SystematicBotany (1987), 12(1): pp. 1-8 C Copyright1987 by the American Society of Taxonomists

Systematicsof Croizatia (Euphorbiaceae)

GRADY L. WEBSTER and LYNN GILLESPIE Botany Department,University of California,Davis, California 95616

JULIAN STEYERMARK Missouri Botanical Garden, St. Louis, Missouri 63166

ABSTRACT.Croizatia, a known fromPanama and (Euphorbiaceae, Phyllanthoi- deae), has been considered closely related to the Old World genus .However, the spinose pollen of Croizatiais very differentfrom the semitectatepollen of Actephila,and the two genera appear clearly distinct.The pollen morphology of Croizatiasuggests a possible relationship with the Oldfieldioideae, and it may be the closest extanttaxon to a connectinglink between the subfam- ilies. Three species of Croizatiaare tentativelyrecognized, including a new species, Croizatia pan- amensis; but species delimitationsmust be regarded as provisional.

The Euphorbiaceous genus Croizatia(Steyer- ferencesobserved are the much more reduced mark 1952), named for the eminent student of petals of staminateflowers and the twice bifid the systematics of the Euphorbiaceae, Leon style tips of Croizatia.Although there are other Croizat-Chaly(Steyermark 1983), has long been diagnostic characters pointed out by Steyer- regarded as of uncertain affinity.The first mark (1952), such as the enlarged columella of species published, C. neotropica(Steyermark Croizatia,the overall gross resemblance to Ac- 1952), was described froma fruitingspecimen tephilais striking.Indeed, Gentry(1982) has at- collected in the stateof Anzoategui, Venezuela. tributedto Rodriguez an unpublished combi- Later,Steyermark (1978) added a second species, nation under Actephilafor C. neotropicaand C. naiguatensis,from another fruitingspecimen quotes Rodriguez as statingthat Croizatiais in- found on the Cerro Naiguata, DistritoFederal, distinguishable fromActephila. Venezuela. Despite the uncertaintyregarding the gener- Croizat (in Steyermark1952) stated that the ic status of Croizatia,the floral,fruit, and leaf affinitiesof Croizatiawere close to the Old World charactersappear to provide sufficientdistinc- genus Actephila.In 1978, when C. naiguatensis tions to maintain it as separate fromActephila. was described,the flowersof both species were Furthermore,examination of the pollen grains still unknown, and the affinitiesof the genus with scanning electron microscopyshows that thereforeremained problematical.In the most the pollen of Croizatia(figs. 2, 3) is very differ- recent enumeration of Euphorbiaceous genera ent fromthat of Actephilacollinsae Hunter (fig. (Webster1975), Croizatawas omittedbecause of 4). The pollen grains of A. collinsaeare 40-49 its dubious status;but in the synopticarrange- (x = 46) Amin diameter,subprolate, and tricol- ment of tribes of Phyllanthoideae (Webster porate with large, well-definedcolpi; the exine ined.) its placement appeared to be with either is rather finely semitectate-reticulate.These the Wielandieae or Amanoeae. Levin (1984), in charactersagree ratherwell with the light mi- a survey of leaf charactersin the Phyllanthoi- croscopic observations on other species of Ac- deae, reportsa greatersimilarity of Croizatiato tephila:A. excelsa (Punt 1962), A. nitidulaGag- and Blotiathan to Actephila. nep., and A. ovalisGagnep. (Kohler 1965). Both Recentlyin June 1983, our knowledge of the Punt and Kohler have noted the resemblance reproductive morphology of Croizatia im- of the pollen grains of Actephilato those of An- proved dramatically due to the collection of drachne(s.l.). In contrast,the pollen grains of floweringspecimens of C. naiguatensis.The pet- Croizatianaiguatensis are spherical, 46-56 (x = aliferous flowers(fig. 1) resemble those of Ac- 51) Am in diameter, with the colpi greatly tephila,as Croizat had predicted; there is an es- shortened (ca. 10 ,umlong), the lalongate germ pecial similarity to the flowers of the pore well developed (ca. 10 Am broad); the ex- widespread paleotropical species Actephilaex- ine is tectate-perforateand ornamented with celsa (Dalz.) Muell. Arg. The major floral dif- conical,sharply-pointed spines ca. 3-4 Amhigh.

This content downloaded from 169.237.8.35 on Mon, 16 Sep 2013 18:53:23 PM All use subject to JSTOR Terms and Conditions 2 SYSTEMATIC BOTANY [Volume 12

15cm

1m $75mm A \ j

~~5mm (~~~~)

2mm \Y 2.5mm-H

FIG. 1. Habit and flowersof Croizatianaiguatensis. A. Pistillate branch with fruitsand flowers. B. Part of staminate branch with flowers. C. Staminate flower with one petal enlarged. D. Pistillode. E. Sta- men. F. Pistillate flower with one petal enlarged. G. Gynoecium with disk, petals, and staminode. H. Cross-sectionof ovary. I. Columella and persistentcalyx on fruitingpedicel.

This content downloaded from 169.237.8.35 on Mon, 16 Sep 2013 18:53:23 PM All use subject to JSTOR Terms and Conditions 1987] WEBSTER ET AL.: CROIZATIA 3

7 v,

FIGS.2-7. Pollen grains of Euphorbiaceae; scales = 10 ,um. 2, 3. Croizatianaiguatensis (Berry et al. 4124, DAV). 4. Actephilacollinsae (Larsen 33631, MO). 5. Tetracoccusdioicus (Carlquist s.n., cult. Rancho Santa Ana Botanic Garden). 6. Podocalyxloranthoides (Williams et al. 18249,DAV). 7. Amanoaguyanensis (Tillett & Tillett 45256,DAV).

This content downloaded from 169.237.8.35 on Mon, 16 Sep 2013 18:53:23 PM All use subject to JSTOR Terms and Conditions 4 SYSTEMATIC BOTANY [Volume 12

TABLE 1. Comparison of morphological charactersof Croizatiawith those of Phyllanthoideae and Oldfield- ioideae.

Phyllanthoideae Croizatia Oldfieldioideae Phyllatoxy alternate alternate alternate,opposite, or whorled Stipules usually + + +, reduced, or 0 Petals + or 0 + 0 Staminate mostlyextrastaminal extrastaminal mostlyintrastaminal or 0 disk Pollen mostly3-4-colporate, brevicolpor- brevicolporateor pororate, not echinate ate, echinate echinate Ovules anatropous/hemitropous hemitropous anatropous Styles mostlybifid twice bifid mostlysimple Seeds ecarunculate ecarunculate carunculate or ecarunculate Endosperm +/0 0 +

This striking differencebetween the pollen (1984b); the pollen is oblate, tricolporatewith grains of Croizatiaand those of Actephilarein- colpi shortbut not greatlyreduced and with a forcesthe impression,based on gross morpho- tectate-regulateexine with supratectal spinu- logical characters,that the two genera are quite lose sculptural elements. Although distinct. has been traditionallyplaced in the Phyllan- On the other hand, the echinate, tribrevicol- theae, this appears to have been due to the er- porate pollen found in Croizatiais clearly sim- roneous inclusion of species properlyreferable ilar to pollen of Oldfieldioideae as that group to Flueggea(Webster 1984a). The pollen of the was palynologically defined by Kohler (1965) type species S. durissimais quite anomalous and taxonomicallydelimited by Webster(1967). within the tribeand resemblesto a much great- There is a close resemblance between the pol- er extentpollen of membersof the Wielandieae, len of C. naiguatensisand thatof Tetracoccusdioi- such as Discocarpus. The "S. durissima" cus Parry (fig. 5) and Podocalyx loranthoides pollen type might conceivably bear an ances- Klotzsch (fig. 6), which are considered to be tral relationship to typical "Oldfieldioid" pol- two of the more primitivemembers of the Old- len; however, much furtherstudy is needed fieldioideae on the basis of floral and vegeta- concerning the taxonomic position of S. duris- tive characters(Webster ined.). Podocalyxdif- simaand otherWielandieae beforepossible links fersfrom Croizatia in having tetraporatepollen to the Oldfieldioideae can be substantiated. with a tectate-imperforateexine and in its dis- When the charactersof Croizatiaare tabulated tinctlysmaller diameter (ca. 30 A,m),while Tet- in comparison with the Phyllanthoideae and racoccusdiffers in having 4- to 6-colpoidorate Oldfieldioideae (table 1), it clearlyagrees better pollen and a tectate-regulateexine with gem- with the Phyllanthoideae in five characters mae on the ridges. (petals, staminate disk, ovules, styles,and de- Within the Phyllanthoideae, true spines of velopment of endosperm) and with the Old- the "Oldfieldioid" type have heretoforebeen fieldioideae in only one (pollen). However, unrecorded. The pollen grains of some species most of the resemblance to the Phyllanthoi- of Amanoa are irregularlybaculate; the sculp- deae involves the common possession of prim- turalelements are rod-likewith the apex some- itive (plesiomorphic) characters. When only times expanded or tapered (fig. 7). The differ- derived (apomorphic) characters are consid- ent structureof the "spines" in both Croizatia ered, Croizatiahas two in common with Phyl- and Amanoasuggests thatthe resemblance may lanthoideae (hemitropous ovules, exalbumi- be due to convergence and, therefore,not in- nous seeds) and one with Oldfieldioideae dicative of a close phylogenetic relationship. (echinate pollen). Another taxon within the Phyllanthoideae In table 2, the charactersof Croizatiaare con- which has pollen that bears some resemblance trasted with two putatively primitive genera, to pollen of the Oldfieldioideae is Securinega and Podocalyx, in the subfamilies durissimnaJ. F. Gmel., illustrated by Webster Phyllanthoideae and Oldfieldioideae. Here the

This content downloaded from 169.237.8.35 on Mon, 16 Sep 2013 18:53:23 PM All use subject to JSTOR Terms and Conditions 1987] WEBSTER ET AL.: CROIZATIA 5

TABLE 2. Comparison of morphological charactersof Croizatiawith a primitivegenus of Phyllanthoideae (Wielandia)and a primitivegenus of Oldfieldioideae (Podocalyx).

Wielandia Croizatia Podocalyx Petals + + 0 Staminate extrastaminal,annular extrastaminal,annular 0 disk Anthers introrse introrse extrorse Pollen 3-4-colporate,not echinate 3-brevicolporate,echin- 4-pororate,echinate ate Ovary glabrous glabrous pubescent Ovules anatropous hemitropous anatropous Styles bifid twice bifid stigmatiform Seeds not black and shiny not black and shiny black and shiny Endosperm 0 0 + resemblance with Wielandiais especially strik- subfamilies Phyllanthoideae and Oldfieldioi- ing, since there are six charactersin common deae. If Croizatiaindeed representsan ancestral that are not shared with Podocalyx.Actephila, link to the Oldfieldioideae, its inclusion on cla- the genus resemblingCroizatia in aspect, shares distic grounds would resultin the introduction these same characters. In contrast, Podocalyx of several discordant characters (petaliferous shares only a single character(pollen exine or- flowers,extrastaminal disk, hemitropous ovules, namentation) not possessed by Wielandiaand bifid styles,and exalbuminous seeds) into the Actephila.However, Croizatiahas only one de- subfamilial diagnosis of the Oldfieldioideae. rived characterin common with Wielandia(lack However, its placement in the Phyllanthoideae of endosperm) and one with Podocalyx(echi- would not change the status of the Phyllan- nate pollen). thoideae as a paraphyletic group, since that The available morphological data on the subfamily would be paraphyletic in any cir- Phyllanthoideae are still so incomplete that at cumscription(due to its having given rise to present it would be prematureto attempta de- the subfamilyAcalyphoideae as well) that has tailed cladistic or phenetic analysis. For the been proposed up until now. In the face of in- present, we are dependent on the character adequate data and the problems of analysis dis- contrastspresented in tables 1 and 2, and un- cussed above, it seems expedient to assign Croi- fortunatelythey tell discordant stories. Croiza- zatia provisionally to the Phyllanthoideae, tia shares two derived characters(hemitropous where it can be placed between Blotiaand Ac- ovules and exalbuminous seeds) with the Phyl- tephila. lanthoideae (although only one with any sin- Although these problems of classificationare gle genus), while it shares only the echinate challenging and interesting,study of the mor- pollen characterwith the Oldfieldioideae. One phology of Croizatiais most significantin elu- could argue that on cladistic, as well as phe- cidating phylogenetic and biogeographic rela- netic, grounds placement of Croizatia in the tionships. The link between Croizatia and Phyllanthoideae is indicated. On the other Podocalyxclearly suggests that the Oldfieldioi- hand, the single pollen charactershared with deae is of South American origin,even though Oldfieldioideae involves a complex of features the living taxa now occur widely in otherparts (spine morphology,shortening of colpus, germ of the southern hemisphere (Africa,Madagas- pore configuration),and it seems rather less car,Ceylon, Indonesia, and Australasia). Croiza- likely to have evolved independently in differ- tia and Actephilaare vicariants of West Gon- ent lineages. We thus end on the horns of a dwanaland and East Gondwanaland, dilemma of "quantity" versus "quality" in respectively(sensu Raven and Axelrod 1974). evaluation of derived characters. The progenitorof Croizatiapresumably migrat- These problems in determining the phylo- ed fromAfrica/Madagascar, where the primi- genetic position of Croizatiaalso raise difficul- tive Phyllanthoideae are concentrated,across a ties in demarcatingthe boundary between the narrowerAtlantic; but since Actephilais known

This content downloaded from 169.237.8.35 on Mon, 16 Sep 2013 18:53:23 PM All use subject to JSTOR Terms and Conditions 6 SYSTEMATIC BOTANY [Volume 12

only from India to Austalasia, other now-ex- Unfortunately,the newer collection bears tinct taxa may have been involved. Martin only fruits,and the flowersof Croizatianeotro- (1974, 1982) indicates that relativelyadvanced pica are still unknown. A collection with pis- Oldfieldioideaeof the Austrobuxusalliance were tillate flowers from (Amazonas, Serrania presentin Australasia in the Paleocene, so that de Bagua, Gentryet al. 22989,F, MO) resembles the initial differentiationof Croizatiawould ap- Croizatianeotropica in some respects,but its as- pear to have been in the late Cretaceous. signmentmust be regarded as uncertain.

SYSTEMATICTREATMENT 2. CROIZATIA NAIGUATENSIS Steyermark,Brit- tonia 30:40, fig. 1. 1978.-TYPE: Venezuela, CROIZATIASteyermark, Fieldiana 28:308, fig.57. Distrito Federal, Cerro Naiguatia, upper 1952.-TYPE: Croizatianeotropica Steyerm. reaches of Quebrada del Mata de Platano, Dioecious or ; indumentum sim- 2000 m, 7 Dec 1973,S. Tillett,G. & B. Morillo ple. Leaves alternate,entire, pinnately veined, & B. Manara 82 (holotype: VEN!; isotype: without embedded glands, petiolate; stipules DAV!). deciduous or persistent. Flowers in axillary clusters;bracts inconspicuous. Staminate flow- 5-7 m high; leaf blades elliptic to ob- ers pedicellate; sepals 5, imbricate,entire; petals ovate, acute or obtuse (less often rounded) at 5, much shorterthan sepals, pubescent, entire; tip, long-attenuateat base, 9-21 cm long, 4-9 disk annular, glabrous; stamens 5, free,anthers cm broad, glabrous or inconspicuously strigose introrse;pollen grains globose, exine with con- beneath, lateral nerves 7-10 on each side, ical spines; pistillode trifid.Pistillate flowers somewhat prominulous with the more delicate pedicellate; sepals 5, imbricate,entire, persis- tertiaryveinlet network;petioles 4-9 mm long, tentin fruit;petals 5, much shorterthan sepals, stout, glabrous or strigose; stipules scarious, pubescent,entire; disk annular, glabrous; ovary striate,brownish, 7 mm long, deciduous. Sta- pubescent,3-locular; styles free,slender, twice minate flowersin dense axillary clusters;ped- bifid; ovules paired in each locule, hemitro- icel 3-4 mm long, hirtellous; sepals 5, elliptic, pous. Fruit a capsule; columella distally ex- entire,4-4.5 mm long, 2.5-3 mm broad; petals panded into 3 broad papery wings; seeds paired 5, obovate, subentire, densely long-ciliate, 1- or solitaryin each locule, smooth, not fleshy, 1.3 mm long, and broad; disk massive, smooth ecarunculate; endosperm absent; cotyledons and glabrous, 2.5-3 mm across; stamens 5, fil- greenish, contortuplicate,much broader than aments 2.2-2.5 mm long, hirsutulous below; and about as long as the radicle. anthers elliptic, 1.2-1.4 mm long; pistillode 3-lobed, hirsutulous, 1.8-2.2 mm high. Pistil- KEY TO THE SPECIES late flowerssolitary; pedicel strigillose,becom- ing 8-12 mm long in fruit;sepals 4 or 5, lan- Fruitingpedicels not over 2 cm long; lateral ceolate, obtuse or acute to rounded, strigillose, nervesof leafblades 7-10 . 2. C. naiguatensisbecoming reflexedand persistentin fruit,8-10 2 Fruitingpedicels over cmlong; lateral nerves mm long, 2-3 mm broad; petals (4)5, broadly ofleaf blades 12-15on each side. obovate or suborbicular, ca. 1 mm long and Fruitingpedicels glabrous, 3.5-4.5 cm long; seeds 10.5-12mm long .... 1. C. neotropica broad, densely ciliate; disk cupuliform, gla- Fruitingpedicels pubescent, 2.5-3.5 cm long; brous, entire, ca. 0.4 mm high and 3.5 mm seeds 7.2-10 mm long .... 3. C. panamensis broad; ovary sericeous; styles spreading, 2.8-3 mm long, twice bifid,the unbranched portion 1. CROIZATIA NEOTROPICA Steyermark, Field- 1.5-2 mm long, the primarybranches 1.2-1.5 iana 28:309, fig.57. 1952.-TYPE: Venezue- mm long, ultimate tips 0.3-0.5 mm long. Cap- la, Anzo'ategui, Quebrada Seca, E of Ber- sule oblate, 3-lobed, 1.3-1.4 cm high, 1.7-2 cm gantin, 18 Mar 1945, Steyermark61523 broad, strigillose,reticulate; columella 7-8 mm (holotype: VEN!). long, dilated at apex into 3 obovate-oblong wings rounded at apex, 7 mm long, 5 mm broad. Additional collection. VENEZUELA. Anzoategui: Seeds unknown. DistritoBolivar, Fila El Guacharo,Serrania de Turi- miquire,1200-1350 m, 25 Nov 1981, Davidse & Gon- Additionalcollection. VENEZUELA. Vargas: trail zalez 19452 (F, MO). fromPico de Naiguatato townof Naiguata,Fila del

This content downloaded from 169.237.8.35 on Mon, 16 Sep 2013 18:53:23 PM All use subject to JSTOR Terms and Conditions 1987] WEBSTER ET AL.: CROIZATIA 7

4cm ^

1 cm

FIG. 8. Croizatiapanamensis. a. Habit. b. Fruitingcalyces (Garcia-Barriga 11175, COL).

Corozo,24 June1983, P. Berryet al. 4121, 4124 (DAV, or small tree 1-6 m high, usually with VEN). a single main stem;leaf blades chartaceous,gla- strigose-hispidulousbeneath, Although the recent collection of flowering brous or sparsely short-acuminate,basally at- materialof Croizatianaiguatensis provides forthe obovate, abruptly tenuate, 22-45 cm long, 5-12 cm broad, with firsttime an understandingof the floral mor- arcuate-ascending lateral nerves con- phology of the genus, it is difficultto compare ca. 15 nected by intramarginalloops, veins (and to a it with the other two species because of the lesser extent) veinlets prominulous beneath; incompletenessof the collections.It differsfrom 0.5-1 cm long and 3-4 mm thick; sti- both C. neotropicaand C. panamensisin its small- petioles more or less persistent,oblong-lanceo- er leaves with fewer lateral veins and shorter pules ribbed sericeous, 10-20 mm fruitingpedicels; it agrees with C. neotropicain late, acuminate, 6-7 mm broad. Flowers not observed. its deciduous stipules but differsin its pubes- long, pedicels hirtellous,2.5-3.5 mm long, cent pedicels. Fruiting 1.2-2 mm thick; sepals more or less persistent in fruit,reflexed, elliptic-lanceolate, acute or 3. Croizatia panamensis Webster,sp. nov. (fig. subacute,8-9 mm long, 3-3.2 mm broad, exter- 8).-TYPE: Panama, Panama, primaryforest hispidulous; columella ca. 8-9 mm high, along road fromEl Llano to Carti-Tupile, nally mm broad; seeds trigonous, smooth, 300-500 m, 30 Mar 1973, L. L. Liesner1279 10-11 7.2-10 mm long, 5.3-6.5 mm broad, (holotype:MO!; isotype: DAV!). brownish, hilum medial, ca. 2 mm broad; cotyledons ca. Ab C. neotropicadiffert pedicellis brevioribus 10 mm long, radicle ca. 5 mm long. seminibusparvioribus, ab C. naiguatensediffert foliismajoribus stipulis persistentibus,pedicil- Additionalcollections. COLOMBIA. Choc6: Villa- lis longioribus. conto,cerca de los rios Quito y Palmado,120 m, 1

This content downloaded from 169.237.8.35 on Mon, 16 Sep 2013 18:53:23 PM All use subject to JSTOR Terms and Conditions 8 SYSTEMATIC BOTANY [Volume 12

Aug 1944, Garcia-Barriga11161 (US), Garcia-Barriga LITERATURE CITED 11175(COL). GENTRY,A. H. 1982. Phytogeographic patterns as This species is proposed with some hesita- evidence for a Choc6 refuge. Pp. 112-136 in tion because of the lack of floweringmaterial. Biologicaldiversification in the tropics,ed. G. T. However, it appears to differfrom Croizatia neo- Prance. New York: Columbia Univ. Press. tropicaand C. naiguatensisin habit (the scarcely KbHLER,E. 1965. Die Pollenmorphologie der biov- branched stems contrasting with both other ulaten Euphorbiaceae und ihre Bedeutung fur species) and in the more-or-lesspersistent stip- die Taxonomie. Grana Palynol. 6:26-120. LEVIN,G. 1984. Systematic foliar morphology of ules. It diverges fromC. neotropicain its shorter Phyllanthoideae (Euphorbiaceae). Ph.D. disser- strigillose fruitingpedicels and smaller seeds, tation,Univ. California,Davis. and fromC. naiguatensisin its larger leaves and MARTIN,H. A. 1974. The identificationof some Ter- longer fruitingpedicels. Until flowering ma- tiarypollen belonging to the family Euphorbi- terial of both C. panamensisand C. neotropicais aceae. Austral. J.Bot. 22:271-291. collected, furthercomparisons would be pre- . 1982. Changing Cenozoic barriersand the mature. Australian paleobotanical record. Ann. Missouri A very distinctive collection from Napo Bot. Gard. 69:625-667. Province, Ecuador, 1llgaard et al. 38956 (AAU), PUNT,W. 1962. Pollen morphology of the Euphor- resembles Croizatiapanamensis, as pointed out biaceae with special reference to . Wentia 7:1-116. by Dr. Michael Huft (pers. comm.). However, RAVEN,P. and D. AXELROD.1974. Angiosperm bio- although it has the elongated persistentstip- geographyand past continentalmovements. Ann. ules and oblanceolate leaves of the Panama Missouri Bot. Gard. 61:539-673. species, the Ecuadorian plant has much longer STEYERMARK,J. 1952. Euphorbiaceae, in Botanical fruitingpedicels and more acuminate sepals. It exploration in Venezuela-II. Fieldiana Bot. 28: may well prove to be a differentspecies, but it 304-322. does not seem advisable to describe it as new . 1978. New taxa from the Avila and Nai- on the basis of the single collection. guata mountains,Venezuela. Brittonia30:39-49. . 1983. [Biographical note on] Leon Croizat- ACKNOWLEDGMENTS. Thisstudy was supportedin Chaly. Taxon 32:530-531. partby NSF grantsand in partby grantsfrom the WEBSTER,G. L. 1967. The genera of Euphorbiaceae GraduateDivision and BotanyDepartment, Univer- in the southeasternUnited States. J.Arnold Arb. sityof California,Davis. Scanningelectron micro- 48:303-430. graphswere takenby Ms. Lynn Gillespieand Dr. . 1975. Conspectus of a new classificationof StevenP. Lynch.Drawings of flowers were provided the Euphorbiaceae. Taxon 24:593-601. bySr. Bruno Manara. We are indebtedto Dr. Michael . 1984a. A revision of Flueggea (Euphorbi- Huftof the MissouriBotanical Garden for making aceae). Allertonia 3:259-312. availablesome critical specimens. . 1984b. Jablonskia,a new genus of Euphor- biaceae fromSouth America.Syst. Bot. 9:229-235.

This content downloaded from 169.237.8.35 on Mon, 16 Sep 2013 18:53:23 PM All use subject to JSTOR Terms and Conditions