Content uploaded by Marina Potapova
Author content
All content in this area was uploaded by Marina Potapova
Content may be subject to copyright.
This article was downloaded by: [Consortium Luxembourg]
On: 28 February 2012, At: 10:42
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House,
37-41 Mortimer Street, London W1T 3JH, UK
Diatom Research
Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/tdia20
Ontogenetic and interspecific valve shape variation
in the Pinnatae group of the genus Surirella and the
description of S. lacrimula sp. nov.
Jonathan David English a & Marina G. Potapova a
a Diatom Herbarium, Academy of Natural Sciences of Philadelphia, Philadelphia, PA, USA
Available online: 16 Dec 2011
To cite this article: Jonathan David English & Marina G. Potapova (2012): Ontogenetic and interspecific valve shape variation
in the Pinnatae group of the genus Surirella and the description of S. lacrimula sp. nov., Diatom Research, 27:1, 9-27
To link to this article: http://dx.doi.org/10.1080/0269249X.2011.642950
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any form to
anyone is expressly forbidden.
The publisher does not give any warranty express or implied or make any representation that the contents
will be complete or accurate or up to date. The accuracy of any instructions, formulae, and drug doses should
be independently verified with primary sources. The publisher shall not be liable for any loss, actions, claims,
proceedings, demand, or costs or damages whatsoever or howsoever caused arising directly or indirectly in
connection with or arising out of the use of this material.
Diatom Research
Vol. 27, No. 1, March 2012, 9–27
Ontogenetic and interspecific valve shape variation in the Pinnatae group of the genus Surirella
and the description of S. lacrimula sp. nov.
JONATHAN DAVID ENGLISH∗& MARINA G. POTAPOVA
Diatom Herbarium, Academy of Natural Sciences of Philadelphia, Philadelphia, PA, USA
The morphology and ontogenetic allometric trends of several taxa in the Pinnatae group of Surirella were studied using traditional
morphometrics, landmark-based shape analysis, and light and electron microscopy. We investigated separately two groups of species:
the first consisted of S. brebissonii Krammer & Lange-Bertalot, S. brebissonii var. kuetzingii Krammer & Lange-Bertalot and S. ovalis
Brébisson, and the second included specimens originally identified as S. minuta Brébisson and S. pinnata W. Smith. Morphological
variability within both groups was mainly limited to differences in valve shape and size. Landmark-based shape analysis revealed several
shape groups within both species complexes, although these shape groups were not separated by clear gaps. Additional groups of specimens
were separated on the basis of different ontogenetic allometric trajectories. In both species complexes, valve shapes converged at later
stages of the vegetative life cycle. Within the ‘S. brebissonii–S. ovalis’ species complex, one shape group corresponded to S. ovalis and
another to S. brebissonii +S. brebissonii var. kutzingii. The latter two varieties had similar average shape, but differed in their ontogenetic
trajectories. In samples from the USA, only representatives of S. ovalis and S. brebissonii var. kutzingii were found. In the ‘S. minuta–
S. pinnata’ species complex, three distinct shape groups were found. One of these is described here as a new species, S. lacrimula English
and two others corresponded to S. minuta and S. pinnata. We suggest, therefore, maintaining these two previously synonymized species
as separate taxa.
Keywords: allometry, morphometrics, new species, Surirella, shape analysis, USA
Introduction
While working on the online diatom identification guide
‘Diatoms of the United States’ (http://westerndiatoms.
colorado.edu/), we re-evaluated records of the frequently
reported freshwater Surirella species at the Academy of
Natural Sciences of Philadelphia (ANSP) Diatom Herbar-
ium. We found that the identification of species in
the Pinnatae section of Surirella (Peragallo & Peragallo
1897–1908), characterized by a low keel and absence of
alar canals, was particularly inconsistent. One obvious
reason for such inconsistency was the low number of dis-
crete morphological characters useful for distinguishing
species within species complexes in this group. Diatom
species within so-called ‘species complexes’, or groups of
very similar species, rarely differ from each other by dis-
crete morphological characters. New diatom species within
species complexes are often described upon finding gaps in
distributions of their continuous characters, such as valve
dimensions or density of structural elements: striae, fibulae,
costae, etc. (e.g., Mann et al. 2004, Edlund & Soninkhishig
2009, Falasco et al. 2009). Such continuous characters are
also the most commonly used for species identifications
within the Pinnatae section of Surirella, but some species
∗Corresponding author. Email: english@ansp.org
(Received 16 June 2011; accepted 17 November 2011)
have overlapping ranges of variation for these characters
(e.g., Krammer & Lange-Bertalot 1987).
The second reason for inconsistent identifications was
a tendency for cell shapes to converge near the end of the
cell cycle. For example, Krammer & Lange-Bertalot (1987)
stated that only the large valves of Surirella brebissonii
Krammer & Lange-Bertalot could be distinguished in light
microscopy (LM) from S. ovalis Brébisson, while smaller
valves of two species were indistinguishable. Another
taxon, S. brebissonii var. kuetzingii Krammer & Lange-
Bertalot was based solely on the difference in size range
between it and the nominate variety (Krammer & Lange-
Bertalot 1987). Separating these taxa is thus difficult in
samples in which there are few valves.
Another species complex that caused identification
difficulties, judging from ANSP records, was S. pinnata
W. Smith +S. minuta Brébisson. Krammer & Lange-
Bertalot (1987) synonymized these two species, as reflected
in the ANSP records: most of the earlier records were of
S. pinnata and the later ones of S. minuta. We also noticed
that most USA populations originally identified as S. minuta
differed slightly in their cell proportions from the speci-
mens from either of the two type populations and were on
ISSN 0269-249X print/ISSN 2159-8347 online
© 2012 The International Society for Diatom Research
http://dx.doi.org/10.1080/0269249X.2011.642950
http://www.tandfonline.com
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
10 English & Potapova
average more rounded. These rounded valves resembled
the smaller, rounded valves of S. minuta in shape, but did
not have the same size range as the type population. We
hypothesized that that a quantitative shape analysis may
yield useful characters for separating species within these
two species complexes.
Shape variation within groups of closely related species
may be attributed to several factors: interspecific and
intraspecific genetic differentiation, shape change related
to ontogeny (ontogenetic allometry) and phenotypic plas-
ticity. The usual goal of shape analysis in diatom studies
is to reveal shape groups that may correspond to separate
taxa. Therefore, it is important to take into account other
sources of variation, especially changes that occur during
the vegetative stage of the life cycle. In pennate diatoms,
the reduction in valve length over the life cycle is usually
accompanied by changes in valve proportions and shape
(Geitler 1932), while the numbers of structural elements
(processes, striae) may change with valve length/diameter
in both pennate and centric diatoms. The size-dependent
variation in quantitative characters has long been recog-
nized by diatomists, who have tried to take it into account
by, for example, creating such morphological characters as
ratios of two variables (reviewed in Theriot 1988) or slopes
of regression of one variable against another (R.K. Edgar
et al. 2004, S.M. Edgar & Theriot 2004). This approach is
especially useful when comparing taxa or populations with
similar size ranges and when a linear relationship between
two variables is established. Ontogenetic allometric trends
visualized as plots of various quantitative variables against
cell length (or diameter in centric diatoms) have also been
used to demonstrate morphological differences in closely
related species (e.g., Theriot & Stoermer 1984, R.K. Edgar
et al. 2004, Potapova & Ponader 2004).
The first goal of this study was to clarify taxa bound-
aries by exploring variation in valve shape in two species
complexes: (1) ‘S. ovalis–S. brebissonii’ complex, also
containing S. brebissonii var. kuetzingii; and (2) ‘S. min-
uta–S. pinnata’ species complex. The second goal was to
update morphological descriptions of species from these
complexes found in North America. Instead of eliminating
shape variability attributed to ontogenetic allometry, our
approach was to use ontogenetic (growth) trajectories as
additional morphological characters. A quantitative anal-
ysis of ontogenetic trajectories of valve shape has rarely
been carried out in the past (but see Stoermer & Ladewski
1982, Mou & Stoermer 1992) because of the difficulties
with extracting valve outlines in the outline-based shape
analysis. The development of landmark-based shape anal-
ysis provides a less laborious alternative to shape analysis
and has already been used in diatom studies (Potapova &
Hamilton 2007, Falasco et al. 2009, Veselá et al. 2009). In
landmark-based methods of shape analysis, landmarks are
placed on structures that are considered homologous across
specimens and the landmark coordinates are recorded. This
allows the use of superimposition techniques that remove
non-shape variation, such as size and orientation of speci-
mens, from the shape variation (Rohlf & Slice 1990). The
extracted shape descriptors can then be used as separate
morphological characters or combined using multivariate
analysis (Bookstein 1996). The sliding semi-landmarks
method is an extension of a standard landmark method
suitable for analyzing outlines (Bookstein 1997).
Materials and methods
Materials examined
The diatom slides and samples investigated in this study
are listed in Table 1. All materials are housed at ANSP.
Most of the materials were collected in the USA, but five
slides represented European collections. The first of these
was WmS64, the isotype slide of S. pinnata from Lewes,
England. The second was HLSEx528, the slide from the
H.L. Smith exsiccati set originally from the Brébisson col-
lection and identified by Brébisson as S. minuta. Although
there is no indication of locality on this slide, it is possible
that this slide represents a syntype, or even an isotype of
this species originally described by Brébisson from Falaise,
France. The other European materials were slide GC11960,
from Longpré, France, originally from P. Petit’s collec-
tion, identified as S. pinnata, slide GC11277 from Strehlen,
Germany, identified as S. pinnata and S. minuta and slide
Feb3448 from the Glasgow Botanical Gardens, Scotland,
identified as S. minuta. In addition, for the shape anal-
ysis, we scanned and used several published images of
specimens from the type populations of S. ovalis (Kram-
mer & Lange-Bertalot 1987: figs 16–20), S. brebissonii
(Krammer & Lange-Bertalot 1987: figs 21–26), S. breis-
sonii var. kutzingii (Krammer & Lange-Bertalot 1987: figs
55–56, 58–68) and S. minuta (Krammer & Lange-Bertalot
1988: Tafel 135, figs 1, 4–5, 10).
Microscopy
For LM, we used a ZEISS AxioImager A1 microscope with
Nomarski optics, equipped with an AxioCam MRm digital
camera. For scanning electron microscopy (SEM), diatom
samples were placed on aluminum stubs, coated with
platinum–palladium, and studied with a ZEISS SUPRA 50
VP FE electron microscope at 10 kV.
Morphometric analysis
Valve shape was characterized using two approaches:
plots of valve width against length and multivariate
landmark-based shape analysis. Valve length and width
were measured from the digital images of specimens
observed in LM and scanned images from other publi-
cations. Three hundred and sixty-two specimens from 14
populations of the S. brebissonii–S. ovalis complex were
included in the analysis. LM images were taken from
slides GC6450a, GC49470b, GC104843a, GC107127b,
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
Valve shape variation in the Pinnatae group of Surirella 11
Table 1. Materials examined. Original identification as recorded in ANSP databases or in Diatom Herbarium catalogues.
Original ANSP slide ANSP material Collection
identification numbers numbers date Locality
Surirella minuta Feb3448 – – Glasgow, Scotland
S. panduriformis GC4055a – May 1952 Savannah River, South Carolina, USA
S. ovalis GC6450a – 6 November 1944 Willow Island, Nebraska, USA
S. ovalis GC6727a – 22 April 1936 Kayenta Creek, Arizona, USA
S. pinnata,S. minuta GC11277 – March 1860 Strehlen, Germany
S. pinnata GC11960 – – Longpré, France
S. pinnata GC43541 25 December 1926 Run near Merion Station, Pennsylvania, USA
S. ovalis GC49470b – – Leigh Spring, Virginia, USA
S. ovata,S. ovalis GC100963a GS020021 29 March 1993 Chicod Creek, North Carolina, USA
S. minuta,S. ovata;
type material for
S. lacrimula
GC101350a GS018351 7 June 1994 Accotink Creek, Virginia, USA
S. brebissonii GC104036a – 1 September 1999 Aberjona River, Massachusetts, USA
S. brebissonii GC104843a – 2 February 1999 Dawson Creek, Louisiana, USA
S. brebissonii GC105049a – 17 August 2000 Silver Creek, Utah, USA
S. brebissonii,
S. minuta
GC106518b – 3 September 2003 East Foster Creek, Washington, USA
S. brebissonii,
S. brebissonii
var. kuetzingii
GC107127a – 25 August 2004 Underwood Creek, Wisconsin, USA
S. brebissonii GC109797b – 2 June 2004 Dry Fork Marias, Montana, USA
S. brebissonii,
S. brightwellii
GC109807a MO000434 21 May 2004 Ross Fork Creek, Montana, USA
S. brebissonii GC109819a MO000446 19 June 2004 Wolf Creek, Montana, USA
S. brebissonii GC109825b MO000452 21 June 2004 Coffee Creek, Montana, USA
S. brebissonii GC109827b – 29 June 2004 Mason Gulch, Montana, USA
S. minuta GC110629a – 10 April 2001 Cane Creek, Tennessee, USA
S. angusta GC111668b CHRS0059 25 March 2008 Churchman’s Marsh, Delaware, USA
S. brebissonii var.
kuetzingii
GC112761a – 17 July 2007 Lake Creek, Montana, USA
S. minuta, possible
syntype
HLSEx528 – – Europe, sent by Brebisson to H.L. Smith
S. pinnata, isotype WmS64 – – Lewes, England
GC6727a, GC104036a, GC105049a, GC106518b,
GC109797b, GC109825b and GC112761a, all represent-
ing US samples, and scanned images of specimens from the
type population of S. ovalis,S. brebissonii and S. brebis-
sonii var. kutzingii published in Krammer & Lange-Bertalot
(1987) were used. Four hundred and twelve specimens
from 12 populations of the S. minuta species complex were
analyzed. LM images were taken from slides GC4055a,
GC43541, GC101350a, GC109827b, GC110629a and
GC111668b representing US samples; Feb3448 (Scotland),
GC11277 (Germany), WmS64 (isotype slide of S. pin-
nata, England), HLSEx528 (possible syntype of S. minuta,
Europe) and GC11960 (France).
For the landmark-based shape analysis, the same 362
specimens from 14 populations of the S. brebissonii–
S. ovalis complex were used, while only 338 speci-
mens from 7 samples of the S. minuta complex were
included. These were LM images from slides GC4055a,
GC111668b, GC101350a, GC109827b and GC110629a
(USA), WmS64 (isotype slide of S. pinnata, England)
and HLSEx528 (possible syntype of S. minuta, Europe).
We conducted landmark-based shape analysis using the
‘tps’ series of software (Rohlf 2007) available from
http://life.bio.sunysb.edu/morph/index.html. The tpsDig2
v. 2.16 program was used to digitize landmarks around the
circumference of the valves. Two of the landmarks were
fixed at the head and foot poles (LM 1, 2 respectively) and
two sets of 18 sliding semi-landmarks (LM 3–20 and 21–
38) were each placed on either side of the circumference
between the poles (Fig. 1). Both authors independently dig-
itized the landmarks for the S. minuta–S. pinnata complex
in order to test whether the method used to place the land-
marks affected the results. We each chose a different method
for the placement of the semi-landmarks: one was to place
landmarks at more frequent intervals on the circumference
where there was greater curvature in order to better approx-
imate the shape, and the other was to place the landmarks at
even distances along the outline of the valve. Shape analyses
on these two datasets were conducted separately.
The left and right half asymmetries were removed using
tpsUtil v. 1.46 to separate the landmarks into two groups:
one group (A) contained the two fixed landmarks (LM 1, 2)
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
12 English & Potapova
Fig. 1. Consensus configuration for S. minuta complex repre-
senting the placement of landmarks. Landmarks 1 and 2 are fixed.
Semi-landmarks 3–20 slide between 1 and 2. Semi-landmarks
21–38 slide between 1 and 2.
and the 18 semi-landmarks on the right side of the circum-
ference (LM 3–20); the other group (B) contained the two
fixed landmarks (1, 2) and the 18 semi-landmarks on the left
side (LM 21–38) (Fig. 1). The x-values of the landmarks in
group B were negated so that the two halves of the valves
would have the same direction of curvature when they were
analyzed. In order to make the semi-landmarks homolo-
gous to each other, we used tpsRelw v. 1.49 to slide the
semi-landmarks minimizing the bending energy (Bookstein
1996, Sheets et al. 2004), which allows for the most varia-
tion within a sample. We extracted the ‘aligned specimens’
data, which is based on the Procrustes superimposition and
used that to find the average the x- and y-values of corre-
sponding points (i.e., those symmetrically reflected across
the baseline) on the corresponding halves of the same spec-
imen. In order to maintain the homology of landmarks, we
slid the landmarks a second time using tpsRelw v. 1.49 in
case there were deformations produced by the averaging.
These final averaged and slid landmark values were used in
all shape analyses.
The shape descriptors called partial warps were obtained
from both datasets using tpsRelw v. 1.49 program. Princi-
pal component analysis, also called the relative warp (RW)
analysis of partial warps and uniform components, was
conducted using the same program. The first few relative
warps are considered to capture the major variation in shape
and are treated as composite shape variables. We plotted
RW against valve length to visualize ontogenetic allometric
trends in studied datasets.
Terminology
The sources of the terms used in this article are Ross et al.
(1979), Krammer & Lange-Bertalot (1987) and Ruck &
Kociolek (2004).
Results
Surirella brebissonii and S. ovalis species complex
The plot of valve length and width (Fig. 2) revealed
two groups of specimens corresponding to two groups of
growth trajectories. The group of specimens with a lower
width/length ratio halted abruptly at ∼38 μm, whereas the
group of relatively wider specimens extended in length
to 70 μm. The larger group contained both of the type
populations of S. brebissonii and S. ovalis, whereas the
smaller group contained the type population of S. brebis-
sonii var. kuetzingii. The two groups of specimens were not
well separated and converged at low valve lengths.
In the relative warp analysis, the first, second and third
relative warps accounted for 79.4, 10.53 and 5.15% of
shape variation, respectively. The rest of the relative warps
altogether accounted for less than 5% of variation. The first
warp (RW1) corresponded to the width and roundness of
the valve relative to its length (Fig. 3) and was moderately
correlated with valve length (R=0.44). The plot of RW1
versus length was similar to the plot of valve width versus
length and revealed two groups of shape trajectories that
overlapped in the smaller valves (Fig. 3). Note that for the
calculation of all population centroids and linear growth
trends, obvious outliers from three samples were not
included. One, two and three outliers were excluded from
GC106518b, GC105049a and GC112761a, respectively.
These specimens were excluded because we could not
be certain that they were from the same species as they
lay far outside the growth trend. As in the plot of valve
width versus length, the specimens from the type popu-
lations of S. brebissonii and S. ovalis fell into the same
group, whereas specimens from the type population of
S. brebissonii var. kuetzingii fell into a separate group. The
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
Valve shape variation in the Pinnatae group of Surirella 13
Fig. 2. Scatter plot of length and width for S. ovalis–S. brebissonii species complex. Regression lines represent ontogenetic–allometric
trends in individual populations. Specimens from type populations are represented by the hollow shapes.
Fig. 3. Plot of the first relative warp against valve length for S. ovalis–S. brebissonii species complex. Regression lines represent
ontogenetic–allometric trends in individual populations. Illustrations outside the plot represent shape approximations of the extreme
values of the warp.
populations from the USA fell into the same respective
groups as in the plot of length and width. The longest valves
of both groups had the same RW1 values, indicating that,
despite their difference in size, they have similar shapes.
The rate of change of this shape component with respect to
length in the S. brebissonii var. kuetzingii group was much
higher than in the S. brebissonii–S. ovalis group.
The second warp reflected the degree of iso- or het-
eropolarity or the position of the widest part of the valve
along the apical axis and was not correlated with valve
length (R=0.01). The population with the most heteropo-
lar valves on average was the type of S. ovalis, but RW2 did
not separate any population groups. No clusters were found
in the plot of the first and second relative warps.
The third relative warp (RW3) reflected the elongation
of the head pole and foot pole (Fig. 4), one of the dis-
tinguishing characteristics of S. ovalis. This relative warp
was correlated most strongly with length (R=−0.68).
No distinct groups of growth trajectories corresponding to
shape change captured by RW3 were found (Fig. 4), but
the type populations of S. brebissonii and S. brebissonii
var. kuetzingii did not vary in this shape character with
respect to length, whereas the type population of S. ovalis
did.
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
14 English & Potapova
Fig. 4. Plot of the third relative warp against valve length for S. ovalis–S. brebissonii species complex. Regression lines represent
ontogenetic–allometric trends in individual populations. Illustrations outside the plot represent approximations of the extremes of the warp.
Fig. 5. Plot of the first and third relative warps for S. ovalis–S. brebisosnii species complex. Centroids of individual populations with
standard error bars.
The plot of RW1 and RW3 showed no clear clusters
of specimens, but the centroids of individual populations
formed two groups along the RW3 axis (Fig. 5). The
composition of these groups was the same as in the width–
length plot and the RW1–length plot except that the type
of S. brebissonii was now placed in the same group as the
type of S. brebissonii var. kuetzingii.
Specimens from populations that comprised the
S. brebissonii var. kueztingii group in the length–width
plot and RW1 are considered here as representatives of
S. brebissonii var. kuetzingii, and those from the larger
group as S. ovalis (because the type specimens of these
species fell into respective groups). The type of S. brebis-
sonii fell into both groups, but had different ontogenetic
allometric trends from those in the S. brebissonii var.
kuetzingii group and had a different overall valve shape
from those in the S. ovalis group, and so is considered dis-
tinct from any US population. The following morphological
descriptions are given to describe ranges of variability of
these taxa observed so far in North America.
Surirella brebissonii var. kuetzingii Krammer &
Lange-Bertalot (Figs 6–10, 17–18)
This description is based on the following popula-
tions from the USA: GC6727a, GC104036a, GC105049a,
GC106518b, GC109797b, GC109825b and GC112761a.
The valves are ovate, 13–56 μm long and 10–24μm wide.
They are heteropolar, with a broadly rounded head pole and,
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
Valve shape variation in the Pinnatae group of Surirella 15
Figs 6–16. Surirella brebissonii var. kützingii and S. ovalis, LM. Figs 6–10. Surirella brebissonii var. kützingii. ANSP GC112761a,
Lake Cr., Montana. Fig. 9. Boundary of central area marked with arrow. Figs 11–16. Surirella ovalis.Fig. 11. ANSP GC6450a, Willow
Island, Nebraska. Figs 12–13. ANSP GC49470b, Leigh Spring, Virginia. Figs 14–16. ANSP GC104843a, Dawson Cr., Louisiana. Fig. 14.
Double fibula marked with arrow. Scale bar =10 μm.
in the smaller valves, a rounded foot pole, whereas in the
larger valves the foot pole is cuneate. The raphe fissure is
simple and is discontinuous externally at both the foot and
head poles. The valve face is slightly concentrically undu-
late (Fig. 9). The striae extend from the marginal zone to
the midline. Their density is 18–23 in 10 μm. The striae
are multiseriate, with 2-3 rows of round areolae near the
midline (Fig. 18, white arrow) and 4-5 rows of areolae near
the margin (Fig. 18, black arrow). On the valve face, these
striae pass through marginal depressions (Fig. 18), which
appear as large areolae in LM. Between two striae there is a
costa. There are usually 5–6 fibulae in 10 μm, but we have
observed some to have a density as high as 8 in 10 μm. Occa-
sionally, two fibulae are adjacent (Fig. 17, white arrow) with
only one intervening stria. The interfibular spaces are more
or less rectangular, tapering slightly towards the midline.
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
16 English & Potapova
Figs 17–22. Surirella brebissonii var. kützingii and S. ovalis, SEM. Figs 17–18. Surirella brebissonii var. kützingii. Fig. 17. MO000446,
Wolf Cr., Montana. Internal valve view. Double fibula marked with arrow. Fig. 18. MO000452, Coffee Cr., Montana. External valve view.
Rows of striae near midline (white arrow) and near margin (black arrow) Figs 19–22. Surirella ovalis. MO000434, Eoss Fork Cr., Montana.
Figs 19, 21. External valve views. Lower costa marked with arrow (Fig. 21). Fig. 20. Internal valve view. Fig. 22. External view of valve
face close to margin. Striae (white arrow) and costae (black arrow). Scale bars: Figs 17, 21 =4μm, Fig. 18 =2μm, Fig. 19 =6μm,
Fig. 20 =3μm, Fig. 22 =1μm.
The portulae are oval in shape and either one or two striae
wide (Fig. 17).
Surirella ovalis Brebisson 1838 (Figs 11–16, 19–23)
This description is based on the following populations
from the USA: GC6450a, GC49470b, GC104843a and
GC107127b. The valves are ovate, 17–70 μm long and
14–37 μm wide, and vary in shape from very slightly het-
eropolar to heteropolar. In the shortest valves, both apices
are broadly rounded (Fig. 16); in valves of intermediate
length, the head pole is usually broadly rounded while
the foot pole is cuneate (Fig. 15); in the longest valves,
both apices are cuneate (Figs 11, 12), although the head
pole remains broader throughout whole length range. The
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
Valve shape variation in the Pinnatae group of Surirella 17
width-to-length ratio ranges from 1.5 to 2. The raphe fis-
sure is simple and is discontinuous externally at both the
foot and head poles. The valve face is strongly concentri-
cally undulate in many of the larger frustules (Figs 11, 19).
The striae are radiate across the valve face from the midline
with a density of 14–19 in 10 μm. The multiseriate striae
have 2–3 rows of areolae near the midline (Fig. 22, black
arrow) and 4–5 rows near the margin (Fig. 22, white arrow).
The striae pass through marginal depressions on the valve
face. The striae contain slit-shaped areolae on the external
valve surface (Fig. 22) and round areolae on the internal
valve surface (Fig. 20). Between the striae there are costae
and every third or fourth costa on the external surface of the
valve, which corresponds to the fibulae on the interior of the
valve, is lower than the surrounding costae (Fig. 21). The
fibulae are broad and marginal, 5–6 in 10 μm. The inter-
fibular spaces are rectangular to triangular and contain one
to three round or oval portulae. Occasionally, two fibulae
are adjacent (Fig. 14) with only one intervening stria.
Surirella minuta and S. pinnata species complex
The plot of valve length and width revealed several groups
of specimens (Fig. 23). Specimens from two populations
formed two smaller groups: one included a few specimens
from slide GC111668b (Delaware, DE), which were consid-
erably longer than the others; while another group consisted
of specimens from slide GC11960 (France), which were
considerably wider than the others. There were also much
smaller valves in the sample from Delaware that were the
same size as the majority of specimens, but there was a
significant gap between the small and the large valves.
The rest of the specimens comprised a large cluster, within
which two partially overlapping groups corresponding to
two shape trajectories could be distinguished. The valves
of trajectory A had a smaller length range than those of
trajectory B. The valves of trajectory A did not exceed
∼30 μm, whereas those of trajectory B extended up to
48 μm in length, but remained relatively more narrow.
These trajectories separated the largest valves, but con-
verged in the smaller valves. The majority of specimens,
including all of those from the USA, fell into one of
these two groups, but two European populations spanned
both groups: HLSEx528, the possible syntype population
of S. minuta, and the specimens from the lectotype slide
of S. minuta.
In addition to the two type populations of S. minuta
and S. pinnata, we chose five representative samples from
the USA for geometric–morphometric analysis, which will
hereafter in the text and figures be referred to by the fol-
lowing abbreviations: GC4055a from the South Carolina
(SC), GC111668b from Delaware (DE), GC101350a from
Virginia (VA), GC109827a from Montana (MT) and
GC110629a from Tennessee (TN). In the relative warp anal-
ysis, the first, second and third relative warps accounted for
87.9, 6.22, and 2.15% of variation, respectively. The rest of
the relative warps all together captured <4% of the shape
variation. RW1 reflected the variation in the relative width
and the shape of the middle part of the valve (Fig. 24) and
was strongly correlated with length (R=0.83). This rel-
ative warp reflected the general pattern of shape change
through the life cycle, namely that as the cells become
shorter, their width-to-length ratio increases. Most popu-
lations shared this trend and had similar rates of change
in RW1 with respect to length (Fig. 24). The one popu-
lation that did not have a similar rate of change was DE,
which had significantly longer valves. Samples MT, VA,
TN and the syntype of S. minuta had, on average, low
RW1 scores; thus they were, on the whole, relatively wider
than the other populations. The RW1 ontogenetic trajecto-
ries of these four populations formed a tight group. Three
populations with high average RW1 scores and more elon-
gated valve shapes were DE, SC and the isotype population
of S. pinnata.
The second relative warp reflected the degree of iso- or
heteropolarity, or the position of the widest part valve along
apical axis. It showed a very weak correlation with length
(R=0.14), and did not separate any distinct groups of
specimens, allometric trends or centroids. All populations
showed a tendency to become more isopolar in the smaller
valves, whereas the larger valves had varying degrees of het-
eropolarity. All of the populations converged in the RW2
values of their smaller valves and thus had the same degree
of isopolarity.
The third relative warp reflected the variability of the
head and foot poles, especially how narrow the foot pole
was (Fig. 25). This relative warp also had a very weak
correlation with length (R=−0.10). Specimens were not
separated by this warp into distinct clusters (Fig. 25), but
the centroids of the populations showed a significant sepa-
ration between the syntype of S. minuta and MT, VA and TN
(Fig. 26). The specimens of MT, VA and TN had on aver-
age low RW3 scores (a more narrow footpole), whereas
the syntype of S. minuta had high RW3 scores (a more
rounded footpole).
In the plot of RW3 versus RW1, the centroids of popula-
tions formed three clearly defined groups (Fig. 26). Higher
RW1 values separated the isotype of S. pinnata, DE and SC
from the other populations, which meant that their valves
were narrower on average. Higher RW3 and lower RW1
separated the syntype of S. minuta and thus this popula-
tion had wider valves with more rounded footpoles than the
other populations. Three populations, VA, MT and TN, had
low RW3 and RW1 values, and so their valves are generally
wider with narrower footpoles.
We obtained the same results of the RW analysis using
two somewhat different ways of placing landmarks on valve
outlines (Figs 27–28). Although there was some minor vari-
ability in the placement of individual specimens in the shape
space, the major shape trends revealed along the RW axes
remained the same (Fig. 27). The main difference in the
results was that even after the landmarks had been slid, they
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
18 English & Potapova
Figs 23. Scatter plot of length and width for S. minuta–S. pinnata species complex. Specimens from type populations are represented
by the hollow shapes. A and B represent two growth trajectories.
Figs 24. Plot of the first relative warp against valve length for the S. minuta–S. pinnata species complex. Regression lines represent
ontogenetic–allometric trends in populations. Illustrations outside the plot represent approximations of the extremes of the warp.
still maintained approximately the density in which they
were placed (Fig. 28), i.e., either equally spaced around the
circumference or more dense in areas of greater curvature.
Despite these differences in placement around the valve cir-
cumference, the RW analysis was able to abstract the same
shape components out of the two datasets.
Three groups of populations within S. minuta species
complex revealed by the landmark analysis were suffi-
ciently different in shape and ontogenetic trends to be con-
sidered different species. One group contained the type of
S. minuta and another the type of S. pinnata. Although these
species have been synonymized, our analysis suggested that
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
Valve shape variation in the Pinnatae group of Surirella 19
Figs 25. Plot of the third relative warp (RW3) of shape variation against valve length for the S. minuta complex. Regression lines represent
ontogenetic–allometric trends in populations. Illustrations outside the plot represent approximations of the extremes of the warp.
Figs 26. The plot of the first relative warp (RW1) against the third (RW3) for the S. minuta complex. Centroids of individual populations
with standard error bars.
they need to be maintained. Further examination showed
that these species also differ in the fibula structure. Updated
morphological descriptions of S. minuta and S. pinnata are
given below. The third group of populations (MT, VA and
TN) was different enough in shape from other two to justify
the description of a new species, S. lacrimula.
Surirella minuta Brebisson in Kutzing 1849 (Figs 31–32)
This description is based on the population from the poten-
tial syntype of S. minuta, HLSEx528. The valves are linear
to ovate and 16–36 μm long and 9–11 wide, with a width-
to-length ratio between 1.5 and 4. The head pole is broadly
rounded, while the foot pole is more cuneate. The striae are
fine with a density of 28–30 in 10 μm. The striae are radiate
at the apices and parallel in the middle of the valve. Each
stria begins at the midline and terminates at the keel. The
fibulae, 7–8 in 10 μm, begin at the valve mantle and extend
about half the distance to the apical axial line (Fig. 32).
Surirella pinnata Smith 1853 (Figs 34–38, 41, 43)
This description is based on specimens from two popu-
lations: the isotype of S. pinnata WSM64 and the pop-
ulation from South Carolina, GC4055a. The population
from Delaware, GC11668b, (Figs 33, 42, 44) was not
included in the morphological analysis because, although
the analysis showed strong similarity in shape charac-
ters, the ontogenetic–allometric trends differed significantly
from the other populations of S. pinnata. The population
from slide Delaware was not abundant enough to determine
whether it contained a new species. The small valves in this
population were indistinguishable from S. pinnata, but it
is unknown whether the large valves were from a separate
species.
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
20 English & Potapova
Figs 27. The plot of a second set of data generated by a different placement of landmarks in the same specimens. First relative warp
(RW1) against the third (RW3) of shape variation for the S. minuta complex. Centroids of individual populations with standard error bars.
Figs 28. Comparison of the final position of landmarks for the two different ways of positioning landmarks in the S. minuta–S. pinnata
dataset. Pairs (a and b) of the extreme values of the first three relative warps of shape variation. Hollow circles represent placement at
equal distances around the circumference and filled circles represent placement of more landmarks at areas of higher curvature.
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
Valve shape variation in the Pinnatae group of Surirella 21
Figs 29–38. Representatives of S. minuta–S. pinnata species complex, LM. Figs 29–30. Surirella lacrimula sp. nov. ANSP GC109827b,
Mason Gulch, Montana. Figs 31–32. Surirella minuta ANSP HLSEx528, Europe. Fig. 32. Short fibula marked with arrow. Fig. 33.
Surirella cf. pinnata. ANSP GC111668b, Churchman’s Marsh, Delaware. Figs 34-38. Surirella pinnata. Figs 34–36. ANSP WmS64,
Lewes, England. Fig. 36. Fibula extending to midline marked with arrow. Figs 37–38. ANSP GC4055b, Savannah R., South Carolina.
Scale bar =10 μm.
The valves of S. pinnata are linear, 14–45 μm long and
7–10 μm wide, with a width-to-length ratio of 2–4.5. The
head pole is broadly rounded, whereas the foot pole is more
cuneate. The raphe fissure is simple and is discontinuous
externally at both the foot and head poles. The density of
the striae in the isotype material was lower than observed
in the S. pinnata in the USA at 24–25 in 10 μm rather
than 26–31 in 10 μm. Corresponding to the fibulae on the
internal valve surface, every fourth costa on the valve face is
lowered below the plane of the other costae (Fig. 41). There
are 3–5 rows of circular areolae in each multiseriate stria
and on the valve face these terminate in a marginal depres-
sion (Fig. 43). The narrow fibulae, 6–8 in 10 μm, begin at
the valve mantle and extend to the midline (Fig. 36). The
interfibular spaces are rectangular and contain one portula.
There are 3–5 striae between the fibulae. The portulae are
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
22 English & Potapova
Figs 39–44. Surirella lacrimula sp. nov., S. pinnata and S. cf. pinnata, SEM. Figs 39–40. Surirella lacrimula sp. nov. Type material,
ANSP GS018351, Accotink Cr., Virginia. Fig. 39. External valve view. Fig. 40. Internal view of a head pole. Figs 41, 43. Surirella pinnata.
ANSP GS020021. Chicod Cr., North Carolina. Fig. 41. External valve view. Fig. 43. External view of a head pole. Figs 42, 44. Surirella
cf. pinnata. ANSP CHRS0059, Churchman’s Marsh, Delaware. Fig. 42. Internal valve view. Fig. 44. Internal view of valve margin with
portulae. Scale bars: Figs 39–41 =3μm, Fig. 42 =6μm, Figs 43–44 =1μm.
generally rectangular in shape and are one to two striae
wide.
Surirella lacrimula English sp. nov. (Figs 29-30, 39-40,
45-54, 55-60)
Diagnosis. Valvae ovatae, 19–30 μm longae, 10–13 μm
latae, facie valva leviter circulatim undata, polo ‘capite’
late rotundo et polo ‘pede’ tereti vel anguste cuneato.
Carina humili. Striae tenues densae, ad apices radiantes,
parallelae medio, ad marginem ab linea–mediale exten-
dentes 27–33/10 μm; fibulae tenues margini terminatae
mediatenus de margine ad axem apicalem, 60–80/100 μm.
Description. Valves ovate, 19–30 μm long, 10–13 μm
wide, valve face slightly concentrically undulate with a
broadly rounded head pole and rounded to narrowly cuneate
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
Valve shape variation in the Pinnatae group of Surirella 23
Figs 45–54. Surirella lacrimula sp. nov., LM. Figs 45–50. Type material, Accotink Cr., Virginia. Fig. 45. Holotype specimen.
Figs 51–53. ANSP GC110629a, Cane Cr., Tennessee. Fig. 54. ANSP GC109827b, Mason Gulch, Montana. Scale bar =10 μm.
foot pole. Low keel. Fine dense striae, radiate at apices,
parallel in the middle, extending to margin from midline,
27–33/10 μm; fine marginal fibulae terminating part way
from the valve margin to the midline, 60–80/100 μm.
Holotype. Circled specimen on the slide ANSP GC58959
(Fig. 45), Diatom Herbarium, Academy of Natural Sci-
ences, Philadelphia (ANSP).
Isotypes. Circled specimens on slides CANA 85057,
ANSP GC58964.
Type locality. Accotink Creek near Annandale, Virginia,
USA. 38.8128 N, 77.2286 W. Collected by United States
Geological Survey (USGS) staff on 7 June 1994.
Etymology. The name of this species comes from the Latin
lacrima, which means, ‘tear’. It is thus named for its small
size and its distinctive teardrop shape.
Morphological details
Portulae. Similarly to S. minuta, the portulae are
generally rectangualar with rounded corners (Fig. 59).
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
24 English & Potapova
Figs 55–60. Surirella lacrimula sp. nov., SEM. Type material, ANSP GS018351, Accotink Cr., Virginia. Fig. 55. Internal valve view.
Fig. 56. External valve view. Fig. 57. External view of the foot pole. Fig. 58. External view of the head pole. Fig. 59. Interior view of a valve
margin with portulae. Fig. 60. External view of a valve margin. Marginal depression marked with arrow. Scale bars: Figs 55–56 =4μm,
Figs 57–60 =1μm.
There is usually one portula between two fibulae, but occa-
sionally there are extra, much smaller portulae between a
large portula and a fibula. The portulae open onto a cavity
or canal formed between the mantle wall and the fibulae.
Fibulae. The fibulae are narrow (∼200 nm) and proceed
from the mantle ∼2μm to merge into a single costa
(Fig. 59). On the valve surface, they correspond to costae
that are lower than the surface of the valve.
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
Valve shape variation in the Pinnatae group of Surirella 25
Striae. The multiseriate striae run between the costae from
the midline to the valve margin and into a marginal depres-
sion (Fig. 60). In the interior of the valve, the striae can be
seen to extend through the portulae to into the raphe canal.
There are usually 2–3 rows of areolae at the apical axis and
3–4 rows of areolae at the valve margin.
Areolae. The areolae seen from the valve interior are cir-
cular, while on the valve surface they are more oblong.
There is a single row of widely spaced areolae on the apical
axis, which are slightly raised on the valve surface.
Marginal depression on the valve face. There is a
marginal depression at the end of each intercostal striae
on the valve face. The position of the marginal depression
corresponds to the junction of the interior canal wall and
the valve interior surface.
Mantle. The mantle is slightly concave. The mantle has
costae and rows of areolae which line up with those on the
valve face. No other ornamentations were present.
Raphe. The raphe slits are externally discontinuous at
both the head and the foot poles. The external raphe ends
are simple. The raphe can be seen in the canal cavity behind
the portulae.
Distribution. Surirella lacrimula is apparently a widely
distributed species across North America. It was found
in most of the ANSP slides with records of S. minuta or
S. pinnata and many of the records of S. brebissonii,
S. brebissonii var kuetzingii and S. ovata that we checked.
Discussion
Our analysis revealed multiple ontogenetic allometric
trends and shape groups within species complexes in the
Pinnatae group of the genus Surirella. Although there was
a statistically significant difference in shape among groups,
their ranges of variation overlapped, including striae and
fibulae density (Table 2), particularly in the smaller valves.
This means that at present it is impossible to draw precise
species boundaries within this group based on morphome-
tric characters and some specimens cannot be identified
with certainty to species level. However, we found that
quantitative characterization of valve shape and allomet-
ric trends were useful for exploring species diversity within
this taxonomic group. The landmark-based shape analysis
was shown to be unaffected by the person placing the land-
marks because we obtained almost identical results between
two independently collected datasets.
Landmark-based shape analysis revealed two shape
groups in S. brebissonii–S. ovalis species complex:
one group contained the type population of Surirella
ovalis, whereas the other included type populations of
S. brebissonii and S. brebissonii var. kuetzingii. The two
latter varieties had the same valve shape, but differed in
their dominant ontogenetic trends. The growth trajectory
of S. brebissonii was actually similar to that of S. ovalis.
Although it was already known that S. brebissonii var.
kuetzingii had a much smaller size range than S. brebissonii
(Krammer & Lange-Bertalot 1987), our analysis showed
that this variety passes through the same range of shape
transformation as S. brebissonii, but does so within a
much smaller size range. The valves of S. brebissonii var.
kuetzingii also change more in shape over a given decrease
in length than those of S. brebissonii. The specimens from
the USA, which were identified as a nominate variety of
S. brebissonii or S. brebissonii var. kuetzingii, all conformed
to the growth trajectory of S. brebissonii var. kuetzingii.
There is insufficient data at present to determine whether
the nominate variety of S. brebissonii and var. kuetzingii
represent separate biological species, but it is notable that
we found no populations with intermediate growth trends
between these two varieties. Surirella brebissonii var. kuet-
zingii seems to be more common than the nominate variety
not only in North America, but also in Europe, because
relatively short valves for the most part are illustrated in
European floras (e.g., Germain 1981: pl. 152, fig. 108 as
S. ovata Kützing; Snoeijs 1993).
Most studied populations in the S. minuta–S. pinnata
species complex had similar ontogentic allometric trends.
The shape analysis revealed, however, three different shape
groups. These groups overlapped, but were sufficiently
Table 2. Morphological characteristics of selected low-keel Surirella species.
Taxa, materials Length (μm) Width (μm) Striae/10 μm Fibulae/10 μm
Surirella brebissonii var. kuetzingii, from all USA populations 13–38 10–18 18–24 5–8
S. iowensis, from all USA populations 21–54 15–33 16–26 6–8
S. lacrimula sp. nov. from all USA populations 19–30 10–13 27–33 6–8
S. lacrimula sp. nov., type population, GS018351 19–28 10–12 27–32 6–8
S. minuta, HLSEx528 syntype? 16–36 9–11 28–30 7–8
S. minuta cf., CHRS0059 18–70 7–14 20–31 5–8
S. ovalis, from all USA populations 17–70 14–37 14–19 5–6
S. pinnata, GC4055a 17–35 7–10 26–31 7–8
S. pinnata, WmS64 isotype 14–45 8–10 24–25 6–7
S. suecica, from all USA populations 14–34 7–10 30–36 9–12
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
26 English & Potapova
statistically different to justify their identification as species.
One, S. pinnata, was separated from the other populations
by the first relative warp, or the relative width and round-
ness of the valve. The second, S. minuta, was separated
from the remaining populations by the third relative warp,
or the shape of its poles. The remaining populations had
very similar ontogenetic trends and shape averages, and so
they were described as a new species, S. lacrimula.
Surirella lacrimula differs from S. minuta by its
smaller range of length and by the more acute foot pole.
Surirella pinnata can be distinguished from S. lacrimula by
not only its narrowly linear valves shape, but also fibulae
extending fully to the midline, whereas those of S. lacrimula
stop far from it. Surirella lacrimula differs from S. ovalis,
S. iowensis, S. stiria and S. brebissonii primarily by its
higher density of costae, 27–32 in 10 μm instead of 14–
24 in 10 μm. It is also never longer than 30 μm, whereas
S. ovalis,S. iowensis and S. brebissonii can be >55 μm.
Surirella iowensis can also have a valve face that is twisted
about the apical axis, but S. lacrimula has only a flat valve
face. Surirella lacrimula differs from S. suecica,S. stalgma
and S. atomus in the structure of its fibulae and the shape of
the valve. The fibulae of these three species are very short
and wide and do not extend from the margin of the valve,
whereas S. lacrimula has narrower fibulae that extend part
way across the face of the valve. The foot poles of these
species are narrower than S. lacrimula and the foot pole of
S. stalgma is always capitate.
Based on this examination of the type populations, there
is evidence to suggest that S. minuta and S. pinnata should
not be synonymized. Both the morphometric analysis of
valve shape and differences in the structure of the fibulae
indicate that these are two similar but separate taxa. The
valves of S. pinnata are narrowly linear, whereas those of
S. minuta are relatively wider. The fibulae in S. minuta do
not extend to the midline like those in S. pinnata.
It is difficult to determine whether the studied popula-
tion of S. minuta from the Brébisson collection (slide ANSP
HLSEx528) was necessarily a syntype of this species. The
original description of S. minuta (Kützing 1849) did not pro-
vide many morphological details of the species. The length
range supplied in the description, 21–45 μm, is longer than
the range in the possible syntype HLSEx528, 16–36 μm,
and the fibulae in the syntype are twice as dense (7–8
in 10 μm) as in the original description (3–4 in 10 μm).
Krammer & Lange-Bertalot (1987) examined the original
material of S. minuta, and their illustrations show specimens
with the same dimensions, shape and density of structural
elements as in the possible syntype population that we
studied.
In studied materials from the USA, S. pinnata was fairly
rare although S. lacrimula was common. Surirella lacrim-
ula also appears to be present in Europe (Ector & Hlúbiková
2010: pl. 115, figs 69–76, pl. 116, figs 1–6). No USA
specimens fit the third shape group, S. minuta.
It has already been shown by Stoermer & Ladewski
(1982) that quantitative shape analysis can reveal shape dif-
ferences in diatoms that are not always seen with human
eye. Our study confirms this and demonstrates that shape
analysis including a study of ontogenetic allometry may be
a valuable exploratory tool in discovering diatom species
diversity.
References
Bookstein F.L. 1996. Combining the tools of geometric morpho-
metrics. In: Advances in morphometrics (Ed. by L.F. Marcus
et al.), pp. 131–151. Plenum Press, New York.
Bookstein F.L. 1997. Landmark methods for forms without land-
marks: localizing group differences in outline shape. Medical
Image Analysis 1: 225–243.
Ector L. & Hlúbiková D. 2010. Atlas des diatomées des Alpes-
Maritimes et de la Région Provence-Alpes-Côte d’Azur.
Centre de Recherche Public – Gabriel Lippmann, Belvaux,
393 pp.
Edgar R.K., Kociolek J.P. & Edgar S.M. 2004. Life cycle asso-
ciated character variation in Aulacoseira krammeri sp. nov., a
new Miocene species from Oregon (USA). Diatom Research
19: 7–32.
Edgar S.M & Theriot E.C. 2004. Phylogeny of Aulaco-
seira (Bacillariophyta) based on morphology and molecules.
Journal of Phycology 40: 772–788.
Edlund M. & Soninkhishig N. 2009. The Navicula reinhardtii
species flock (Bacillariophyceae) in ancient lake Hövsogöl,
Mongolia: description of four taxa. Nova Hedwigia, Beiheft
135: 239–256.
Falasco E., Blanco S., Bona F., Gomà J., Hlúbiková D.,
Novais M.H., Hoffmann L. & Ector, L. 2009. Taxon-
omy, morphology and distribution of the Sellaphora stroemii
complex (Bacillariophyceae). Fottea 9: 243–256.
Geitler L. 1932. Der Formwechsel der pennaten Diatomeen
(Kieselalgen). Archiv für Protistenkunde 78: 1–226.
Germain H. 1981. Flore de Diatomées, eaux douces et soumâtres
du massif Armoricain et des contrées voisines d’Europe
occidentale. Societe Nouvelle Des Editions Boubee, Paris.
444 pp.
Krammer K. & Lange-Bertalot H. 1987. Morphology and tax-
onomy of Surirella ovalis and related taxa. Diatom Research
2: 77–95.
Krammer K. & Lange-Bertalot H. 1988. Bacillariophyceae.
2. Teil: Bacillariaceae, Epithemiaceae, Surirellaceae. In:
Süsswasserflora von Mitteleuropa (Ed. by H. Ettl et al.), Vol.
2/2. G. Fischer Verlag, Jena. 596 pp.
Kützing F.T. 1849. Species Algarum. Leipzig. F.A. Brockhaus,
922 pp.
Mann D.G., McDonald S.M., Bayer M.M., Droop S.J.M.,
Chepurnov V.A., Loke R.E., Ciobanu A. & du Buf J.M.H.
2004. The Sellaphora pupula species complex (Bacillario-
phyceae): morphometric analysis, ultrastructure and mating
data provide evidence for five new species. Phycologia 43:
459–482.
Mou D. & Stoermer E.F.1992. Separating Tabellaria (Bacillario-
phyceae) shape groups based on Fourier descriptors. Journal
of Phycology 28: 386–395.
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012
Valve shape variation in the Pinnatae group of Surirella 27
Peragallo H. & Peragallo M. 1897–1908. Diatomées Marines
de France et des Districts Maritimes Voisins (Ed. by J. Tem-
père), Micrographe-Editeur, à Grez-sur-Loing (S.-et-M.).
491 pp.
Potapova M. & Hamilton P.B. 2007. Morphological and eco-
logical variation within the Achnanthidium minutissimum
(Bacillariophyceae) species complex. Journal of Phycology
43: 561–575.
Potapova M.G. & Ponader K.C. 2004. Two common North
American diatoms, Achnanthidium rivulare sp. nov. and
A. deflexum (Reimer) Kingston: morphology, ecology and
comparison with related species. Diatom Research 19:
33–57.
Rohlf F.J. 2007. TPS software series. Department of Ecology and
Evolution, State University of New York, Stony Brook.
Rohlf F.J. & Slice D.E. 1990. Extensions of the Procrustes
method for the optimal superimposition of landmarks.
Systematic Zoology 39: 40–59.
Ross R., Cox E.J., Karayeva N.I., Mann D.G., Paddock T.B.,
Simonsen R. & Sims P.A. 1979. An amended terminology for
the siliceous components of the diatom cell. Nova Hedwigia,
Beiheft 64: 513–533.
Ruck E. & Kociolek J.P. 2004. Preliminary phylogeny
of the family Surirellaceae (Bacillariophyta). Bibliotheca
Diatomologica 50: 1–236.
Sheets H.D., Keonho K, Mitchell C.E. 2004. A combined
landmark and outline-based approach to ontogenetic shape
change in the Ordovician trilobite Triarthrus becki. In:
Applications of morphometrics in paleontology and biology
(Ed. by A. Elewa), pp. 67–81. Springer, New York.
Smith W. 1853. Synopsis of British Diatomaceae. Vol. 1. John
Van Voorst, London. 89 pp.
Snoeijs P. 1993. Intercalibration and distribution of diatom
species in the Baltic Sea, Vol. 1. Opulus Press, Uppsala.
129 pp.
Stoermer E.F. & Ladewski T.B. 1982. Quantitative
analysis of shape variation in type and modern populations
of Gomphoneis herculeana.Nova Hedwigia, Beiheft 73:
347–386.
Theriot E. 1988. An empirically based model of variation in rota-
tional elements in centric diatoms with comments on ratios
in phycology. Journal of Phycology 24: 400–407.
Theriot E. & Stoermer E.F. 1984. Principal component
analysis of Stephanodiscus: observations on two new species
from the Stephanodiscus niagarae complex. Bacillaria 7:
37–58.
Veselá J., Neustupa J., Pichrtová M. & Poulíèková A.
2009. Morphometric study of Navicula morphospecies
(Bacillariophyta) with respect to diatom life cycle. Fottea 9:
307–316.
Downloaded by [Consortium Luxembourg] at 10:42 28 February 2012