Phytoliths in plants and soils of the interior Pacific Northwest, USA

Abstract

Phytoliths are a useful paleoproxy in the arid environments. This modern analog study assessed variability of silica phytoliths in 38 species of plants and 58 modern soil samples from 24 locations in the interior Pacific Northwest. Phytoliths were grouped into 20 broadly defined morphotypes based on their 3D shapes under light microscope and presumed anatomical origin within the plant. Grasses (all C3) have most diverse forms. Most examined conifers, sedges and some shrubs produce identifiable phytoliths as well. Eight different community types can be distinguished based on their modern phytolith record in soils, including shrublands, four regional grassland types, and three forest types. Low percentages of grass phytoliths and high incidence of non-grass forms correspond to forest vegetation in the region today, while certain grass phytoliths allow further differentiation among different grasslands. Phytolith assemblages were further compared to 5 environmental variables, including elevation, mean annual temperature, mean annual precipitation, a moisture index and a growing-degree days index. Some morphotypes tend to occur within relatively narrow environmental windows, which could enable direct paleoenvironmental inferences from phytoliths in geological sediments from the region.

Full text

Phytoliths in plants and soils of the interior Pacific Northwest, USA
Mikhail S. BlinnikovT
Department of Geography, St. Cloud State University, St. Cloud, MN 56301-4498, USA
Received 13 July 2004; received in revised form 14 February 2005; accepted 23 February 2005
Abstract
Phytoliths are a useful paleoproxy in the arid environments. This modern analog study assessed variability of silica
phytoliths in 38 species of plants and 58 modern soil samples from 24 locations in the interior Pacific Northwest. Phytoliths
were grouped into 20 broadly defined morphotypes based on their 3D shapes under light microscope and presumed anatomical
origin within the plant. Grasses (all C
3
) have most diverse forms. Most examined conifers, sedges and some shrubs produce
identifiable phytoliths as well. Eight different community types can be distinguished based on their modern phytolith record in
soils, including shrublands, four regional grassland types, and three forest types. Low percentages of grass phytoliths and high
incidence of non-grass forms correspond to forest vegetation in the region today, while certain grass phytoliths allow further
differentiation among different grasslands. Phytolith assemblages were further compared to 5 environmental variables,
including elevation, mean annual temperature, mean annual precipitation, a moisture index and a growing-degree days index.
Some morphotypes tend to occur within relatively narrow environmental windows, which could enable direct
paleoenvironmental inferences from phytoliths in geological sediments from the region.
D2005 Elsevier B.V. All rights reserved.
Keywords: climate; modern analogs; Oregon state; plant opal; vegetation; Washington state
1. Introduction
Phytolith analysis is a powerful, yet relatively
underutilized, method of paleoenvironmental recon-
struction that can be used to supplement pollen and
macrofossil analyses (Piperno, 1988; Pearsall, 2000).
In North America, both archaeologists (Rovner, 1971;
Mulholland, 1993) and paleoecologists (Kurman,
1985; Fredlund and Tieszen, 1997a,b; Kearns, 2001;
Blinnikov et al., 2002) used phytoliths to infer a range
of paleoenvironmental conditions. Despite some early
applications of phytoliths in paleoenvironmental and
paleopedological work (Smithson, 1958; Witty and
Knox, 1964; Twiss et al., 1969; Rovner, 1971;
Norgren, 1973), the use of phytoliths in paleoecology
remains uncommon (Piperno and Persall, 1993).
Recent studies suggest that any paleoenvironmental
reconstructions using phytoliths must begin with
analyzing modern phytoliths distribution in plants
and soils in the given region (Bowdery, 1998; Carnelli
et al., 2001; Lu and Liu, 2003).
0034-6667/$ - see front matter D2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.revpalbo.2005.02.006
TTel.: +1 320 308 2263; fax: +1 320 308 1660.
E-mail address: mblinnikov@stcloudstate.edu.
Review of Palaeobotany and Palynology 135 (2005) 71 – 98
www.elsevier.com/locate/revpalbo

In this paper I present results of a modern analog
study of phytoliths in plants and soils of the US interior
Pacific Northwest defined as states of Washington and
Oregon east of the Cascade Range. This study provides
a modern analog dataset required for any paleoenvir-
onmental reconstruction of the late Pleistocene and
Holocene vegetation of the region, including the
previously published work (Blinnikov et al., 2001,
2002). In fact, only two other studies of phytoliths in
plants and modern soils are available from the Pacific
Northwest (Witty and Knox, 1964; Norgren, 1973).
The former mostly focused on phytoliths in soils at a
forest-grassland ecotone in central Oregon, while the
latter focused on Douglas-fir forests of the western
Cascade Range and grasslands of north-central Ore-
gon. Together they provide data on phytoliths from ca.
15 species in the region, mostly conifers and grasses.
Neither study provided an adequate sampling design to
compare phytolith distributions among different veg-
etation types.
In temperate North America, phytoliths have
been mostly studied in grasses (Twiss et al.,
1969), sedges (Walter, 1975; Ollendorf, 1992),
conifers (Klein and Geis, 1978), deciduous trees
(Geis, 1973), and some dicotyledonous shrubs and
herbs (Bozarth, 1992). Their distribution in grasses
has been studied most extensively within Alberta
(Blackman, 1971), North Dakota (Mulholland,
1989), central Great Plains (Twiss et al., 1969;
Fredlund and Tieszen, 1994), northern Great Plains
(Brown, 1984), southeastern states (Lanning and
Eleuterius, 1987; Lu and Liu, 2003), and northern
Arizona (Kearns, 2001).
The number of modern soil studies on phytoliths
on the continent is even smaller. Two important
examples include reports of Bozarth (1993) from
Alberta and Fredlund and Tieszen (1994) from the
northern Great Plains. Earlier soil studies included
work of Beavers and Stephen (1958) in Illinois,
Verma and Rust (1969) in Minnesota, and Kurman
(1985) in Kansas, but none of these provide sufficient
data for interregional comparisons.
The following questions are addressed in this
paper:
1. What is the overall diversity of phytoliths in the
main phytolith-producing plants in the interior
Pacific Northwest? Which taxa can be identified
on the basis of their phytoliths and at what level
(e.g., species, genus, family)?
2. What is the pattern of modern phytolith distribu-
tion in soils under present-day vegetation? Can
different vegetation types be distinguished on the
basis of their phytolith record in modern soils?
3. What is the relationship between modern phytolith
assemblages in soils and climate?
2. Area description
Franklin and Dyrness (1988) describe 15 physio-
graphic provinces in Oregon and Washington within
the study area (Fig. 1), of which two are most
important. The Columbia Basin Province is flat and
low (300–600 m elevation) and is covered with
extensive basalt flows of 7–15 Ma (million years BP)
in age (Baker et al., 1991). The Blue Mountains
Province includes the Ochoco, Blue, and Wallowa
mountains and has a more variable relief, ranging from
750 m elevation to 3000 m in the Wallowas. The Pacific
Northwest lies in the path of westerly storm tracks
coming from the Pacific Ocean (Bryson and Hare,
1974). The position of these storm tracks shifts
seasonally from 608N in summer to about 358N
latitude in winter as a result of shifts in the position and
strength of the polar jet stream. Westerly winds during
fall, winter, and spring bring most of the precipitation.
The summers are sunny and dry. The study area,
located on the leeward side of the Cascade Range, sits
in a pronounced rain shadow with distinctly more
continental climate than west of the range. In my study,
the mean annual temperature (MAT) ranged from 0 8C
at subalpine sites in the Wallowas, OR (WA on Fig. 1)
to +12 8C at the Boardman Range, OR (BR on Fig. 1).
The mean annual precipitation (MAP) ranged from
semi-desert (172 mm) at the Boardman Range to 1492
mm at subalpine sites in the Wallowas. The climate data
for the research sites were interpolated using an
appropriate topographic adjustment from datasets in
Thompson et al. (2000).
In the dry interior of the Pacific Northwest,
vegetation zones are often discontinuous, with dis-
tribution determined by available moisture (Franklin
and Dyrness, 1988). The driest sites are located at low
elevations in the western Columbia Basin near the
Cascade Range. The wettest sites are located at high
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9872

elevations on the windward western slopes of the Blue
Mountains (Fig. 1).
Although the landscape today is altered by agri-
culture, the potential natural vegetation (PNV) was
reconstructed by Daubenmire (1968, 1970). The shrub
steppe, represented by the Artemisia tridentata–Agro-
pyron spicatum association, occupies the driest sites.
As precipitation increases to the east, the PNV changes
to dry perennial bunch grassland with A. spicatum–
Poa sandbergii and, higher up, more mesic A.
spicatum–Festuca idahoensis association. The latter
is in turn replaced by F. idahoensis–Symphoricarpos
albus meadow steppe association in the Blue Moun-
tains foothills. Festuca-dominated meadow steppe is
in turn replaced by Pinus ponderosa zone common
between 600 and 1200 m. Abies grandis–Pseudotsuga
menziesii dark coniferous forest zone occupies higher
elevations (900–1500 m). The summits of the Blue
Mountains above 1500 m elevation support Abies
lasiocarpa–Picea engelmannii association. In the
Wallowa Mountains, subalpine parklands of Pinus
albicaulis and A. lasiocarpa and shrub communities of
A. tridentata ssp. vaseyana are found above 2000 m
(WA on Fig. 1). True alpine communities exist at the
highest elevations (N2500 m) with Festuca viridula
grassland being the most common association.
3. Phytolith extraction and data analysis methods
I analyzed phytoliths from 38 species of plants and
58 surface soil samples collected in 1996–1998.
Above-ground biomass of grasses and herbs was
collected to include both vegetative and generative
Fig. 1. Location of modern phytolith sample sites in the interior Pacific Northwest, USA. Physiographic province boundaries are after Franklin
and Dyrness (1988). See Table 4 for the explanation of site codes.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 73

organs. Leaves, needles, and twigs of trees and shrubs
were included, but no attempt was made to collect
underground parts. For each species, about 10 g of dry
plant matter was obtained from a minimum of 3
different plants at each location. No attempt was made
to quantify opal content of various species, although
grasses were clearly the most heavily silicified overall.
Samples of topsoil were collected at 24 locations (Fig.
1). To ensure that samples were collected from
relatively undisturbed vegetation, most sampling was
done in research natural areas, designated wilderness
areas, and the Nature Conservancy and state preserves.
Aggregated soil samples were obtained by ten random
pinches for a total of about 20 g of topsoil (0–2 cm) on
plots of approximately 4 4 m within each commun-
ity. Some litter was present in the topsoil samples.
Some locations had more than one plot thus sampled
for a total of 58 soil samples. Vegetation descriptions
were made in field recording dominant plants and their
percent cover. Each of the samples was then assigned
to one of the eight recognized vegetation types, based
on the composition of dominants.
Phytoliths were extracted using modified wet
oxidation technique of Piperno (1988). Plant material
was carefully rinsed with distilled water, cut into small
fragments of about 5 5 mm, and placed in a sand bath
in glass beakers. Approximately 50 ml of 70% nitric
acid and a pinch of KClO
3
was added. Digestion was
allowed to proceed near boiling point for 1.5 h. The
remaining residue was washed twice with distilled
water, kept in warm hydrochloric acid (10%) for 15–20
min to remove carbonates, washed twice again, and
dried. The phytoliths were stored in ethyl alcohol.
Soil samples were treated using the approach of
Blinnikov (1994). Approximately 10 g of soil was
sieved through a 700-Am mesh sieve to remove
pebbles and large plant fragments. The organic
material was then removed by wet oxidation in 70%
nitric acid for 1.5 h. To remove carbonates, samples
were additionally treated with 10% hydrochloric acid
for 15 min in a warm bath. Samples were then
subjected to deflocculation in 5% sodium hexameta-
phosphate solution for 1 h. Clays were removed by
gravity sedimentation repeated twice (Soil Survey
Laboratory Methods Manual, 1996). The remaining
residue was subjected to heavy liquid flotation
(potassium and cadmium iodide solution and/or
sodium polytungstate) using the density of 2.3 g/
cm
3
. The samples were mixed thoroughly with a glass
rod and centrifuged for 5 min at slow speed
Plate I. Light microscopy digital photographs of the phytolith morphotypes used in the study. Scale bar= 10 Am. Morphotypes 1–13 are grass
phytoliths, 14 and 15 are grass or non-grass and 16–20 are non-grass phytoliths. The morphotype name, 2-letter shorthand, and source species
are listed for each morphotype. All are shown in top view, unless otherwise noted. See Table 2 for full description of grass morphotypes
according to the International Code for Phytolith Nomenclature (Madella et al., 2003):
1. Plate rectangular, straight edges (pr), oblique view, Poa sandbergii.
2. Plate wavy, short — 3–4 undulations (ws), P. sandbergii.
3. Plate wavy, long — 5+ undulations (wl), Calamagrostis rubescens.
4. Long cell, smooth, parallel sides (ls), Festuca idahoensis.
5. Long cell indented (li), Stipa comata.
6. Long cell deeply indented (ld), F. idahoensis.
7. Long cell angular/nonparallel sides (la), Koeleria cristata.
8. Dendritic long cell from seed epidermis (sd) Elymus cinereus.
9. Rondel oval/elongated (ro), oblique view, Agropyron spicatum.
10. Rondel rounded, keeled (rk), side view, F. idahoensis.
11. Rondel rounded, horned (rh), side view, A. spicatum.
12. Rondel pyramidal (rp), side view, S. comata.
13. Bilobate Stipa-type (bs), Oryzopsis hymenoides.
14. Silicified trichome (ht), C. rubescens.
15. Silicified hair or hair base (hh), Sitanion hystrix.
16. Blocky (bl), Picea engelmannii.
17. Spiked of Pinus ponderosa (pp).
18. Epidermal polygonal (ep), Artemisia tridentata.
19. Asterosclereid, Pseudotsuga menziesii, from bother coniferousQmorphotype group.
20. Unevenly thickened cell walls, Larix occidentalis, from bother coniferousQmorphotype group.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9874

M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 75

(approximately 1000 rpm). The floating phytoliths
were collected from the top 5 mm of the solution,
transferred to clean test tubes and sunk by adding
distilled water in proportion of 3 : 1. The phytolith-rich
residue was stored in ethyl alcohol.
Phytoliths were counted in immersion oil type A
(refractive index 1.52) under Zeiss optical microscope
(400) to examine 3D shapes under rotation.
Morphotypes were tallied using the standard approach
of Pearsall (2000). Between 200 and 300 phytoliths
were counted per slide. Phytolith morphotypes were
documented by hand drawings, light microphoto-
graphs, SEM, and permanent reference slides.
All identifiable phytoliths larger than 5 Am were
counted, not only short cells (rondels, bilobates,
polylobates, and saddles), but also long cells and
other grains of identifiable shape. I followed my own
shorthand classification system, modified from Twiss
et al. (1969) and Fredlund and Tieszen (1994) in
describing grass morphotypes, and followed Piperno
(1988) and Bozarth (1993) in describing non-grass
morphotypes (Plate I,Fig. 2). I also provide descrip-
tive code following the recently released Glossary for
the International Code for Phytolith Nomenclature 1.0
(Madella et al., 2003) for each morphotype.
Raw percentages of morphotypes were plotted in
the final diagram for 58 soil samples. Detrended
correspondence analysis (DCA) using PC-ORD
(1997) was performed to examine the indirect environ-
mental gradients implied by the phytolith data (Ter
Braak and Prentice, 1988). The scores for samples and
phytolith morphotypes were shown on the same
scatterplot to illustrate the degree of distinction
between different vegetation types. Boxplots (MINI-
TAB, 1998) were used to further explore variability of
the phytolith percentage data within and between
vegetation types as a form of direct gradient analysis.
To evaluate the relationship between the phytolith
data and climate variables, scatterplots were drafted
in MINITAB (1998) showing the correlation between
selected phytolith types and climate variables. Each
scatterplot has a locally weighted regression
(LOESS) curve to show trends between phytoliths
and climate variables (Cleveland, 1993). Because
climate variables are usually highly correlated with
each other, Principal Components Analysis of the
climate data was performed to identify climate
variables best characterizing regional climate patterns
(Ter Braak and Prentice, 1988). Climate variables for
each site were derived from climate stations’ data
(1951–1980) interpolated onto a 25-km equal-area
grid with a locally fitted regression trend-surface
model with latitude, longitude, and elevation as
predictors (Bartlein et al., 1994). The predicted
climate values at each grid point, the values at sea
Fig. 2. Hand drawing of the phytolith morphotypes used in the
study. Scale bar= 10 Am. Morphotypes 1–13 are grass phytoliths, 14
and 15 are grass or non-grass and 16–20 are non-grass phytoliths.
Morphotypes 1–13 re are shown in top view and side view with
variants, if available. See Plate I for the digital microscopy pictures
of all of these forms and Table 2 for the full description of grass
morphotypes according to the International Code for Phytolith
Nomenclature (Madella et al., 2003): (1) Plate rectangular, straight
edges (pr). (2) Plate wavy, short — 3–4 undulations (ws). (3) Plate
wavy, long — 5 or more undulations (wl). (4) Long cell, smooth,
parallel sides (ls). (5) Long cell indented (li). (6) Long cell deeply
indented (ld). (7) Long cell angular/nonparallel sides (la). (8)
Dendritic long cell from seed epidermis (sd). (9) Rondel oval/
elongated (ro). (10) Rondel rounded, keeled (rk). (11) Rondel
rounded, horned (rh). (12) Rondel pyramidal (rp). (13) Bilobate
Stipa-type (bs). (14) Silicified trichome (ht). (15) Silicified hair or
hair base (hh). (16) Blocky (bl). (17) Spiked of Pinus ponderosa
(pp). (18) Epidermal polygonal (ep). (19) Asterosclereid of
Pseudotsuga menziesii from bother coniferousQmorphotype group.
(20) Unevenly thickened cell walls, Larix occidentalis, from bother
coniferousQmorphotype group.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9876

level (a constant), and the local lapse rates determined
in the regression analysis allowed bilinear interpolation
of the climate values from the four-grid points
surrounding each sample site. The interpolated values
of monthly temperature and precipitation were used
to calculate December–February and June–August
temperature, MAT, MAP, and a few other climate
characteristics. Growing-degree days from 0 and 5 8C
bases (i.e. the accumulated growing-season warmth)
were obtained by calculating pseudo-daily temper-
atures from monthly temperature estimates for each site
and then summing the difference between pseudo-daily
temperatures and temperatures of 0 and 5 8C, respec-
tively (Prentice et al., 1992). The moisture index a(the
ratio of actual evaporation to potential evaporation)
was calculated using either the Priestley–Taylor equa-
tion (Prentice et al., 1992) or the Thornthwaite–Mather
approach (Willmott et al., 1985).
4. Results
4.1. Morphotypes used in the study
Many dominant plant species of the Pacific North-
west contain abundant silica phytoliths (Table 1,Plate
I). I studied dominant grasses and conifers, two
species of upland sedges, a few species of shrubs
and forbs, and one species of fern. All these taxa were
expected to contribute to the phytolith assemblages.
As expected, each of the grasses produces a few
different morphotypes. Most species, however, have
only two or three morphotypes representing about 75%
of the total. In addition to the silicified short and long
cells, grasses also produce abundant silicified hairs and
trichomes, bulliform cells, and scutiform and dendritic
phytoliths that come from seed epidermis.
The short-cell phytoliths are divided into three large
groups: rectangular plates with straight edges (mor-
photype 1 in Tab l e 2 ), plates with sinuous edges, or
wavy, (morphotypes 2 and 3), also known as
bcrenatesQ(Fredlund and Tieszen, 1994), and rondels
(morphotypes 10–13), also known as bshort tra-
pezoidsQ(Brown, 1984), or bhatsQ(Smithson, 1958).
All these phytoliths come from specialized silica cells
in the grass epidermis.
The long cell phytoliths (morphotypes 4–7 in
Table 2) come from non-specialized long cells in the
grass epidermis. Although the overall variation in this
group is high, four morphotypes were distinguished
for convenience: rectangular long cells, indented long
cells, deeply indented long cells, and angular long
cells. The sum of all long cells was also calculated to
simplify the analysis and compare the results with
less-detailed classifications.
Category 8 in Table 2 combines two very different
forms of phytoliths from seed epidermis: dendritic
long cells and scutiform opal. Only the dendritic form
is shown in Fig. 2. Scutiform opal drawings can be
found in Smithson (1958) and Kaplan et al. (1992).
Rondels (bhatsQ, short trapezoids) constitute
another important group of grass phytoliths. They
form in highly specialized short cells in the
epidermis. Rondels were divided into four groups,
following in part Mulholland (1989) and Fredlund
and Tieszen (1994): rondels elongated in top view,
rondels rounded in top view with keel underneath,
rondels rounded with double protrusions, or horns,
underneath, and rondels rectangular in top view and
pyramidal in side view. This classification should be
approached with caution, as there are transitional
types. The sum of all rondels can be additionally
used, because it minimizes the risk of confusing two
similar morphotypes.
Bilobates were found only in the Stipa-group
(Stipa and Oryzopsis) and in Aristida longiseta.Of
these species, the Stipa-group produces primarily
Stipa-type bilobates, which appear bilobate in the
top view, but are actually trapezoids in the side view
(Mulholland, 1989; Fredlund and Tieszen, 1994)with
few true bilobates. Aristida longiseta produces Aris-
tida-type true bilobates with very long shafts, bilobate
in both side and top views. Because no Panicoid
grasses are common in the study area today, presence
of bilobates in soil samples likely indicates presence
of Stipa s.l. or Aristida.
Trichomes (Mulholland, 1989), also called
bpricklesQ(Pearsall, 2000)orbpoint-shaped phyto-
lithsQ(Brown, 1984), are thick-pointed epidermal
appendages developing on the margins of grass
blades. Their base length is equal to three to four
heights of the phytolith (#14 in Ta b le 2). All
trichomes were treated together as a single morpho-
types in this study. Silicified hairs and hair bases
constitute a separate morphotype (#15). These are
longer and thinner appendages (height Hbase length)
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 77

than trichomes. While trichomes are primarily found
in grasses, Asteraceae and some other forbs can
produce multicellular trichomes and microhairs.
Non-grass forms comprise the remaining mor-
photypes and will be discussed in greater detail
below.
Table 1
Dominant trees, shrubs, grasses, sedges, and herbs of the Pacific Northwest analyzed for phytoliths
Species Phytoliths
present?
Morphotypes
a
Reference
TREES
Abies amabilis No
Abies grandis No
Abies lasiocarpa Yes 16, 18, tracheids Klein and Geis (1978),Bozarth (1993)
Alnus sinuata Yes Anticlinal epidermis
Juniperus occidentalis Yes Tracheids
Larix occidentalis Yes 2 0 Norgren (1973),Carnelli et al. (2004)
Picea engelmanni Yes 16, 18, sinuous Norgren (1973),Bozarth (1993)
Pinus albicaulis No
Pinus contorta Yes 1 8
Pinus ponderosa Yes 1 7 Norgren (1973),Kearns (2001)
Pseudotsuga menziesii Yes 19 Norgren (1973),Klein and Geis (1978)
Thuja plicata No
SHRUBS
Artemisia rigida Yes 1 6
Artemisia tridentata Yes 16, 18, tracheids
Chrysothamnus nauseosus Yes Segmented hairs
Chrysothamnus viscidiflorus Yes Segmented hairs
Purshia tridentata No
Tetradymia canescens No
GRAMINOIDS (SUBFAMILY shown for grasses)
Agropyron dasystachyum POOIDEAE Yes See Table 2 Blackman (1971),Brown (1984)
Agropyron spicatum POOIDEAE Yes See Table 2 Norgren (1973),Brown (1984)
Aristida longiseta ARUNDINOIDEAE Yes See Table 2 Brown (1984),Mulholland (1989)
Bromus tectorum POOIDEAE Yes See Table 2
Calamagrostis rubescens POOIDEAE Yes See Table 2 Blackman (1971),Norgren (1973)
Carex geyerii Yes 4, 7, sedge conical Norgren (1973),Ollendorf (1992)
Carex rossii Yes 4, 7, sedge conical Ollendorf (1992)
Elymus cinereus POOIDEAE Yes See Table 2 Norgren (1973)
Festuca idahoensis POOIDEAE Yes See Table 2 Blackman (1971),Norgren (1973)
Festuca viridula POOIDEAE Yes See Table 2
Koeleria cristata POOIDEAE Yes See Table 2 Brown (1984),Mulholland (1989),
Oryzopsis hymenoides STIPOIDEAE Yes See Table 2 Brown (1984)
Poa sandbergii POOIDEAE Yes See Table 2 Blackman (1971)
Sitanion hystrix POOIDEAE Yes See Table 2 Norgren (1973),Brown (1984)
Stipa comata STIPOIDEAE Yes See Table 2 Norgren (1973),Barkworth (1981)
Stipa occidentalis STIPOIDEAE Yes See Table 2 Barkworth (1981)
Stipa thurberiana STIPOIDEAE Yes See Table 2 Barkworth (1981)
FORBS
Balsamorrhiza spp. Yes Segmented hairs Bozarth (1992)
Lupinus sericeus Yes Rough blocky, hairs
Pteridium aquilinum Yes Epidermal anticlinal, tracheids
Plant nomenclature follows Hitchcock and Cronquist (1994).
a
Numbers correspond to the morphotypes shown on Fig. 2.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9878

4.2. Phytoliths in grasses
Fifteen grass species common to the interior
Pacific Northwest were examined in this study (Table
3). All produce abundant phytoliths and are discussed
below in the alphabetical order of their scientific
names.
Thick-spike wheatgrass Agropyron dasystachyum
Hook (Scribn.) is an important dominant in a rare
grassland association (A. dasystachyum–Oryzopsis
hymenoides) on exceedingly dry soils in north-central
Oregon, e.g., in The Nature Conservancy’s Lindsay
Prairie Preserve. This species has a large diversity of
Festucoid phytoliths with a mixture of wavy plates,
long cells, and rondels (Table 3). Its assemblage also
contains dendritic and scutiform phytoliths from seed,
also common in cultivated grains with large inflor-
escences, e.g., barley and wheat.
Table 2
Morphotypes recognized in this study described according to the International Code for Phytolith Nomenclature (Madella et al., 2003)
Name and reference
number used
in this study (Fig. 2)
Acronym Shape descriptor Texture and ornamentation Anatomical origin
1. Plates rectangular with
straight edges
pr Parallelepiped/trapeziform in 3D,
rectangular in 2D
Smooth (psilate),
linear ridges common
Epidermal short cell
2. Short wavy ws Trapeziform in 3D,
lobateb4 lobes on each side in 2D
Smooth (psilate) top,
sinuate margin
Epidermal short cell
3. Long wavy wl Trapeziform in 3D,
lobateN4 lobes on each side in 2D
Smooth (psilate) top,
sinuate margin
Epidermal short cell
4. Long cells rectangular lc Elongate, flat, parallel sides Smooth (psilate) to
mildly aculeate
Epidermal long cell
5. Long cells indented li Elongate, flat Aculeate Epidermal long cell
6. L. c. deeply indented ld Elongate, flat Echinate
(N1/2 the width of cell)
Epidermal long cell
7. L. c. angular la Elongate, flat,
non-parallel sides
Smooth (psilate) Epidermal long cell
8. Dendrictic from seed sd Elongate, flat Dendriform Epidermal long cell
Scutiform from seed sd Scutiform Papillate Papillae cell
9. Oval rondel ro Trapeziform in 3D,
oblong to oval in
top view 2D
Smooth (psilate) Epidermal short cell
10. Keeled rondel rk Trapeziform in 3D,
oval in top view 2D
Smooth (psilate) top,
keeled bottom
Epidermal short cell
11. Horned rondel rh Same as 10, but with
horned-like protrusions
near bottom
Smooth (psilate) top,
horned bottom
Epidermal short cell
12. Pyramidal rondel rp Truncated pyramidal
3D, square top view 2D
Smooth (psilate) Epidermal short cell
13. Bilobate of Stipa-type bs Trapeziform in 3D Smooth (psilate) Epidermal short cell
14. Silicified trichome ht Acicular/lanceolate Smooth (psilate) Trichome
15. Silicified hair/base hh Cylindric/conical Smooth (psilate) Hair cell or hair base
16. Blocky bl Cubic or globose Smooth (psilate) or cavate Unknown, probably
epidermal or mesophyll
17. Spiked of P. ponderosa pp Clavate Tuberculate Epidermal?
18. Epidermal polygonal ep Irregular polygonal
in top view, very
flat on the side
Smooth Epidermal, non-grass
Epidermal anticlinal ea Anticlinal in top view,
very flat on the side
Smooth to sinuate margins Epidermal, non-grass
19. Asterosclereid of P. meniesi co Very large (N200 Am),
star-shaped
Striate, very dark because
of occluded carbon
Vascular cell
20. Uneven cells Larix-type co Cuneiform Facetate Epidermal cell
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 79

Tab le 3
Percentages of different phytolith morphotypes in grasses of the Pacific Northwest. All counts rounded to the nearest percent of the total (0% for values b0.5%). Morphotype numbers correspond to Fig. 2. Nomenclature follows Hitchcock and
Cronquist (1994)
Morphotypes
Species
1 2 3 4 5 6 7 Sum of all
long cells
(4 +5 + 6 + 7)
8 9 10 11 12 Sum of all
rondels
(9+10+11+12)
13 14 15 Other
types
Total
number
of phytoliths
counted
Plates
rectangular
with
straight
edges
Plates
short
wavy
Plates
long
wavy
Long cells
rectangular
Long
cells
indented
Long
cells
deeply
indented
Long
cells
angular
Dendritic
and
scutiform
(seeds)
Rondels
elongated
Rondels
rounded
keeled
Rondels
rounded
horned
Rondels
pyramidal
Bilobates Trichomes Hairs
and
hair-bases
Agropyron
dasystachyum
6 8 6 6 4 0 1 11 31 10 0 0.5 22 33 0 0 4 0 202
Agropyron
spicatum
10 5 0 6 9 7 9 31 1 18 2 0.5 27 48 0 1 3 1 616
Aristida longiseta 0 1 0 0 2 0 0 2 0 0 7 12 0 19 67 4 4 3 223
Bromus tectorum 4 24 0 2 0 9 0 11 21 1 0 0 28 29 0 2 8 1 305
Calamagrostis
rubescens
10 17 20 15 4 0 0 20 0 0 4 0 0 4 0 26 2 1 414
Elymus cinereus 3 14 7 17 14 1 1 34 6 3 7 3 0 14 0 14 6 3 236
Festuca
idahoensis
6 2 0 1 23 10 0 34 0 0 9 36 1 47 0 3 3 4 295
Festuca viridula 4 1 0 0 41 0 0 43 0 0 20 11 5 37 0 13 1 0 152
Koeleria cristata 2 37 7 11 0 0 0 12 0 1 10 0 0 11 0 24 2 6 315
Oryzopsis
hymenoides
5 0 0 1 2 0 0 3 1 0 5 6 28 39 47 1 4 0 209
Poa sandbergii 8 56 6 0 4 0 1 6 0 0 17 2 0 18 0 3 0 2 324
Sitanion hystrix 10 12 0 7 1 6 3 17 0 3 11 0 9 23 0 0 35 4 222
Stipa comata 3 0 0 7 2 2 0 11 0 0 4 36 23 64 11 6 3 1 228
Stipa occidentalis 0 12 0 2 0 0 0 3 0 0 0 9 37 46 35 1 2 0 348
Stipa thurberiana 6 4 0 3 6 3 0 12 0 0 3 8 4 14 51 5 7 1 237
All counts rounded to the nearest percent of the total (0% for values b0.5%).
Morphotype numbers correspond to Fig. 2. Nomenclature follows Hitchcock and Cronquist (1994).
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9880

Bluebunch wheatgrass (Agropyron spicatum
(Pursh) Scribn. and Smith) is one of the three most
common grasses in the Columbia Basin native grass-
land. It dominates dry Artemisia tridentata–A. spica-
tum sagebrush steppe association and slightly wetter A.
spicatum–Festuca idahoensis grassland association
(Daubenmire, 1970). Agropyron spicatum produces
smooth or indented long cells (brodsQ), and some
elongated rondels (Norgren, 1973). Deeply indented
and angular long cells may help distinguish this
species from other species, except Sitanion hystrix.
These two species are in fact closely related and even
hybridize (Hitchcock and Cronquist, 1994). The over-
all proportion of the long cells in S. hystrix is only half
that of A. spicatum.
Red threeawn (Aristida longiseta Steud.) is a rare
C
4
species in the study area. It contains over 66% of
bilobate phytoliths. Many of these are of the
diagnostic Aristida-type with very long shafts (Mul-
holland, 1989), that had also been recently reported
from Aristida species in Australia (Bowdery, 1998)
and South America (Piperno and Persall, 1998).
The non-native cheatgrass (Bromus tectorum L.)
can be readily distinguished from native grasses on
the basis of the high content of hairs (8%) and seed
phytoliths (21%) that come from inflorescences.
Cultivated annual grains and annual Bromus species
with relatively large inflorescences are likely to have a
high proportion of such forms as well (Ball et al.,
1999).
The dominant grass of pine forests in the study area
is pinegrass (Calamagrostis rubescens Buckl.). It
produces wavy plates with more than five undulations
(20%). This morphotype (#3 in Fig. 2) is characteristic
of the genus, similar phytoliths were found in
Calamagrostis sp. in Russia (Blinnikov, 1994).
Another abundant morphotype is trichomes (26%),
but it is not genus-specific. Calamagrostis rubescens
also has a large proportion of rectangular long cell
phytoliths (15%), short wavy plates (17%), and
rectangular plates (10%). It produces few rondels
(b5%) or hairs (b2%). Because this species is
common in pine forests today but not found in
grasslands, it can be an important indicator of the
past forested environments.
Giant wildrye (Elymus cinereus Scrib. and Mer.)
has limited distribution today in small seasonally wet
depressions in non-forested areas. It produces diverse
phytoliths making it hard to distinguish from other
grasses, especially closely related Agropyron and
Sitanion.Elymus assemblage has a high proportion
of long cells (34%), including smooth rectangular
(17%) and indented (14%) morphotypes. However,
deeply indented and angular forms, typical of A.
spicatum, are uncommon in this species.
Idaho fescue (Festuca idahoensis Elmer) is a
dominant species of the moderately dry Agropyron
spicatum–F. idahoensis grassland, and it also dom-
inates the more mesic F. idahoensis–Symphoricarpos
meadow steppe (Daubenmire, 1970). Agropyron and
Festuca can be distinguished based on their rondels.
Agropyron spicatum contains mostly elongated and
pyramidal forms of rondels (#9 and 12), while F.
idahoensis contains mostly horned rondels of #11,
similar to bconicalQphytoliths of Fredlund and
Tieszen (1994) (Table 3). Festuca idahoensis also
produces long cell phytoliths (34%), most of them
indented (23%). In contrast to A. spicatum, angular
long cells appear to be absent from F. idahoensis,
which may help differentiate these two species
further.
Green fescue (Festuca viridula Va se y ) i s a n
important grass of the alpine zone in the study area. It
is one of the least silicified species and requires further
research. Preliminary results show that F. viridula is
similar to Festuca idahoensis in overall phytolith
composition. Most phytoliths are indented long cells
(41%) and rounded keeled (20%) and horned (11%)
rondels.
Junegrass (Koeleria cristata Pers.) is a species
with circumpolar distribution. In the study area, it
co-dominates with Festuca idahoensis in the F.
idahoensis–Symphoricarpos meadow steppe. Its
common phytoliths are wavy plates with 3 or 4
undulations. Some of these forms with diagonally
slanted ends appear to be diagnostic (Kearns, 2001).
In addition, K. cristata contains about 24% silicified
trichomes, but few long cells (b4%) or rondels
(b11%).
The needlegrass species from the Stipae tribe
(Oryzopsis hymenoides (R. and S.) Ryker, Stipa
occidentalis Thurb., Stipa thurberiana Piper) are
readily distinguished from other grasses on the basis
of their Stipa-type bilobates (36–51%). These species
grow in dry open grasslands or subalpine meadows,
typically on soils with low effective moisture content.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 81

Stipa-type phytoliths in soils may thus indicate dry
conditions. These three species contain few long cells
or rectangular plates (b12% each). Stipa comata Trin.
and Rupr. belongs to a different section than S.
occidentalis and S. thurberiana (Barkworth, 1981).
Its phytoliths are primarily horned and pyramidal
rondels (36% and 23%, respectively) with a lower
percentage of bilobates (11%).
Sandberg bluegrass (Poa sandbergii Va s ey ) i s
found in dry grasslands and scablands with Agropyron
spicatum–P. sandbergii,Artemisia tridentata/A. spi-
catum and A. tridentata/P. sandbergii associations
(Daubenmire, 1970). Poa sandbergii contains a large
proportion of short wavy forms (56%, #2 in Fig. 2).
Usually these phytoliths have more pronounced
undulations than those of Koeleria cristata and
pointed rather than diagonally slanted ends. Poa
sandbergii has 18% rondels, mainly rounded keeled
morphotype #10.
Squirreltail (Sitanion hystrix Nutt. (Smith)) grows
in disturbed grasslands in the study area. Its phytolith
composition resembles that of the related Agropyron
spicatum,Agropyron dasystachyum,andBromus
tectorum. It produces a high share of silicified hairs
(N34%) and a lower share of wavy forms (12%).
Although S. hystrix cannot be unambiguously identi-
fied, high proportion of silicified hair phytoliths in
soil assemblages may indicate its presence.
4.3. Phytoliths in non-grasses
Twenty-three species of non-grasses were analyzed
for phytoliths, including common trees, shrubs and
forbs in the study area (Table 1). Of these species, 17
contained silicified material. Among the conifer tree
species (Pinaceae family), ponderosa pine (Pinus
ponderosa Dougl.) and Douglas-fir (Pseudotsuga
menziesii Mirbel (Franko)) contain abundant diagnos-
tic phytoliths. Pinus ponderosa produces spiked
forms in its needles (Norgren, 1973; Kearns, 2001;
cf. Pinus banksiana form in Bozarth, 1993). It also
produces potentially confusing wavy phytoliths,
similar to wavy plates of grasses (#2 and 3 in Fig.
2,Klein and Geis, 1978). Pseudotsuga menziesii
contains large branched silicified asterosclereids
(Norgren, 1973; Klein and Geis, 1978).
Other common pines in the region, for example,
lodgepole (Pinus contorta Dougl.) and whitebark
(Pinus albicaulis Engelm.) do not appear to produce
phytoliths of any kind, although some cell wall
silicification was observed in P. contorta. Among
other conifers, two species with the greatest potential
to be recognized in soils are western larch (Larix
occidentalis Nutt.) and Engelmann spruce (Picea
engelmannii Parry). Larch needles contain abundant
cell wall phytoliths of uneven thickness (cf. Larix
decidua in Carnelli et al., 2004). Spruce features
polyhedral endodermal cells (blocky phytoliths),
found also in other spruces and firs (Klein and Geis,
1978; Bozarth, 1993), and flat epidermal forms
characteristically sinuous on all four sides (cf. Picea
glauca in Bozarth, 1993).
Of the three examined fir species (Pacific silver fir
Abies amabilis (Dougl.) Forbes, grand fir Abies
grandis (Dougl.) Forbes, and subalpine fir Abies
lasiocarpa (Hook.) Nutt.), only the latter contains
identifiable phytoliths—endodermal blocky #16, elon-
gated blocky forms (#18), flat polygonal epidermal
morphotype #19, and some silicified tracheids (Klein
and Geis, 1978).
Big sagebrush Artemisia tridentata Nutt. and stiff
sagebrush Artemisia rigida (Nutt.) Gray (Asteraceae)
are the most common sagebrush species of the
Columbia Basin (Daubenmire, 1970). They produce
similar phytoliths: epidermal flat polyhedral cells,
blocky forms, silicified tracheids, and a limited
amount of silicified hairs. Another common dominant
shrub Chrysothamnus, or rabbitbrush, is poorly
silicified. Two species Chrysothamnus nauseosus
Pall. (Britt.) and Chrysothamnus viscidiflorus (Hook.)
Nutt. produce limited amount of segmented hairs,
common in other Asteraceae. Phytoliths were not
found in bitterbrush Purshia tridentata (Pursh) DC.
(Rosaceae) or Tetradymia canescens DC. (Astera-
ceae). Sitka alder (Alnus sinuata (Regel) Rydb.,
Betulaceae) contains silicified epidermal cells with
anticlinal (bjigsaw puzzleQ) walls.
Two upland sedges (Carex geyerii Boott and Carex
rossii Boott) contain abundant conical phytoliths,
diagnostic of genus Carex and Cyperaceae family
(Ollendorf, 1992), as well as some smooth elongated
forms similar to the rectangular long cells of grasses.
The balsamroot, Balsamorhiza spp., widespread in
the Columbia Basin, produces distinct segmented
hairs, diagnostic of Asteraceae. Another common forb,
silky lupine (Lupinus sericeus Pursh, Fabaceae),
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9882

produces thin silicified hairs and also verrucate blocky
forms (not shown). The only examined fern (Pteridium
aquilinum L.) produces abundant anticlinal epidermal
phytoliths (bjigsaw puzzleQ) and silicified tracheids.
4.4. Phytoliths in modern soils
Fig. 3 illustrates the range of regional variability in
the phytolith record from 58 modern soil samples.
Tab le 4 provides information on the geographic
coordinates, elevation, and vegetation composition
for each sample. The samples were grouped into eight
vegetation types (see Table 4 footnote for details). The
vegetation types are necessarily subjective, but they
generally correspond to the common associations
found in the study area (Daubenmire, 1970; Franklin
and Dyrness, 1988).
Samples from sagebrush steppe were distinguished
by high percentages of blocky forms (24 F4%, n=11)
and epidermal polygonal phytoliths (17 F3%), both
probably produced by Artemisia tridentata, but the
latter may also be from other dicot shrubs or herbs.
Anticlinal epidermal cells, and some Asteraceae
phytoliths (e.g., perforated plates and segmented hairs,
Bozarth, 1992) were also found in a few samples.
Among grass phytoliths, the most common were
different kinds of long cells (18 F4%) probably
contributed by Agropyron spicatum. Short wavy
phytoliths with 3–4 undulations (#2 on Fig. 2) were
from Poa sandbergii (12 F4%). Sample 3 (A.
tridentata/P. sandbergii shrub steppe) and sample 9
(Artemisia rigida/P. sandbergii shrub steppe) con-
tained about 30% of this morphotype, which is
consistent with P. sandbergii being the only grass
present in the sampling plot. Dendritic and scutiform
phytoliths from seed epidermis were present in
samples where Bromus tectorum cover was high.
Samples from grasslands with Stipa or Oryzopsis
species present were easily distinguished from other
vegetation types by the presence of Stipa -type
bilobates (9 F2%, n=9). In other samples, this
morphotype (#13, Fig. 2) never exceeds 3%. Other
common grass morphotypes are long cells (23 F4%)
and rondels (17 F6%). Of the former group, rectan-
gular smooth were the most common morphotype
(common in Stipa comata). Among the rondels, the
most common morphotype was rounded horned rondel
(#11), which is also common in S. comata (N36%),
and well represented in other species of Stipa. This
morphotype, however, is also found in Festuca,soit
cannot be a reliable indicator of Stipa or Oryzopsis
presence. Blocky forms were also found in soil
samples from this vegetation type, but in lower
percentages than in those from sagebrush steppe
(4 F0.5%). Most of these phytoliths probably came
from Artemisia, because the Stipa grassland sampling
sites were located near Artemisia stands. As expected,
dendritic and scutiform phytoliths from seeds (#8)
were found in the samples with high presence of
Agropyron dasystachyum and/or Bromus tectorum.
There also was a large proportion of both epidermal
polygonal (12 F3%) and anticlinal (3 F2%) phyto-
liths in this vegetation type. Rabbitbrush (Chryso-
thamnus spp.) and some Asteraceae forbs, e.g.,
Balsamorhiza and Centaurea, were the likely con-
tributors of these phytoliths.
The single sample of bother grasslandQ(sample 21,
Fig. 3) came from a dense patch of Elymus cinereus.
The most abundant forms in the Elymus grassland
were the long cells (34%) and trichomes (14%),
consistent with the species’ assemblage (see above).
Samples from Agropyron grassland contain equal
proportions of short wavy morphotype (23 F5%),
long cells (26 F6%), and rondels (22 F4%). The first
morphotype (#2, Fig. 2) was apparently contributed
by Poa sandbergii, while the other two came from
Agropyron spicatum and, when present, Festuca
idahoensis. Rectangular plates were more abundant
in samples from Agropyron grassland than in those
from sagebrush steppe or Stipa grassland. Among
non-grass phytoliths, blocky and epidermal polygonal
morphotypes were present in samples near sagebrush
steppe or forest boundary.
Samples from Festuca grassland contained the
same morphotypes as those of the Agropyron
grassland, but could be distinguished based on
different phytolith proportions. Short wavy morpho-
type (#2) made up 12 F3% (n= 9), long cells (#4–7)
26 F8%, and rondels (#9–12) 31 F8%. The pro-
portion of wavy phytoliths thus was lower (due to
less Poa in this grassland), while the proportion of
rondels was higher, reflecting the added contribu-
tions of Festuca and Agropyron. When individual
rondel and long cell morphotypes were examined
(Fig. 3), samples from Festuca-dominated grasslands
could be differentiated from those from Agropyron-
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 83

2
4
6
8
10
12
14
16
18
20
22
24
26
28
30
32
34
36
38
40
42
44
46
48
50
52
54
56
58 1000 2000 3000
ma.s.l.
Elevation
20
Rectangularplates
20 40
Wavy/lobed,slightly
andwith3-4undul.
20
Wavy/lobedwith
5ormoreundul.
20 40 60
Longcells
20
Dendriticandscutiform
20 40
Rondels
20
Stipa-typebilobates
20
Trichomes
20
Hairsandhairbases
20
Conical(Carex)
20 40 60
Blocky
20 40
Epidermalpolygonal
20
Epidermalanticlinal
20 40
Spiked(P.ponderosa)
20 40 60
Other conifer
Asteraceae
Veg.Type
Code
SS
SG
OG
AG
FG
PF
OF
SF
ModernPhytolith Assemblages from the PNW:Morphotype Percentages
BR BR
U2 LP
SH KR
KR SH
LG WF
RS LP
LP
LP
LP SH
WA
WA WA
WA
LP LP
U4 SH
SH LG
LGMA
HR PA
U1 U5
LG HR
HR PA
PA
PA
WA SC
MR WS
WS
PA WF
RS
RS U6
WA U3
U3 FG
DL DL
DT
DT DT
DT
SiteSampleID
Fig. 3. Percentage of phytolith morphotypes in 58 modern soil samples from the interior Pacific Northwest. Vegetation codes: SS — sagebrush steppe, SG—Stipa grassland, OG —
Other (Elymus) Grassland, AG — Agropyron Grassland, FG — Festuca Grassland, PF — Pinus ponderosa Forest, OF — other (Abies grandis and Pseudotsuga menziesii) forest, SF —
Subalpine Forest. See Table 4 for the explanation of site codes and sample ID numbers.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9884

Table 4
Surface sample sites in the Pacific Northwest
Site
code
Site name Latitude
(N)
Longitude
(W)
Elevation
(m)
Vegetation Sample numbers
(vegetation type
a
)
HR Horse Ridge Research Natural
Area (RNA), OR
43856V00W121802V00W1230 Juniperus/Artemisia/
Agropyron–Festuca woodland
29 (AG) 34,
35 (FG)
FG Frog Camp Rd., Willamette
National Forest, OR
44812V00W121852V30W1475 Pinus contorta–Abies
lasiocarpa forest
52 (SF)
WS West of Sisters, OR, on Hwy. 242 44818V00W121837V00W1025 Pinus ponderosa/Agropyron forest 42, 43 (PF)
SC Stevens Canyon Rd.,
Deschutes National Forest, OR
44821V00W121832V00W1000 Pinus ponderosa/Festuca forest 40 (PF)
MR Metolius RNA, Deschutes
National Forest, OR
44830V00W121837V30W1000 Pinus ponderosa/Festuca forest 41 (PF)
SH Sheep Rock Trail, the
Peninsula, OR
44831V30W121817V30W700 Juniperus/Artemisia/
Agropyron–Festuca–Stipa woodland
24, 25 (AG),
16 (SG)
LG Lawrence Memorial Grassland
Preserve (TNC), OR
44856V30W120848V00W1020 Agropyron–Festuca grassland
and Artemisia rigida–Poa scabland
9 (SS), 26, 27
(AG), 33 (FG)
MA Hwy. 97 near Madras, OR 45810V00W120850V00W1000 Artemisia –Agropyron shrub steppe 28 (AG)
WA Mt. Howard, Eagle Cap
Wilderness Area, OR
45815V30W117810V30W1710, Pinus ponderosa forest, 49 (PF). 17,
18, 19, 20 (SG),
2450 Festuca viridula and Stipa
occidentalis grasslands
39 (FG)
LP Lindsay Prairie Preserve
(TNC), OR
45835V30W119839V00W275 Agropyron dasystachyum–
Oryzopsis grassland
12, 13, 14, 15
(SG) 22 (AG)
Artemisia–Stipa shrub steppe, 4 (SS)
Elymus grassland 21 (OG)
WF Rd. 46 at Rd. 080,
Wallowa-Whitman National
Forest, OR
45837V00W117813V00W1275 Artemisia–Poa scabland, 10 (SS)
Pinus ponderosa–
Calamagrostis forest
44 (PF)
BR Boardman RNA, OR 45844V00W119838V00W170 Artemisia–Agropyron shrub steppe 1,2 (SS)
U6 Robinette Mtn.,
Columbia Co., WA
46805V00W117852V00W1450 Pinus ponderosa forest 48 (PF)
U3 Rd. 64 at Chase Mtn., Umatilla
National Forest, WA
46805V30W117851V30W1450 Abies grandis–Pseudotsuga forest 50, 51 (OF)
DT Devil’s Tailbone Rd.,
Umatilla National Forest, WA
46809V00W117828V00W1750 Abies lasiocarpa forest 55, 56, 57,
58 (SF)
U4 Prairie in Walla Walla Co., WA 46812V00W118815V00W500 Agropyron–Festuca grassland 23 (AG)
U5 Bundy Hollow Cemetery, WA 46812V00W118805V00W650 Festuca grassland 32 (FG)
PA Pataha River Canyon, Umatilla
National Forest, WA
46816V00W117831V30W1400 Festuca–Koeleria grassland 36, 37, 38 (FG)
Agropyron–Festuca grassland 30 (AG)
Pinus ponderosa–Calamagrostis forest 45 (PF)
RS Rose Springs, Umatilla
National Forest, WA
46816V30W117833V00W1400 Artemisia rigida–Poa shrubland 11 (SS)
Pinus ponderosa–Carex forest 46, 47 (PF)
U2 Sagebrush Scrub, Benton Co., WA 46817V00W119830V00W250 Artemisia/Agropyron shrub steppe 3 (SS)
KR Kahlotus Ridgetop Natural Area, WA 46842V00W118833V00W475 Artemisia/Agropyron–Festuca 6, 7 (SS)
SH Sand Hills, near Hampton Rd., WA 46847V30W118831V00W450 Artemisia shrub steppe 5, 8 (SS)
U1 Dry Prairie, Adams Co., WA 46850V00W118830V00W450 Agropyron–Festuca grassland 31 (FG)
DL Dewey Lake, near Mt. Rainier
National Park, WA
46851V15W121829V00W1550 Tsuga mertensiana–
Abies lasiocarpa forest
53, 54 (SF)
Site codes correspond to Fig. 1.
a
Eight vegetation types were distinguished based on the dominant species found within the 44 m plots: SS = Artemisia-dominated
Sagebrush Steppe, AG = Agropyron-dominated Grassland, FG = Festuca-dominated Grassland, SG= Stipa /Oryzopsis-dominated Grassland,
OG = Other Grassland (Elymus-dominat ed), PF = Pinus ponderosa Fore st, OF = Other Forest (Abies grandis–Pseudotsuga menziesii),
SF = Subalpine Forest (Abies lasiocarpa and/or Tsuga mertensiana-dominated).
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 85

dominated grasslands based on the higher percent-
age of rounded horned rondels (12 F3% versus
1F0.5%), and higher percentage of indented long
cells (14 F6% versus 6 F1%). One sample from
this vegetation type (39) came from an alpine stand
of Festuca viridula growing at 2550 m elevation,
and was very different in its composition. Additional
samples from other high-elevation sites are neces-
sary to more reliably identify such alpine commun-
ities in the soil phytolith record. The occurrence of
conifer phytoliths in some samples of Festuca-
dominated grassland can be explained by proximity
to Pinus ponderosa forest zone.
Pinus ponderosa forestwasidentifiedbythe
presence of the diagnostic spiked morphotype of P.
ponderosa (#21, Fig. 2). Spiked phytoliths were
abundant in soils under present-day P. ponderosa
forest (28 F6%, n= 10). Two groups of samples can
be distinguished within this vegetation zone (Fig. 3):
samples from low-elevation dry forests of P. ponder-
osa/Agropyron spicatum and P. ponderosa/Festuca
idahoensis associations (samples 40–44), and samples
from more mesic P. ponderosa/Calamagrostis rubes-
cens association at higher elevations (samples 45–49).
The former group had few wavy or rectangular
phytoliths, but abundant rondels and long cells. The
latter group had a high proportion of wavy forms,
including those diagnostic of Calamagrostis,and
trichomes.
Two samples were taken from bOther ForestQ
vegetation type, sample 50 from an almost pure Abies
grandis stand and sample 51 from A. grandis–
Pseudotsuga menziesii forest with some Pinus pon-
derosa nearby. Unstudied forest grasses probably
contributed their phytoliths to both samples.
Subalpine forest samples could be readily distin-
guished from those from other vegetation types by the
high proportion of blocky forms (probably Picea and
Abies,9F3%, n= 7), epidermal polygonal phytoliths
(13 F2%) probably contributed by Abies, and a suite of
bother coniferQphytoliths (41 F22%). The bother
coniferQsuite included silicified tracheids, phytoliths
with unevenly thickened cell walls of Larix, and
phytoliths with four sinuous sides (Klein and Geis,
1978; Bozarth, 1992) from Picea engelmannii and
Abies lasiocarpa. The variability in the bother coniferQ
group was quite high, and further research of the
phytoliths from subalpine forests is needed. Sam-
ple 54 had a high proportion of conical phytoliths
of Carex (15%) and some Stipa-type bilobates.
This sample came from a subalpine bog on the
shore of Dewey Lake near Mt. Rainier National
Park. Stipa occidentalis and Carex both grew at
the site.
Detrended correspondence analysis (DCA) was
performed in PC-ORD (1997) on 58 modern soil
samples to test how well vegetation types could be
differentiated based on phytolith assemblages, and if
any gradients could be extracted from the data by this
indirect method. In the resulting scatterplot (Fig. 4,
only first two axes are shown) samples from all
vegetation types grouped together well, except sam-
ples from bother forestsQ, which appeared similar to
grasslands. The main reason for this was lack of
phytolith-producing conifers in this community,
which made phytoliths from grasses excessively
important. Artemisia shrub steppe, Pinus ponderosa
forests and subalpine forests appeared as the most
distinct vegetation types.
The other four vegetation types were less well
differentiated. Samples from Stipa-dominated grass-
lands contained high proportion of bilobate Stipa
phytoliths. Agropyron spicatum-dominated grasslands
contained high percentage of long cells and wavy
forms, while those from Festuca-dominated grass-
lands were distinguished by the high proportion of
rondels and silicified hairs.
Axis 1 in Fig. 4 can be interpreted as an artificial
gradient stretching from the assemblages with the high
proportion of spiked morphotype (#21 on Fig. 2) to the
assemblages with the high proportion of the bother
coniferQmorphotypes or blocky forms, with grassland
assemblages (usually N80%) taking up the middle.
Axis 2 apparently represents the main environmental
moisture gradient extending from very dry commun-
ities (assemblages from Artemisia steppe) near bottom
through moderately dry grasslands to mesic Pinus
ponderosa forests, and finally to the moist subalpine
forests. The exceptions to this pattern are the samples
from the more mesic Festuca idahoensis-dominated
grasslands that were placed below, not above, drier
Agropyron spicatum and Stipa-dominated grasslands
on Axis 2.
Boxplots allow further direct examination of the
phytolith data by plotting each morphotype abun-
dance for the eight vegetation types (Fig. 5).
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9886

4.5. Phytoliths and climate
Interpolated values of 5 selected environmental
variables from 30 sampling sites provide a basis for
direct phytolith–climate comparisons. Principal Com-
ponents Analysis was performed on the climate
variables and elevation to explicitly characterize
climatic gradients between the sites and to reduce
the number of variables for the subsequent analysis.
The first two components resulting from PCA
explained 90.8% of the variance in the data. PC1
(72.5% of the total variation) represented the main
environmental gradient in the data dependent on
elevation. As expected, a strong negative correlation
was found between elevation and temperature, and a
strong positive one between elevation and precipita-
tion. Loadings for PC1 were highest for growing-
degree days above 0 8C (0.254), mean annual
temperature (0.253), mean July temperature (0.253),
June–August temperature (0.253), and mean temper-
ature of the warmest month (0.253). Loadings were
lowest for June–August precipitation (0.251), July
precipitation (0.248), and elevation (0.250).
PC2 appeared to describe seasonality of precip-
itation (18.3% of the total variance). PC2 loadings
were highest for July precipitation to annual precip-
itation ratio (0.442) and January–July temperature
range (0.351), and were lowest for the January
precipitation to annual precipitation ratio (0.423),
and the absolute minimum temperature (0.405).
Five variables were then chosen for subsequent
analysis against phytolith data, elevation, mean
annual temperature (MAT), mean annual precipitation
(MAP), number of growth degree-days with temper-
atures above 0 8C, and the moisture index of the ratio
of actual evaporation to potential evaporation
(Thornthwaite–Mather method). The first 3 variables
highlight the main environmental gradient apparent in
PC1, while the latter two were chosen as useful
bioclimatic variables representing two main parame-
ters essential for plant growth: amount of useful
energy and amount of available moisture (Thompson
et al., 2000).
Scatterplots (MINITAB, 1998) display abundan-
ces of 13 selected morphotypes (Fig. 6) against the
five selected environmental variables. Rare morpho-
Stipa or Oryzopsis grassland
P.menziesii-A.grandis forest
Axis 1, SD
Axis 2, SD
34
29
3
52
42
43
40
41
24
25
16
26
33 9
27
28
49
39
17 18
19
20
22
12 13
14
21
15 4
10
44
1
2
50
57
58
55
56
36 37
38
45
30
46
47
11
6
7
5
8
53
54
31 3
23
32
48
Pinus ponderosa forest
Festuca grassland
Agropyron grassland
Artemisia steppe
phytolith morphotypes
subalpine forest
Elymus grassland
other conifer
epidermal anticlinal
epidermal
polygonal
conical
wavy long
bilobates
trichomes
wavy
short
long
cells
rondels
hairs
seed
spiked
Asteraceae
-2 -1 0 +1 +2 +3
-1
0
+1
+2
-2
plates
blocky
Fig. 4. Detrended Correspondence Analysis scatterplot showing the first two axes of the simultaneous ordination results of phytolith surface
samples and morphotypes from the interior Pacific Northwest. Both axes are in units of standard deviations of DCA scores from the mean
sample score, which was assigned a zero value. Phytolith morphotypes are from Fig. 2 (#9–12 rondels and #4–7 long cells were merged into
single categories). Shape symbols represent 8 recognized vegetation types. Numbers correspond to 58 modern soil sample IDs from Table 4.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 87

types were excluded from the analysis. After initial
analysis of all 58 samples, a few alpine sites, the
only bother grasslandQand two bother coniferQforest
sites were dropped, leaving a total of 45 samples
for the direct gradient analysis. Despite the high
degree of variation in the data, several trends are
evident:
Rectangular plates, common in all grasses, were
more abundant at low elevations and most abundant
(about 13%) at approximately 500 m above sea
level. They did not show a strong relationship with
any factor.
Short wavy morphotype had higher values at sites
at intermediate elevations, with higher annual temper-
ature, lower precipitation, and intermediate number of
GDD (N2000). These were the dry grassland sites
dominated by Agropyron spicatum with high presence
of Poa sandbergii, where values close to 30% were
observed with MAT of 7 8C and MAP of 300 mm.
Overall, the presence of short wavy phytoliths
suggests presence of dry conditions with a very weak
negative linear trend for MAP (Fig. 6,R
2
= 3%).
Long wavy cells share increased with increasing
temperature and decreasing precipitation. The main
contributors of these phytoliths were grassland
species of Agropyron spicatum and Festuca idaho-
ensis. Over 30% long cells in a sample may indicate
moderately dry conditions (b600 mm MAP). Few
plots contained these, and thus no strong linear trend
was evident.
A sum of all long cells from grasses demonstrated a
weak negative trend with increasing elevation (Fig. 6,
R
2
= 9.2%), MAP (Fig. 6,R
2
= 23%)and AE / PE (Fig.
6,R
2
= 9.7%), and a weak positive trend with MAT
(Fig. 6,R
2
= 12%) and GDD N08C(Fig. 6,R
2
= 9.1%).
These are weak regression values, but stronger than
for individual long cell or for wavy morphotypes.
Long cells seem to be a good overall indicator of
grasslands.
The abundance of rondels (brondel sumQin Fig. 6)
decreases with increasing precipitation (R
2
= 5.5%)
and barely increases with increasing temperature
(R
2
= 2.9%). Samples with the highest proportion of
rondels (N30%) come from moderately moist Festuca
idahoensis-dominated grasslands with less than 670
mm MAP.
Bilobate phytoliths of Stipa-type had a bimodal
distribution for all environmental variables except
annual temperature range. These forms were found in
both cool and moist conditions (MAT b38C,
MAP N1500 mm) at subalpine sites, where Stipa
occidentalis is common, and in dry warm conditions
on the plains, where Stipa comata,Stipa thurberiana,
and Oryzopsis are found (MAT N10 8C, MAP b300
mm). Increase in GDD N08C had the strongest
positive linear trend (R
2
= 11.1%), but because of the
bimodal distribution, quadratic regression would be
more appropriate (R
2
= 13.6%).
Trichomes were most abundant in samples from
middle elevations (ca. 1300 m), moderately warm
temperatures (MAT 5 8C), and moderately moist
conditions (MAT 790 mm). These came largely from
Pinus ponderosa forests with Calamagrostis (N5% of
trichomes). Thus, trichomes may indicate presence of
forested, primarily Pinus-dominated, environments.
In contrast, silicified hairs were common in samples
from low elevations (b1000 m), with high MAT (N8
8C for samples with N10% hairs) and low MAP (b670
mm for samples with N10% hairs). Although the
relationship was again weak (e.g., R
2
= 9% for MAT
and 7.5% for MAP), such phytoliths were definitely
more common in dry, warm environments of grass-
lands and shrub steppe.
The sum of all grass phytoliths (bgrass sumQin Fig.
6) showed some of the strongest linear trends of all
morphotypes: a positive linear trend with increasing
MAT (R
2
= 11.7%) and GDD N08C(R
2
= 7.0%) and a
negative one with elevation (R
2
=8.9%), MAP
(R
2
= 23.7%), and AE / PE (R
2
= 3.3%).
Blocky phytoliths were most common in samples
from dry and warm sites at low elevations (shrub steppe
sites at b700 m, N10 8CMAT,b300 mm MAP;
R
2
= 4.9%). A few samples with abundant blocky forms
came also from subalpine forests at N1500 m elevation.
Fig. 5. Boxplots showing different phytolith morphotypes found in modern soils of the interior Pacific Northwest under eight vegetation types
arranged from the driest (steppe) to the wettest (subalpine forest). Vegetation codes: 1 — sagebrush steppe, 2 — Stipa grassland, 3 — Other
(Elymus) Grassland, 4 — Agropyron Grassland, 5 — Festuca Grassland, 6 — Pinus ponderosa Forest, 7 — other (Abies grandis and
Pseudotsuga menziesii) forest, 8 — Subalpine Forest. Percentage value of morphotype concentration in a sample is shown as dots. The left and
right sides of each box indicate the 25th and 75th percentiles respectively, the median is shown as the vertical line near the middle of the box.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9888

M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 89

Fig. 5 (continued).
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9890

They exhibited a positive trend with increased number
of GDD N08C(R
2
= 6.8%), and a negative trend with
elevation (R
2
= 7.6%), MAP (R
2
= 2.4%) and especially
AE / PE index (R
2
= 12.2%).
Many different taxa contribute epidermal polyg-
onal cells to the soil assemblages. No apparent
trendwasfoundintheclimatedataforthis
morphotype. It does exhibit a bimodal distribution
with elevation with mid-elevation pure grassland
sites having the lowest proportion (quadratic trend’s
R
2
= 17.1%).
In contrast, spiked phytoliths of Pinus ponderosa
had a very narrow distribution, since as a diag-
nostic form of P. ponderosa, they were restricted to
P. ponderosa forests. These phytoliths were found
only in samples from higher elevations (1000–1400
m), with 7–8 8C MAT and 600–1000 mm MAP.
Because of the narrowness of distribution, no linear
trend was apparent, except for AE / PE index
(R
2
= 19.8%).
bOther coniferQsuite of morphotypes was restricted
to the subalpine forests at even higher elevations
(1600–2000 m), with lower temperatures (3–4 8C
MAT, R
2
= 30%) and higher precipitation (900–2000
mm MAP, R
2
= 27.2%).
5. Discussion
5.1. Phytoliths in plants
Identifiable phytoliths were found in many domi-
nant plants in the interior Pacific Northwest (Table 1).
Phytoliths from two Artemisia shrubs, two Chryso-
thamnus shrubs, Bromus tectorum,Carex rossii,
Festuca viridula,andLupinus sericeus are described
here for the first time.
Characteristic phytolith shapes from four species of
Pinus, two species of Abies , two of Larix, three of
Picea, two of Tsuga, and Pseudotsuga were described
earlier from New York, Arizona and the European
Alps (Klein and Geis, 1978; Kearns, 2001; Carnelli et
al., 2004). My findings confirm conclusions about
diagnostic value of asterosclereids of Pseudotsuga
menziesii and bendodermal polyhedronsQ(i.e., blocky)
phytoliths of Picea (Norgren, 1973; Bozarth, 1993).
Fig. 5 (continued).
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 91

Fig. 6. Scatter diagrams (MINITAB, 1998) showing the abundance of selected phytolith morphotypes plotted against selected environmental
variables. Long cell sum includes morphotypes #4–7, rondel sum — #9–12, grass sum — #1–13. Environmental variables: MAT—mean annual
temperature in 8C, MAP—mean annual precipitation in mm, AE/ PE – actual to potential evapotranspiration index of Thornthwaite–Mather and
GDDN0 — growing-degree days with base N08C. LOWESS curves show the locally weighted regression of the relationship of phytoliths to
climatic variables.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9892

Klein and Geis (1978) and Carnelli et al. (2004)
reported that Larix sp. produces abundant epidermal
cell wall fragments, also found in my study.
Spiked phytoliths of Pinus ponderosa were pre-
viously described by Norgren (1973) and Kearns
(2001) andappearsimilarinformtothespiny
irregular bodies diagnostic of Pinus banksiana
(Bozarth, 1993). In the same study, a four-sided wavy
plate diagnostic of Picea glauca was reported, which I
found in Picea engelmannii, and elongated polyhe-
drons in Abies balsamea, corresponding to my blocky
form in Abies lasiocarpa. Some conifers in the study
area contained little (Juniperus occidentalis,Pinus
contorta) or no silicified material (Abies amabilis,
Abies grandis,Pinus albicaulis ,Thuja plicata). The
paucity of phytoliths in these trees can be attributed to
the overall low level of silicification in these species.
Given the high diagnostic value of the examined
conifer phytoliths, a thorough study of phytoliths for
all Holarctic conifers is desirable.
As reported in Geis (1973) and Bozarth (1992),
few deciduous trees in North America produce
abundant distinct phytoliths. Exceptions are Quer-
cus,Ulmus , and Acer, but these genera are not
prominent in my study area today. Deciduous tree
species produce polygonal and anticlinal (bjigsaw
puzzleQ) epidermal phytoliths, also produced by
some dicot shrubs and forbs. Polygonal epidermal
forms were reported from Quercus,Betula ,Corylus,
Populus, and Ulmus (Bozarth, 1992). Anticlinal
forms were found in Acer,Platanus,Populus and
Salix. In the study area, representatives of some of
these genera are restricted to floodplain forests or
upper treeline. Alnus sinuata, a shrub growing near
the upper treeline, was found to contain some
anticlinal phytoliths.
Unlike trees, shrubs have received little attention
with respect to their phytolith production. My study
demonstrated potential utility of Artemisia phyto-
liths. Blocky phytoliths of Artemisia of unknown
anatomical origin (perhaps from subepidermal tissue
or bark) appear to be similar to blocky forms of
Abies and Picea.Artemisia also produces epidermal
polygonal phytoliths (morphotype #18) common to
other Asteraceae.
In their classic paper, Twiss et al. (1969) proposed
abthree-groupQclassification model that divided all
grass phytoliths into Panicoids (i.e., bilobates and
crosses), Chloridoids (i.e., saddles), and Festucoids
(i.e., rondels and wavy forms). A thorough evaluation
of this system can be found in Mulholland (1989).
Because few Chloridoid or Panicoid species grow in
the study area, I chose to develop a more detailed
system of grass phytolith classification based primar-
ily on common Festucoid forms. Patterns in Festucoid
phytolith production in the PNW are similar to those
in Alberta (Blackman, 1971). In particular, Blackman
observed short wavy phytoliths in both Poa sand-
bergii (identified as Poa secunda) and Koeleria
cristata, long wavy phytoliths in Calamagrostis
rubescens, scutiform seed opal in Agropyron, and
diverse wavy and smooth phytoliths in Elymus. Her
illustrations of phytoliths from Stipa comata and
three other species of Stipa appear similar to my
pyramidal and Stipa-type morphotypes (Blackman,
1971).
An earlier attempt at classifying grass phytoliths
(Norgren, 1973) provided data and illustrations of
phytoliths in seven species of grasses and one sedge
from Oregon. He distinguishes three classes: brodsQ
(corresponding to plates (#1) and long cells (#4–7) in
this study), bhookbasesQ(apparently corresponding to
trichomes (#14) and rondels (#9–12) in this study),
and hairs (#15). The findings of Norgren (1973)
closely match my results, although lack of clear
definitions for some of his morphotypes makes direct
comparisons difficult. For example, he found Agro-
pyron spicatum to contain large proportion of bwavy
rodsQ(around 27%), similar to my bdeeply indented
rods and angular rodsQ(15%). Festuca idahoensis,on
the other hand, contained mostly brough rods b(60%),
apparently corresponding to my bindented long cellQ
(24% in my study), and hairs (17% in Norgren (1973),
4% in my study). No distinction was made by
Norgren (1973) between trichomes and rondels, both
were treated as bhookbasesQ(i.e., silicified epidermal
appendages), which confuses interpretation. For
example, Stipa comata had about 40% bhookbasesQ
(bspheroidal to peanut-shapedQ,Norgren, 1973, p. 57),
apparently corresponding to pyramidal rondels and
Stipa-type bilobates (this study), which anatomically
are not epidermal appendages at all, but epidermal
short cells. Surprisingly, Norgren (1973) does not
describe any bhookbasesQin F. idahoensis, which is
hard to explain given the widespread occurrence of
rondels in this species, as well as in other Festuca spp.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 93

worldwide (Smithson, 1958; Kiseleva, 1982; Blinni-
kov, 1994).
Norgren (1973) mentions high presence of tri-
chomes (blarge hookbasesQ)inCalamagrostis rubes-
cens, which my study also confirmed. I also found a
high proportion of bsmooth rodsQ(i.e., rectangular
long cells) in Elymus cinereus.Theonlysedge
mentioned in Norgren (1973) was Carex geyerii,
which contained both brodsQ(i.e., long cells), and
bsmall hookbasesQ(i.e., diagnostic conical forms of
Carex).
Brown (1984) examined 112 species of grasses
common to central North America. Direct compar-
isons are difficult to make, however, because his
system is very detailed. Overall, Brown’s results
support my findings with respect to phytolith forms
found in Agropyron spicatum and Agropyron dasys-
tachyum,Calamagrostis rubescens,Festuca idaho-
ensis,Koeleria cristata ,Poa sandbergii,Sitanion
hystrix, and Stipa comata .
Mulholland (1989) identified a large number of
bpolylobate sinuateQforms in Koeleria cristata and
Elymus canadensis from North Dakota (cf. short
wavy plates in this study), entire rondels (apparently,
pyramidal) in Stipa comata, and abundant Stipa-type
bilobates in other Stipa species (cf. Stipa-type
bilobates in this study).
Kiseleva (1982) described phytoliths from arid
Mongolian grasslands, including some from the genera
foundinmystudyarea.Someofhermatching
observations include high abundance (about 80%) of
rondels in seven closely related species of Festuca , the
presence of Stipa-type phytoliths in six Stipa species
(ca. 20%), presence of pyramidal rondels in Elymus
and short wavy forms in Koeleria cristata, and a high
percentage of long wavy forms in Calamagrostis
macrolepis. Thus, different species of the same genera
of Festucoids in Mongolia contain phytolith morpho-
types similar in appearance and proportion to the North
American species. This conclusion is further corrobo-
rated by Blinnikov (1994), who reported many of the
same phytolith morphotypes in the same genera of
Festucoid grasses from the Caucasus (e.g., Calama-
grostis,Festuca,Poa).
In a new study from Europe, Carnelli et al.
(2004) analyzed 21 species from the Swiss Alps.
Their results confirm many of my findings both with
respect to grass and non-grass forms. While their
classification is considerably more detailed (e.g.,
they distinguish 18 types of trichomes, while I lump
them all together), ample illustrations make compar-
isons easy. Their analysis of another Calamagrostis
species (Calamagrostis villosa) corroborate my
findings that this genus tends to have high incidence
of long wavy cells and trichomes (Blinnikov, 1994;
this study). Three Festuca species tend to have high
incidence of rondels (their btrapezoidQcategory).
Larix in their study contained spiky cells (cf. my
morphotype #20), etc.
5.2. Phytoliths in modern soils
Phytolith assemblages in soils can differentiate
common vegetation types of the interior Pacific
Northwest. Overall, the most distinct assemblages
come from Pinus ponderosa forests and Stipa grass-
lands. Both could be identified on the basis of a single
diagnostic morphotype, spiked of P. ponderosa (#17)
and Stipa-type bilobate (#13) respectively. Stipa-type
bilobates (Fredlund and Tieszen, 1994) and spiked
phytoliths (Kearns, 2001) can persist for centuries in
the modern soil record. Another diagnostic morpho-
type, the large silicified asterosclereids of Pseudotsuga
menziesii, helped identify the presence of P. menziesii
in the past (Norgren, 1973), but their extremely large
size makes them rare in the silt fraction. Although
Artemisia shrub lacks phytoliths diagnostic of the
genus, Artemisia shrub steppe can be reliably identified
on the basis of the high presence of blocky forms
apparently from Artemisia and polygonal epidermal
forms of other shrubs and forbs.
Grasslands can be distinguished from either forests
or shrub steppe on the basis of high proportion of
grass phytoliths (N90%), particularly long cells and
wavy forms in Agropyron-dominated grassland and
rondels in Festuca-dominated grassland. Although
both arid Agropyron-dominated and mesic Festuca-
dominated grasslands have similar assemblages, they
can be distinguished based on different types of
rondels and long cells: high presence of short wavy
forms indicate presence of Poa sandbergii or Koe-
leria cristata.
Few modern reference studies of phytoliths in soils
exist anywhere in North America. Norgren (1973)
analyzed soil samples from ten different locations in
Oregon, including three samples from the central
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9894

Coast Range, two from the Cascades, one from
eastern Oregon, three from the Columbia Plateau,
and three from the Wallowa Mountains. Despite
confusing terminology, the findings match my data
reasonably well. For example, a Tolo series soil from
Pinus ponderosa–Pseudotsuga menziesii–Larix occi-
dentalis forest in the Wallowa Mountains, which
contained 10% Calamagrostis rubescens and Carex
geyerii understory, featured 35% trichomes (referred
to as bhookbasesQ)ofC. rubescens. In my study, 20%
trichomes were observed in a sample from P.
ponderosa–C. rubescens forest in the northern Blue
Mountains in Washington State (sample 45 on Fig. 3).
According to Norgren (1973),Agropyron-dominated
grassland assemblages, such as Condon silt loam
sample from northern Oregon near the Columbia
River, contained about 80% of brodsQ(i.e., long cells
and wavy and rectangular plates), and 12%
bhookbasesQ(i.e., rondels and hair bases), similar to
my study.
A more recent example of a modern analog
phytolith study comes from the Great Plains (Fredlund
and Tieszen, 1994). Although only short-cell forms
were considered, some interesting comparisons can be
made. Most of the 50 assemblages studied in that
work contained a high proportion of saddles, coming
from the short-grass prairie species. None of my
assemblages contained saddles, not surprisingly, given
the fact that Chloridoid grasses were not found in the
study area. Two samples from Stavely, Alberta, from
Agropyron –Stipa–Danthonia community (Fredlund
and Tieszen, 1994) are similar to some of the driest
assemblages (Stipa-dominated grasslands with Agro-
pyron dasystachyum) from Lindsay Prairie in the
Columbia Basin (samples 12–14 on Fig. 3). The
Stavely assemblages were dominated by three kinds
of rondels (60%), Stipa-type bilobates (10%), and
crenate (wavy/lobed plates in my classification)
phytoliths (10%). My samples were likewise domi-
nated by rondels (40%), Stipa-type (15%), and wavy
forms (40%), if all the non-grass, long cell, and
trichomes/hair phytoliths are excluded. It is also
noteworthy that Fredlund and Tieszen (1994) bkeeledQ
Qmorphotype appear to be the same form as my
belongated rondelQ, and is likewise found in a member
of Agropyron genus (Agropyron smithii). The Stavely
assemblages contained about 20% of this morphotype,
while Agropyron-dominated grasslands in my study
contained only 9% of this morphotype (median
value), or about 14% if non-grass and long cell forms
are excluded. Although it is true that long cell
taphonomy and production is less known than that
of the short-cell forms (Fredlund and Tieszen, 1994),
the results of my research suggest that such forms
should be considered in the phytolith analysis because
they yield additional interesting data.
Some further comparisons can be made with a
modern analog study of phytoliths assemblages from
subalpine and alpine communities in the northwestern
Caucasus (Volkova et al., 1995). Despite different
species, genera of grasses and conifers were the same,
and some similarities can be pointed out. Phytolith
assemblage from Pinus hamata–Calamagrostis arun-
dinacea forest from Teberda Nature Reserve, Russia,
is similar to the Pinus ponderosa–Calamagrostis
rubescens assemblages from the Blue Mountains,
Washington. Both assemblages contain a high propor-
tion of diagnostic Pinus phytoliths, trichomes, and
long wavy phytoliths of Calamagrostis. The more
prominent role of Pinus phytoliths in the Blue
Mountains may be explained by the higher phytolith
production in P. ponderosa. As more modern phytolith
studies are done elsewhere, there will be more
opportunities for further interregional comparisons.
A reliable internationally recognized nomenclature of
morphotypes is being developed to facilitate such
comparisons in the future (Madella et al., 2003).
5.3. Phytoliths and climate
Poor understanding of phytolith production and
preservation patterns hampers application of the
phytolith analysis in direct paleoclimatic reconstruc-
tions (Fredlund and Tieszen, 1997b). Phytoliths are
commonly thought to reflect subtle shifts in vegetation
composition at the local scale, because they are
primarily deposited in situ (Piperno, 1988). However,
phytoliths are displaced from the original deposition
site both vertically and horizontally as a result of fire,
flood, grazing, burrowing, or another disturbance.
Such mixing may blur the vegetation signal offered
by phytolith assemblages in modern soils. Indeed,
Fredlund and Tieszen (1994) found that phytoliths
from a given geographic location appeared to reflect
vegetation of a larger area than previously thought.
They concluded that the phytolith assemblages were
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 95

likely to reflect regional climate. For example, at the
regional scale, grassland composition is closely asso-
ciated with temperature and moisture gradients. Cali-
bration of grass phytolith assemblages in climatic terms
opened a way to the direct paleoclimatic reconstruc-
tions from phytoliths by developing response surfaces
and transfer functions (Fredlund and Tieszen, 1997b).
The phytolith data could thus supplement paleoclimatic
reconstructions provided by pollen (Bartlein et al.,
1986; Williams et al., 2001).
More extensive and even sampling of different
vegetation and climate zones would be required
before quantitative paleoclimatic reconstructions
could be made with phytoliths. My sampling strategy
was to maximize the number of different native
vegetation types visited, which necessitated sampling
dissimilar communities near ecotones (e.g., a grass-
land site sampled in proximity to a forested site). In
macroclimatic terms, such sites would have an
identical climate, which would blur the phytolith-
derived signal. More even sampling of sites along a
regular grid and over a larger geographical area, as
was recently done with the regional pollen (Minckley
and Whitlock, 2000), would better illustrate current
phytolith–climate relationships.
6. Conclusions
The objective of this study was to provide a
basis for the interpretation of fossil phytolith
assemblages recovered from the late Pleistocene
loessal paleosols in the Columbia Basin Province of
the Pacific Northwest. Phytoliths from 38 plant
species and 58 samples of modern soils were
examined. Using twenty phytolith morphotypes
from modern soils (13 of which were from grasses)
it was possible to distinguish eight vegetation types,
including Artemisia-dominated shrub steppe, Stipa-
dominated lowland and subalpine grasslands, Agro-
pyron-dominated dry grassland, Festuca-dominated
mesic grassland, Elymus cinereus grassland, Pinus
ponderosa-dominated dry forest, Abies grandis–
Pseudotsuga menziesii mesic forest and Abies
lasiocarpa–Picea engelmannii subalpine forest.
Shrub steppe, Stipa-dominated grasslands, P. pon-
derosa forests, and subalpine forests produced the
most distinctive phytolith assemblages. Use of three
morphotypes within the group of long cells and four
different rondels enabled further differentiation of
grasslands.
Grass phytoliths are the most diagnostic at the
subfamily level, but can also distinguish certain
genera of Festucoid grasses. For example, Agro-
pyron,Calamagrostis,Festuca ,Koeleria,Poa and
Stipa may be distinguished based on their phytolith
record. Non-grass phytoliths appear to be most
distinctive at the family level, but some genera,
and even species, could be identified, e.g., Pseudot-
suga menziesii and Pinus ponderosa in my study
area. Some morphotypes are redundant, e.g., rectan-
gular plates are found in almost all grasses,
polygonal epidermal cells are common in many
conifers, shrubs and forbs, and silicified tracheids are
common in most conifers.
Study of direct phytolith–environmental relation-
ships proves that certain morphotypes occur prefer-
entially in certain climates and at certain elevations.
For example, long cells as a group and short wavy
morphotype tend to occur in higher proportion in drier
and warmer habitats at lower elevations, while spiked
and other conifer morphotypes occur under moist and
cool conditions at higher elevations. Rondels tend to
be more abundant in samples from moderately dry to
moderately moist habitats. Bilobates of Stipa-type
have a distinct bimodal distribution, occurring in
either the dry and warm climate of the lowlands, or in
the cool climate of the subalpine zone. Blocky forms
are mostly restricted to dry and warm habitats in
lowland areas, but are also found in subalpine forest
samples. While the data show only weak climate–
phytolith relationships, selection of a wider range of
sites across longer environmental gradients could
further improve our understanding of the phytolith–
climate relationships.
This study did not address the issues of variable
production of opal by individual species and variability
in preservation of various forms in soils. In general,
paleoassemblages may be matched against their
modern analogs based on the whole assemblage com-
position or indicator forms. Additional research of
production and preservation is required, because some
phytoliths may prove to be systematically under-
represented in the paleoassemblages based on either
their low production today or low preservation rate in
soils.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9896

Acknowledgements
I. Blinnikova, A. Gubin and S. Sumstine pro-
vided help in the field. I. Blinnikova also greatly
assisted with lab processing. P. Bartlein provided the
modern climate dataset and helped with climate–
phytolith interpretations. A. Busacca, P. Bartlein, P.
MacDowell, C. Whitlock, and G. Retallack provided
helpful comments on an earlier draft of this paper.
The paper presents some of the results of a Ph.D.
dissertation research supported by the Mazamas and
Sigma Xi research grants and the Graduate Doctoral
Research Fellowship of the University of Oregon.
References
Baker, U.R., Bjornstad, B.N., Busacca, A.J., Fecht, K.R., Kiver,
E.P., Moody, U.L., Rigby, J.G., Stradling, O.F., Tallman, A.M.,
1991. Quaternary Geology of the Columbia Plateau. In:
Morrison, R.B. (Ed.), Quaternary Non-Glacial Geology: Con-
terminous U.S. The Geology of North America. Geol. Soc. Am.,
Boulder, CO, vol. K-2.
Ball, T.B., Gardner, J.S., Anderson, N., 1999. Identifying inflor-
escence phytoliths from selected species of wheat (Triticum
monococcum,T. diccoccum,T. diccocoides, and T. aestivum)
and barley (Hordeum vulgare and H. spontaneum) (Gramineae).
Am. J. Bot. 86, 1615 – 1623.
Barkworth, M.E., 1981. Foliar epidermis and taxonomy of North
American Stipeae (Gramineae). Syst. Bot. 6, 136 – 152.
Bartlein, P.J., Prentice, I.C., Webb III, T., 1986. Climatic response
surfaces from pollen data for some eastern North American taxa.
J. Biogeogr. 13, 35 – 57.
Bartlein, P.J., Lipsitz, B.B., Thompson, R.S., 1994. Modern climate
data for paleoenvironmental interpretations. AMQUA Abstracts
13th Biennial Meeting, p. 197.
Beavers, A.H., Stephen, I., 1958. Some features of the distribution
of plant opal in Illinois soils. Soil Sci. 86, 1 – 5.
Blackman, E., 1971. Opaline silica bodies in the range grasses of
southern Alberta. Can. J. Bot. 49, 769 – 781.
Blinnikov, M., 1994. Phytolith analysis and the holocene dynamics
of alpine vegetation. In: Onipchenko, V., Blinnikov, M. (Eds.),
Experimental Investigation of Alpine Plant Communities in the
Northwestern Caucasus. Verfffentlichungen des Geobotani-
schen Institutes der ETH, Stiftung Rqbel, Zqrich. H., vol.
115, pp. 23– 40.
Blinnikov, M., Busacca, A., Whitlock, C., 2001. A new 100,000-year
record from the Columbia basin, Washington, U.S.A. In:
Meunier, J.D., Colin, F. (Eds.), Phytoliths—Applications in
Earth Science and Human History. A.A. Balkema, Rotterdam,
Netherlands.
Blinnikov, M., Busacca, A., Whitlock, C., 2002. Reconstruction of
the Late Pleistocene Columbia basin grassland, Washington,
USA, based on phytolith records in Loess. Palaeogeogr.
Palaeoclimatol. Palaeoecol. 177, 77 – 101.
Bowdery, D., 1998. Phytolith analysis applied to Pleistocene–
Holocene archaeological sites in the Australian arid zone. Br.
Archaeol. Rep. Int. Ser., 695.
Bozarth, S.R., 1992. Classification of opal phytoliths formed in
selected dicotyledons native to the Great Plains. In: Rapp Jr., G.,
Mulholland, S.C. (Eds.), Phytolith Systematics: Emerging
Issues. Plenum Press, New York, NY, pp. 193 – 214.
Bozarth, S.R., 1993. Biosilicate assemblages of boreal forests and
aspen parklands. In: Pearsall, D.M., Piperno, D.R. (Eds.),
Current Research in Phytolith Analysis: Applications in
Archaeology and Paleoecology. MASCA Research Papers in
Science and Archaeology, University of Pennsylvania, Phila-
delphia, PA, pp. 95 – 105.
Brown, D.A., 1984. Prospects and limits of a phytolith key for
grasses in the central United States. J. Archaeol. Sci. 11,
345 – 368.
Bryson, R.A., Hare, F.K., 1974. The climates of North America. In:
Bryson, R.A., Hare, F.K. (Eds.), Climates of North America.
Elsevier, Amsterdam, pp. 1– 47.
Carnelli, A.L., Madella, M., Theurillats, J.-P., 2001. Biogenic silica
production in selected alpine plant species and plant commun-
ities. Ann. Bot. 87, 425 – 434.
Carnelli, A.L., Theurillat, J.-P., Madella, M., 2004. Phytolith types
and type-frequencies in subalpine–alpine plant species of the
European Alps. Rev. Palaeobot. Palynol. 129, 39 – 65.
Cleveland, W.S., 1993. Visualizing Data. Hobart Press, Summit,
NJ.
Daubenmire, R.F., 1968. Forest vegetation of Eastern Washington
and Northern Idaho. Technical Bulletin of the Washington
Agricultural Experimental Station, vol. 60. Pullman, WA.
Daubenmire, R.F., 1970. Steppe vegetation of Washington. Tech-
nical Bulletin of the Washington Agricultural Experimental
Station, vol. 62. Pullman, WA.
Franklin, J.F., Dyrness, C.T., 1988. Natural Vegetation of Oregon
and Washington. Oregon State University Press, Corvallis, OR.
Fredlund, G.G., Tieszen, L.L., 1994. Modern phytolith assemb-
lages from the North American Great Plains. J. Biogeogr. 21,
321 – 335.
Fredlund, G.G., Tieszen, L.L., 1997a. Phytolith and carbon
evidence for late quaternary vegetation and climate change
in the Southern Black Hills, South Dakota. Quat. Res. 47,
206 – 217.
Fredlund, G.G., Tieszen, L.L., 1997b. Calibrating grass phytolith
assemblages in climatic terms: application to late pleistocene
assemblages from Kansas and Nebraska. Palaeogeogr. Palae-
oclimatol. Palaeoecol. 136, 199 – 211.
Geis, J.W., 1973. Biogenic silica in selected species of deciduous
angiosperms. Soil Sci. 116, 113– 130.
Hitchcock, C.L., Cronquist, A., 1994. Flora of the Pacific North-
west, 9th printing. University of WA Press, Seattle and London.
Kaplan, L., Smith, M.B., Sneddon, L.A., 1992. Cereal grain
phytoliths of southwest Asia and Europe. In: Rapp Jr., G.,
Mulholland, S.C. (Eds.), Phytolith Systematics: Emerging
Issues, Advances in Archaeological and Museum Science,
vol. 1. Plenum Press, New York, pp. 149 – 174.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–98 97

Kearns, B., 2001. Diagnostic phytoliths for a ponderosa pine–
bunchgrass community near Flagstaff, Arizona. Southwest. Nat.
46, 282 – 294.
Kearns, B., 2001. Diagnostic phytoliths for a ponderosa pine–
bunchgrass community near Flagstaff, Arizona. Southwest. Nat.
46, 282 – 294.
Kiseleva, N.K., 1982. Izuchenie fitolitov v pochvah dlya vyjasne-
nija istorii rastitel’nosti stepej vostochnoj Mongolii. Izv. AN
SSSR, Ser. Geogr. 2, 95 – 106 (in Russian).
Klein, R.L., Geis, J.W., 1978. Biogenetic opal in the Pinaceae. Soil
Sci. 126, 145 – 156.
Kurman, M.H., 1985. An opal phytolith and palynomorph study of
extant and fossil soils in Kansas (USA). Palaeogeogr. Palae-
oclimatol. Palaeoecol. 49, 217 – 235.
Lanning, F.C., Eleuterius, L.N., 1987. Silica and ash in native plants
of the central and southeastern regions of the United States.
Ann. Bot. 63, 395 – 410.
Lu, H., Liu, K., 2003. Phytoliths of common grasses in the coastal
environments of southeastern USA. Estuar. Coast. Shelf Sci. 58,
587 – 600.
Madella, M., Alexandre, A., Ball, T., 2003. International code for
phytolith nomenclature 1.0. The Phytolitharien 15, 7 – 16.
Minckley, T., Whitlock, C., 2000. Spatial variation of modern pollen
in Oregon and southern Washington, USA. Rev. Palaeobot.
Palynol. 112, 97– 123.
MINITAB, 1998. MINITAB Reference Manual Version 12.1, State
College, PA.
Mulholland, S.C., 1989. Phytolith shape frequencies in North
Dakota grasses: a comparison to general patterns. J. Archaeol.
Sci. 16, 489 – 511.
Mulholland, S.C., 1993. A test of phytolith analysis at Big
Hidatsa, North Dakota. In: Pearsall, D.M., Piperno, D.R.
(Eds.), Current Research in Phytolith Analysis: Applications
in Archaeology and Paleoecology. MASCA Research Papers
in Science and Archaeology. University of Pennsylvania,
Philadelphia, PA, pp. 131 – 145.
Norgren, J.A., 1973. Distribution, Form and Significance of Plant
Opal in Oregon Soils. Ph.D. Dissertation, Oregon State
University, Corvallis, OR.
Ollendorf, A.L., 1992. Toward a classification scheme of sedge
(Cyperaceae) phytoliths. In: Rapp Jr., G., Mulholland, S.C.
(Eds.), Phytolith Systematics: Emerging Issues. Plenum Press,
New York, NY, pp. 91 – 106.
PC-ORD, 1997. PC-ORD for Windows. Multivariate Analysis of
Ecological Data Version 3.0. MJM Software, Gleneden Beach,
OR.
Pearsall, D.M., 2000. Paleoethnobotany: a Handbook of Procedures,
2nd edition. Academic Press, San Diego, CA.
Piperno, D.R., 1988. Phytolith Analysis: an Archeological
and Geological Perspective. Academic Press, San Diego,
CA.
Piperno, D.R., Persall, D.M., 1993. The nature and status of
phytolith analysis. In: Pearsall, D.M., Piperno, D.R. (Eds.),
Current Research in Phytolith Analysis: Applications in
Archaeology and Paleoecology. MASCA Research Papers in
Science and Archaeology, University of Pennsylvania, Phila-
delphia, PA, pp. 9 – 18.
Piperno, D.R., Persall, D.M., 1998. The Silica Bodies of Tropical
American Grasses. Smithson. Contrib. Bot. Washington, DC.
Prentice, I.C., Cramer, W., Harrison, S.P., Leemans, R., Monserud,
R.A., Solomon, A.M., 1992. A global biome model based on
plant physiology and dominance, soil properties and climate.
J. Biogeogr. 19, 117 – 134.
Rovner, I., 1971. Potential of opal phytolith for use in paleoeco-
logical reconstruction. Quat. Res. 1, 343 – 359.
Smithson, F., 1958. Grass opal in British soils. J. Soil Sci. 9,
148 – 154.
Soil Survey Laboratory Methods Manual, 1996. Soil Survey
Investigations Report No. 42, version 3.0, NRCS, USDA.
Ter Braak, C.J.F., Prentice, I.C., 1988. A theory of gradient analysis.
Adv. Ecol. Res. 18, 271 – 317.
Thompson, R.S., Anderson, K.H., Bartlein, P.J., 2000. Atlas of
relations between climatic parameters and distributions of
important trees and shrubs in North America. U. S. Geol. Surv.
Prof. Pap. 1650.
Twiss, P.C., Suess, C.E., Smith, R.M., 1969. Morphological
classification of grass phytoliths. Soil Sci. Soc. Am. Proc. 33,
109 – 115.
Verma, S.D., Rust, R.H., 1969. Observation on opal phytoliths in a
soil biosequence in southeastern Minnesota. Soil Sci. Soc. Am.
Proc. 33, 749 – 751.
Volkova, E.V., Onipchenko, V.G., Blinnikov, M.S., 1995. Subrecent
phytolith assemblages of alpine communities in the Teberda
nature reserve, the north-western Caucasus, Russia. Oecol.
Mont. 4, 1 – 8.
Walter, K.S., 1975. A preliminary study of the achene epidermis of
certain Carex (Cyperaceae) using scanning electron microscopy.
Mich. Bot. 14, 67 – 72.
Williams, J.W., Shuman, B.N., Webb III, T., 2001. Dissimilarity
analyses of late-quaternary vegetation and climate in eastern
North America. Ecology 82, 3346 – 3362.
Willmott, C.J., Rowe, C.M., Minz, Y., 1985. Climatology of the
terrestrial seasonal water cycle. J. Climatol. 5, 589 – 606.
Witty, J.E., Knox, E.G., 1964. Grass opal in some chestnut and
forest soils in north central Oregon. Soil Sci. Soc. Am. Proc. 28,
685 – 687.
M.S. Blinnikov / Review of Palaeobotany and Palynology 135 (2005) 71–9898