The silica bodies of tropical American grasses : morphology, taxonomy, and implications for grass systematics and fossil phytolith identification

48 Pages·1998·3.3 MB·English

by  Piperno, Dolores R

The Silica Bodies of Tropical
American Grasses: Morphology,
Taxonomy, and Implications
for Grass Systematics and
Fossil Phytolith Identification
SERIESPUBLICATIONSOFTHESMITHSONIANINSTITUTION
Emphasisuponpublicationasameansof"diffusing knowledge"wasexpressedbythefirst
Secretary of the Smithsonian. In his formal plan for the institution, Joseph Henry outlined a
program that included the following statement: "It is proposed to publish a series of reports,
giving anaccountofthenewdiscoveriesinscience,andofthechanges madefromyeartoyear
in all branches of knowledge." This theme of basic research hasbeen adhered tothrough the
years by thousands of titles issued in series publications under the Smithsonian imprint,
commencing with Smithsonian Contributions to Knowledge in 1848 and continuing with the
following activeseries:
Smithsonian Contributions toAnthropology
Smithsonian Contributions toBotany
Smithsonian Contributions to the EarthSciences
SmithsonianContributions totheMarineSciences
Smithsonian Contributions toPaleobiology
Smithsonian Contributions toZoology
SmithsonianFolklifeStudies
SmithsonianStudiesinAirandSpace
Smithsonian StudiesinHistoryand Technology
Intheseseries,the Institutionpublishessmallpapersandfull-scalemonographsthatreport
the research and collections ofitsvarious museums and bureauxorofprofessional colleagues
in the world of science and scholarship. The publications are distributed by mailing lists to
libraries, universities, andsimilarinstitutionsthroughouttheworld.
Papers or monographs submitted for series publication are received by the Smithsonian
Institution Press, subjectto itsownreviewforformatand style,onlythrough departmentsofthe
variousSmithsonianmuseumsorbureaux,wherethemanuscriptsaregivensubstantivereview.
Press requirementsformanuscriptand artpreparation areoutlined onthe insidebackcover.
Michael Heyman
I.
Secretary
Smithsonian Institution
SMITHSONIAN CONTRIBUTIONS TO BOTANY NUMBER
• 8 5
The Silica Bodies of Tropical
American Grasses: Morphology,
Taxonomy, and Implications
for Grass Systematics and
Fossil Phytolith Identification
Dolores R. Piperno
and Deborah M. Pearsall
ISS
JUL - 6 1998
SMITHSONIAN
INSTITUTION
SMITHSONIAN INSTITUTION PRESS
Washington, D.C.
1998
—
ABSTRACT
Piperno, Dolores R., and Deborah M. Pearsall. The Silica Bodies of Tropical American
Grasses: Morphology, Taxonomy, and Implications forGrass Systematics and Fossil Phytolith
Identi—fication. Smithsonian Contributions toBotany, number85,40pages, 76figures, 2 tables,
1998. Silica bodies from over 200 species ofNeotropical grasses comprising 80 different
generafromall currentlyrecognized subfamilies have been isolated fromplant tissue and have
beendescribed. Silica-body shapesaresignificantatvaryingtaxonomic levels, fromthefamily
to the genus. The Bambusoideae, especially, contribute large numbers of tribal- and
genus-specificforms. Significantcorrelationsarefoundbetweenphytolith shapeandthesource
plant's taxonomic relationships and postulated phylogeny. Disarticulated short-cell phytoliths
occurringinancientsoilsandsedimentscanbeusedtomakeidentificationsofcertaintaxainthe
Poaceae below the level offamily. Silica bodies observed in fossil grasses may elucidate the
evolutionary history ofthe Poaceae.
Official publication date is handstamped in a limited number of initial copies and is
recordedin the Institution's annual report,AnnalsoftheSmithsonian Institution. SERIESCOVER
DESIGN: Leafclearing from the katsura tree Cercidiphyllumjaponicum Siebold and Zuccarini.
LibraryofCongressCataloging-in-Publication Data
Piperno. DoloresR.
Thesilicabodiesoftropical Americangrasses : morphology,taxonomy,andimplicationsforgrasssystematicsand
fossil phy—tolith identification / DoloresR. PipernoandDeborah M. Pearsall.
p. cm. (Smithsonian contributions tobotamy : no. 85)
Includesbib—liographical refere—nces(p. 8). — —
1. Grasses Latin America Classific—ation. 2. Grasses Latin America Cytology. 3. Grasses, Fossil
Identification. 4. Silicabodies(Plants) LatinAmerica. I. Pearsall, Deborah M. II. Title. III. Series.
QK1.S2747 no. 85 [QK495.G74] 580s-dc21 [584'.9'098] 97-22711
®
The paper used in this publication meets the minimum requirements of the—American
National Standard forPermanence ofPaper for Printed Library Materials Z39.48 1984.
Contents
Page
Introduction 1
Acknowledgments 2
Sampling Methods 2
General Patterns at the Subfamily Level 2
Examples ofPhytolith Discrimination at and below the Subfamily Level 3
Phytoliths with Limited Distributions in the Poaceae 4
Other Kinds ofSilica Bodies and Patterns ofSilicification 4
Other Important Phytoliths not Derived from Short Cells 5
Summary 5
Grass Taxon Identification and Environmental Reconstruction from
Phytoliths Retrieved from Ancient Tropical Soils and Sediments 5
The Place ofPhytoliths in Paleoagrostology and Grass Systematics 6
Literature Cited 8
Tables 10
Figures 22
in

The Silica Bodies of Tropical
American Grasses: Morphology,
Taxonomy, and Implications
for Grass Systematics and
Fossil Phytolith Identification
Dolores R. Piperno
and Deborah M. Pearsall
Introduction subfamily to subfamily to be ofcritical use in the systematics
Silica bodies are a type ofphytolith that form in specialized and evolution of the Poaceae (Thomasson, 1987; but see
epidermal cells of grass leaves. They have long been Muholland, 1989, and Ollendorfet al., 1988).
recognized as distinctive to the Poaceae (Grob, 1896; Prat, Compared to research carried out in temperate regions, far
1936; Blackman, 1971). Various workers have considered fewerstudies involvinggrassphytoliths havebeenattempted in
silica bodies to be diagnostic to family, and often they reflect theOldandNewWorldtropics (Piperno, 1988; Pearsall, 1989).
the plants' subfamily affilations (Twiss et al., 1969; Brown, As the Poaceae may have had its origins in moist tropical or
1984; Mulholland, 1989). In paleoecological and archaeologi- subtropical lowlands (Stebbins, 1986), and as there has been a
cal sequences fromNorth America, silicabodies haveprovided heightened interest in the determination ofnatural and human
information on past environments and human subsistence, effectson lowlandtropical environmentsduringtheQuaternary
having been recovered in a virtually unaltered state from period (Leyden, 1987; Leyden et al., 1993; Piperno et al., 1990,
sediments up to 600,000 years old (Fredlund, 1986; Mulhol- 1991), the availability ofphytolith keys with broad application
land, 1993). They also have been recognized in fossilized to the grass family will become increasingly important in
remains ofgrasses datingto the Miocenewhere, inconjunction botanical research.
withothermicromorphological charactersofthe leaf,theyhave We have completed the first major survey ofsilica bodies in
provided information on the fossil's phylogeny and taxonomic New World tropical grasses. Our goals were two-fold: (1) to
relationships (Thomasson, 1984, 1987; Thomasson et al., provide a key to the kinds and distributions of short-cell
1986). phytoliths in the tropical American flora that will be ofuse in
Although well established as significant and taxonomically plant systematics andpaleobotany,and(2)todeterminetowhat
important components of grasses, phytoliths rarely have been extentidentificationofgrass subfamilies, tribes,andgeneracan
used independently eitherto positively identify specific taxaor be made from individual phytoliths retrievedfrom ancient soils
to elucidate grass phylogeny. The prevailing assumption has and sediments in Central and South America.
beenthat grassesproduce a limitedrangeofsilica-bodyshapes, Ourresults indicate thatcertain kinds ofsilicabodies maybe
which, at any rate, too often demonstrate overlap from found in all ofthe subfamilies ofthe Poaceae and they can be
used for discrimination below the family level only with
Dolores R. Piperno, Smithsonian Tropical Research Institute, Box caution. However, other short-cell phytoliths disarticulated
2072, Balboa, Panama (mailing address: STRI Unit 0948, APOAA
fromplant tissue are valid indicators ofindividual subfamilies,
34002-0948). Deborah M. Pearsall, Department ofAnthropology,
American Archaeology, University ofMissouri, Columbia, Missouri tribes, and genera of grasses, and these do provide valuable
65211. information on grass taxonomy and phylogeny.
1
2 SMITHSONIAN CONTRIBUTIONS TO BOTANY
—
Acknowledgments. This research was supported by a although the latter two structures often have a very low
grant to the authors from the National Science Foundation phytolith content and usually contribute other types of
(BNS-89-2365), the Smithsonian Tropical Research Institute phytoliths.
(STRI), and a grant to the STRI from the Andrew W. Mellon In Table 1, the placement of phytoliths under the saddle-,
Foundation. Part ofthe work was carried out at The Museum rondel-, bilobate-, orcross-shaped category signifies that these
Applied Science Center for Archaeology (MASCA), Univer- bodies are of the classic shapes that have been commonly
sity of Pennsylvania, which Piperno thanks for its generous reported in previous grass studies. In other words, they are
support. usually tabular, mirror-image forms that almost alwaysassume
saddle-, bilobate-, or cross-like orientations when observed in
Sampling Methods tissue andalsowhen removed from plant parts and mountedon
slides as separate entities.
Leaves, culms, and inflorescences from over 200 species of In this study, many variations of these conventional,
grasses comprising 80 different genera and representing a short-cell phytoliths were observed. Such phytoliths typically
broad survey ofall fivesubfamilies now recognized(Watsonet had only a single developed saddle- or bilobate-like face, and,
al, 1985; Kellogg and Campbell, 1987) were studied. asoriginally formedintheplant, theywerethickwith relatively
Assignation of individual genera to subfamily and tribe deep extensions into the epidermal tissue. When removed from
followed the current consensus in grass systematics (Soder- the plant andplaced on slides they turned and assumed various
strom et al., 1987). Most ofthe species analyzedwere sampled kinds ofwide, irregularshapes, with the saddle orbilobate face
from annotated herbarium sheets at the Missouri Botanical that had been observed in tissue becoming more difficult to
Garden (in St. Louis and housed at the Smithsonian Tropical define because it had become one of the lateral edges ofthe
Research Institute, Panama) and the National Herbarium in phytolith. In Table 1, these phytoliths were given either their
Washington, D.C. A few species were collected in the field by own categories or were placed in the "Other" category. Most
the authors during field trips made to Latin America over the are described and illustrated.
past 15 years. Twiss et al. (1969) proposed three major divisions of
For most species under consideration, at least two replicate short-cell phytoliths corresponding to three dominant subfami-
specimens derived from different, local populations were lies native to the Great Plains ofthe United States: bilobate/
analyzed. Asbamboos inhabit awide rangeofhabitatsandmay cross = Panicoideae, saddle = Chloridoideae, and circular/oval/
constitute a significant component of the tropical flora, our rectangular= Pooideae. They recognized that some deviation
sample included representatives of almost every genera and from this typology occurred; for example, circular to oval
many species ofthe Bambusoideae reported from the Neotro- bodies were observed in some panicoid grasses. Brown (1984)
pics (Soderstrom and Ellis, 1987). and Mulholland (1989) carried out more extensive studies of
Phytoliths were extracted from plant material by the wet North American grasses and also found that although the
oxidation method described in Piperno (1988) and were three-part division generally held, there were significant
mounted on slides in Permount, a histological fixative. deviations from the expectedpattern. The same was true ofthe
A multitude ofshort-cell phytolith types were isolated from tropical grasses studied herein.
the grasses analyzed. Many of them have not been reported Circularto oval phytoliths or "rondels" (Mulholland, 1989),
previously in studies ofNorth American and European grasses which are most closely associated with the Pooideae (Table 1),
(e.g., Parry and Smithson, 1964, 1966; Twiss et al., 1969; also are found in the Arundinoideae, Panicoideae, and, most
Brown, 1984; Mulholland, 1989). This paper will discuss and prominently, in the inflorescences of the Bambusoideae
illustrate the major distributional patterns demonstrated by the (Figures 1-4, Table 1). Bilobate phytoliths, the most character-
phytolith types, concentrating on those forms considered to be
istic markers ofthe Panicoideae (Table 1), also are present in
diagnostic ofgrasses below the family level, as well as on the
the Arundinoideae, Pooideae, Bambusoideae, and Chloridoi-
forms that appearto be repeated often across majorboundaries deae (Figures 5-15). Indeed, bilobates are the most common
ofthe family.
kind of silica body in phytolith assemblages from certain
genera in these subfamilies, such as Aristida (Arundinoideae),
General Patterns at the Subfamily Level Eragrastis (Chloridoideae), and Stipa (Pooideae). The cross-
Table 1 summarizes the distribution of the short-cell shaped phytolith, another panicoid marker, is common in
phytoliths isolated from the grasses studied. Silica bodies certain Bambusoideae (tribe Olyreae) (Table 1) and occurs in
classified as circular to oval- (rondels), saddle-, bilobate-, or small numbers in the Chloridoideae, Arundinoideae, and
cross-shapedarewell-establisheddiagnostic featuresofthe leaf Pooideae {Brachypodium, Polypogon) (Figures 16, 17).
epidermis, occurringbothoverandbetweenthe leafveins. This Saddle-shapedphytoliths are the dominant phytolith class of
study shows that they also may be conspicuous components of the Chloridoideae (Table 1). They are also common in two
the epidermis of inflorescence bracts, culms, and seeds, subtribes ofthe Bambusoideae (Guaduinae and Chusqueinae)
NUMBER
85 3
(Table 1) and are present in two genera ofthe Arundinoideae one-and two-spiked-side body, saddle with ridged platform,
(Aristida, Phragmites) (Figures 18-22). Chusqoid body, bilobate/saddle both sides; Figures 26-3
1 ).
Individual generain the Arundinoideaearemarkedbyhighly Phytoliths just described are commonly found in the bamboo
divergent sets of phytoliths that may exhibit Panicoideae, subtribes Guaduinae andChusquineaeandoccurrarely in other
Chloridoideae, or Pooideae tendencies (Table 1). Phytoliths bamboo taxa, not being noted at all in the tribe Olyreae (Table
with both saddle and bilobate tendencies, called "saddle/ 1).
bilobates," are common in this family, whereas the conven- The Bambusoideae also contribute a large number of
tional kinds of saddle and rondel forms are relatively rare. phytoliths diagnostic at the tribal, subtribal, and genus levels.
These determinations are consistent with findings that the Bodies whose long axes have two points and two large,
Arundinoideae is actually a hetereogeneous andpoorly defined rounded lobes are diagnostic ofChusquea, and they have been
group of loosely related genera (Conert, 1987), and with the observed thus far only in two species of the genus, C.
current beliefthat this subfamily is primitive and basal to all simpliciflora and C. pittieri (Figure 32). Chusquea bodies with
others ofthe Poaceae (Kellogg and Campbell, 1987). multifaceted faces were observedonly in C. pittieri, a highland
It is evident from this analysis that the assignation of representative of the genus (Figure 33). This distribution
short-cell phytoliths found in ancient sediments and fossil probably allows differentiation ofthe single lowland species of
plants to a particular grass subfamily will not always be the genus currently recognized, C. simpliciflora. Large,
possible. Compounding the difficulty is that several different distinctive bodies are produced in Streptochaeta spp. (Figure
plant structures may contribute confuser phytoliths. For 34), whereas Neurolepis contributes a diagnostic small and
example, rondels may be found in both the leaves and wide tent-shaped body (Figure 35). Pharus contributes a
inflorescences of Aristida, whereas saddles occur in leaves, unique, wide body (Figure 36), whose smaller, lateral face
culms, and inflorescences of bamboos. However, a large presents as a biloboid structure in tissue. Raddiela spp.
numberofothershort-cell phytoliths that were observed inthis contributebilobatesandcross shapesthat appeartobe enclosed
study do appear to be both disjunct in distribution and in siliceous plates (Figures 37, 38).
diagnosticatandbelowthe subfamily level. Someofthemneed The Olyreae produce a number of different kinds of
tobe isolated fromplanttissue forproperstudy. Many havenot genus-specific forms that are irregular versions of complex
been described previously in other phytolith studies. A bilobates or crosses. Their orientation in plant tissue reveals
summary of these phytoliths can be found in Table 2. There these basic, short-cell shapes (Figure 39). When isolated from
also appears to be considerable potential for discrimination of tissue, they are oriented differently and assume a number of
theconventional short-cellphytolithtypesusingmicrodifferen- diagnostic forms. For example, those in Maclurolyra and
tiation ofshape characteristics and phytolith size. Pariana have one sinuous shape (representing the turned
bilobate) and one sloping edge, and they are extremely wide.
Examples of Phytolith Discrimination atand (Figures40,41). Phytoliths fromeach ofthesegeneraappearto
belowthe subfamily level be differentiable on the basis ofthe regularity ofthe edge slope
and the width of the body. Arberella contributes bodies with
"Long, wavy trapezoids" (Brown, 1984) appearto be unique various irregular points and concavities on one edge (Figures
tothe Pooideaeandcanbeusedtoidentifythis groupofmainly 42-44).
high-elevationtropical grasses (Figure 23). Squaretorectangu- Other phytoliths found in the Bambusoideae are diagnostic
lar short-cell phytoliths of the kind described by Twiss et al. at the tribal level. Phytoliths previously described as irregular
(1969) also may be diagnostic of tropical Pooideae, as they mesophyll (Piperno, 1988), but which are now also known to
were not observed in taxa from other subfamilies analyzed be irregular, complex short-cell phytoliths, are found in the
herein. Phytoliths called "plateaued saddles" were isolated in Olyreae (Figures 45, 46). Cross-shaped bodies with partly or
great number from the leaves ofPhragmites (Figures 24, 25). fully developed conical projections on one side, the "Variant 3
More study is needed before they can be assigned genus- and 8"types (Piperno, 1988), alsoarediagnostic ofthe Olyreae
specific status, but they can be used as a marker of the (Figure 17). Many bilobates from this tribe also display the
Arundinoideae and to search for, or rule out, the possible same Variant 3 and 8 attributes (Figures 47, 48). Cross-shaped
presence ofPhragmites in soil phytolith assemblages. bodies with three indentations, blocky structures, and concave
The Bambusoideae contribute a large number of types faces, and with blocky structures, concave faces, and serrated
diagnostic at the family level (Table 2). Many of these are shortaxesarecharacteristic ofOtateafimbriata and Chusquea,
variations on saddle or bilobate themes. Typically, the respectively (Figures 49, 50).
phytolith, as found in tissue, is a thick structure with only one The documentation of many diagnostic phytoliths in
bilobate- or saddle-like face. When isolated from tissue, bamboos is ofconsiderable significance because the subfamily
phytoliths turn from their original orientations to assume often exhibits habitat preferences that are very different from
various kinds of shapes with acute points, "collapsed" sides, other tropical grasses. Bamboos are commonly found in the
and other irregularities of structure (e.g., collapsed saddle, shady, cool understorey oftropical forests rather than in open
4 SMITHSONIAN CONTRIBUTIONS TO BOTANY
and disturbed vegetation. Bamboo phytoliths, then, can be assemblages in which over 90% of saddles were longer than
important markers of various kinds of tropical forests in wide (Figures 21, 22, 60). Another difference between saddles
archaeological andpaleoecological studies (seePiperno, 1988). from chloridoid and non-cloridoid grasses is that very tall
The distribution of bamboo phytoliths by tribe and subtribe saddles, those measuring longer than 15 microns, dominate
mirrors current taxonomic assessments ofthe family based on bamboo assemblages, whereas saddles shorterthan 15 microns
other anatomical and structural features (Soderstrom and Ellis, long predominate in chloridoids (Figure 22).
1987), indicating the fundamental relationship of silica-body
shape to Bambusoideae systematics.
Phytoliths with Limited Distributions in the Poaceae
Smaller numbers of one-to-one associations between a
phytolith and a tribe or genus were observed in other Phytoliths called "narrow elliptates" were isolated in great
subfamilies. They occurred in Arundinella (Panicoideae), number from the Bambusoideae, especially the Guaduinae and
Ahstida (Arundinoideae), Polypogon elongatus (Pooideae), Chusqueinae, where they are a major component of the
and Aegopogon and Jouvea (Panicoideae) (Figures 51-56). intercostal areas ofthe leafepidermis (Figures 61, 62). They
Phytoliths from Ahstida are called "rondeloid/saddleloid" were not observed in the Olyreae, Streptochaeteae, Phareae,
because they display characteristics intermediate between Streptogyneae, orNeurolepidinae. Narrow elliptates also occur
classic rondels and saddles. Arundinella possesses an unusual in small numbers in the Arundinoideae, having been observed
phytolith that is tall, thick, and more-or-lesssaddle shapewhen thus far only in Gynerium (Figure 63). These phytoliths have
found in tissue, and then assumes a diagnostic form when biloboid characteristicswhen viewed in tissue (Figures 61, 63),
removed from the leaf (Figures 51, 52). It is very similar to but they may assume othershapes when removed from the leaf
phytoliths found in the Arundinoideae taxon Gynerium (Figure (Figure 62).
57) andto "saddle/bilobate, both sides"types ofphytoliths that Complex bilobates ortrilobates andpolylobates arecommon
are common in bamboos (Figure 31). in phytolith assemblages from panicoid taxa. They were not
Stebbins (1987) believed that the panicoid grasses derived observed, however, in the Bambusoideae and were extremely
viaArundinella from the Arundinoideae. The divergent set of rare in the Arundinoideae, occurring in very small numbers
phytoliths in Arundinella as compared to other species (less than 1% of the short-cell assemblage) in a few culms.
examined in the Panicoideae, as well as the similarity ofthe Complex bilobates also were observed in Eragrostis (Chlori-
Arundinella leaf phytolith assemblage to that from some doideae) and in Stipa (Pooideae).
Arundinoideae taxa (predominance ofwide, angled bodies and
odd saddle shapes; few bilobates and crosses and no complex Other Kindsof Silica Bodiesand Patterns
bilobates) are consistent with this proposed phylogeny. OF SlLIClFICATION
As noted above, there also is considerable potential for
short-cell phytolith discrimination using microshaped features There are some types ofphytoliths found in tropical grasses
of the conventionally defined short-cell phytolith types. For that do not lend themselves to easy description and classifica-
example, bilobates with rounded or semirounded lobes and tion. Many ofthem derive from culms, in which are typically
long, thin shafts, orwith squared lobes and distinct, thin shafts found small, irregularphytoliths, especially in members ofthe
are common in panicoid taxa, but they were not observed in Arundinoideae (Figures 64, 65). Other unusual phytoliths are
bamboos (Figures 5, 7). Bamboo bilobates, in contrast, usually insignificantcomponentsofassemblages from leavesand floral
are squat bodies with no or unremarkable shafts (Figures bracts(Figure 66). As no overlapoccursbetweenthese types of
11-13, 58, 59). Bilobates from the Chloridoideae (Figure 14) phytoliths and those from otherfamilies, they can be identified
exhibit distinctively flared, convex edges that were first in soils orfossils asgrassbodies. Thesephytoliths would repay
reported by Mulholland 1989). furtherinvestigation, as someofthemare likelyto beofprecise
(
Size differences in bilobates from different subfamilies also taxonomic value.
are apparent. Many bilobates isolated from panicoid grasses Patterns of silicification in certain members ofthe bamboo
exceed 20 microns in length, whereas those from the subtribe Arthyrostilidiinae are most interesting, as it appears
Bambusoideae, Cloridoideae, and Pooideae(withtheexception that leafsilicabodiescharacteristicofbamboosorothergrasses
ofStipa) are almost without exception shorter. are seldomproduced. Most genera are characterizedinstead by
Significant differences also are apparent from subfamily to the presence ofsmall, silicified cone-like structures that arise
subfamily in the type of saddle-shaped phytolith produced. from the epidermis (Figures 67-69). The comment that "this
Many phytolith assemblages from chloridoid grasses have subtribe is rather far removed from the Arundinariinae and
significant proportions (32%-86%) of squat saddles, that is, Bambusinae and appears to have arisen from a single ancestor
the axis ofthe phytolith exhibiting the double-edge, saddle-like different from that which gave rise to otherNew World woody
outline is wider than the other, or the two axes are subtribes such as the Chusqueinae, Guaduinae, and Neurole-
equidimensional (Figures 18-20). In contrast, every species pidinae" (Soderstrom and Ellis, 1987:234) is amply supported
studied from the Bambusoideae and Arundinoideae produced by its silicification patterns.