phylogenetic implications. P1. NEWELL, C. A. and T. HYMOWITZ. 1978. Seed coat Syst. Evol. 142:23-47. variation in GlycineWilld. subgenus Glycine(Le- GEESINK, R. 1969. Account of the genus Portulacain guminosae) by SEM. Brittonia30:76-88. Indo-Australia and the Pacific (Portulacaceae). POELLNITZ, K. VON. 1934. Versuch zur einer Mono- Blumea 17:275-301. graphie der GattungPortulaca. Feddes Repret Z. HAUPTLI, H., B. D. WEBSTER, and S. JAIN. 1978. Vari- Bot. Taxon. Gebot. 37:240-320. ation in nutletmorphology of Limnanthes.Amer. SMITH, E. B. 1978. An atlas and annotatedlist of the J.Bot. 65:615-624. vascular plants of Arkansas,+ supplements.Fay- HILL, R. J. 1976. Taxonomic and phylogenetic sig- etteville,Arkansas: Published by the author. nificance of seed microsculpturingin Mentzelia STEYERMARK, J.A. 1963. Floraof Missouri. Ames: Iowa (Loasaceae) in Wyoming and adjacent western St. UniversityPress. states. Brittonia28:86-112. WOFFORD, B. E. 1981. Externalseed morphology of JUDD, W. S. and R. P. WUNDERLIN. 1981. Firstreport Arenaria(Caryophyllaceae) of the southeastern of Portulacaamilis (Portulacaceae) in the United United States. Syst. Bot. 6:126-135. States. SIDA 9:135-138. WYATT, R. 1984. Intraspecificvariation in seed mor- LEGRAND, C. D. 1962. Las especies americanas de phology of Arenariauniflora (Caryophyllaceae). Portulaca.Anales Mus. Nac. Montevideo 7:1-147. Syst. Bot. 9:423-431.

SystematicBotany (1986), 11(2): p. 308 C Copyright1986 by the American Society of Plant Taxonomists

Announcement International Organization of Plant Biosystema- the Chairperson, Dr. KrystynaUrbanska, Geobota- tists. The Executiveand Council of the Internation- nisches Institut,ETH, Stiftung Rubel, Zurichberg- al Organization of Plant Biosystematists(IOPB) will strasse 38, CH-8044, ZURICH, SWITZERLAND. meet during the IOPB 1986 Symposium, "Differen- IOPB publishes the IOPB NEWSLETTER. Infor- tiation Patternsin Higher Plants", Zurich, Switzer- mation for the IOPB NEWSLETTER may be sent to land, 13-18 July 1986. Anyone wishing to place an the Editor,Dr. KrystynaUrbanska. item on the agenda for discussion should write to Application forms for membership in IOPB may Dr. Liv Borgen, Secretary,IOPB, Botanical Garden be obtained from the Secretaryby sending US $25 and Museum, Trondheimsveien 23B, N-OSLO 5, (for the period 1983-1987) directlyto the Secretary- NORWAY. Treasurer of IOPB, Dr. L. Borgen at the above ad- Informationon participationmay be obtained from dress.

This content downloaded from 83.204.224.234 on Fri, 9 May 2014 05:55:35 AM All use subject to JSTOR Terms and Conditions