Molecular Analyses of the Genus Ilex (Aquifoliaceae) in Southern South America, Evidence
from AFLP and ITS Sequence Data
Author(s): Alexandra M. Gottlieb, Gustavo C. Giberti and Lidia Poggio
Source:
American Journal of Botany,
Vol. 92, No. 2 (Feb., 2005), pp. 352-369
Published by: Botanical Society of America, Inc.
Stable URL: https://www.jstor.org/stable/4123880
Accessed: 18-12-2018 17:13 UTC
REFERENCES
Linked references are available on JSTOR for this article:
https://www.jstor.org/stable/4123880?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Botanical Society of America, Inc.
is collaborating with JSTOR to digitize, preserve and
extend access to
American Journal of Botany
This content downloaded from 157.92.6.34 on Tue, 18 Dec 2018 17:13:49 UTC
All use subject to https://about.jstor.org/terms
American Journal of Botany 92(2): 352-369. 2005.
MOLECULAR ANALYSES OF THE GENUS ILEX
(AQUIFOLIACEAE) IN SOUTHERN SOUTH AMERICA,
EVIDENCE FROM AFLP AND ITS SEQUENCE DATA'
ALEXANDRA M. GOTTLIEB,1,3 GUSTAVO C. GIBERTI,2 AND
LIDIA POGGIO'
'Universidad de Buenos Aires, Facultad de Ciencias Exactas y Naturales, Ciudad Universitaria, Pabell6n II, 4to piso, Departamento
de Ecologifa, Gen6tica y Evoluci6n, Lab. 62, C1428EHA, Buenos Aires, Argentina; and 2Instituto de Quifmica y Metabolismo del
Ffirmaco (IQUIMEFA), Junin 956, 1113-Buenos Aires, Argentina
In order to clarify the relationships among southern South American (sSA) representatives of the genus Ilex, an amplified fragment
length polymorphism (AFLP) analysis was accomplished. In addition, the phylogenetic relationships of the species were studied using
ribosomal internal transcribed spacer (ITS) sequence data alone and in combination with AFLP data, taking into account the possible
existence of paralogous sequences and the influence of alignment parameters. To explore stability of phylogenetic hypotheses, a
sensitivity analysis was performed using 15 indel-substitution models. Within each species assayed, the AFLPs allowed the recognition
of several diagnostic bands. Furthermore, the AFLP analysis revealed that individuals belonging to the same morpho-species formed
coherent clades. In addition, some cases of geographical association were noted. Studies on ITS sequences revealed divergence between
data obtained herein and sequence data downloaded from GenBank. The sensitivity analyses yielded different interspecific hypotheses
of relationships. Notwithstanding, analyses of the ITS data alone and in combination with AFLPs, rendered clades stable to variation
in the analytical parameters. Topologies obtained for the AFLPs, the ITS data alone and the combined analyses, demonstrated the
existence of a group formed by I. argentina, I. brasiliensis, I. brevicuspis, L integerrima, and L theezans, and that L dumosa and I.
paraguariensis were distantly related to the former. Incongruence with traditional taxonomical treatments was found.
Key words: AFLP; direct optimization; Ilex; ITS; sensitivity analysis.
The plant genus Ilex L., which belongs to the holly family,
Aquifoliaceae Bartl., comprises deciduous and evergreen
bushes or trees with economic importance as crops and or-
namentals. These plants are functionally dioecious; rudimen-
tary pistils in staminate flowers and sterile stamens in pistillate
flowers are borne on different plants. A remarkable morpho-
logical diversity has been described in the genus, including
trees to creeping prostrate plants or epiphytic shrubs. It has
more than 400 species, dispersed mainly in America and Eur-
asia, but also in Oceania and Africa (Giberti, 1994b).
South America is considered one of the main areas of di-
versification of Ilex, together with East Asia (Loesener, 1942;
Cuenoud et al., 2000). In southern South America (sSA) the
species are mostly found in areas with tropical or subtropical
climates, mainly in northeastern Argentina, southeastern Brazil
and eastern Paraguay. Twelve species have been described for
sSA (Loesener, 1901; Lillo, 1911; Edwin and Reitz, 1967; Gi-
berti, 1998), namely, Ilex affinis Gardner, I. argentina Lillo, L
brasiliensis (Sprengel) Loes., I. brevicuspis Reissek, I. cha-
maedryfolia Reissek, L dumosa Reissek, L integerrima (Vell.)
Reissek, L microdonta Reissek, L paraguariensis A. St. Hil.,
I. pseudobuxus Reissek, I. taubertiana Loes., and I theezans
C. Martius ex Reissek. Of those mentioned, I. paraguariensis
is the most relevant from an economic perspective. In Argen-
tina, southern Brazil, Paraguay, and Uruguay, the aerial parts
of this plant are used to prepare a very popular beverage called
mate, which is highly appreciated for its flavor and stimulating
properties due to its caffeine and theobromine contents (Filip
et al., 2001).
About a hundred years after Linnaeus (1753) published the
diagnosis for the genus Ilex, Gray (1856) attempted to estab-
lish an infrageneric classification. In 1861, Reissek published
the greatest species inventory for the time, followed by Hooker
(1862) who reported the occurrence of about 145 species
worldwide. Loesener (1891, 1901, 1908, 1942) published a
series of seminal studies recording almost 280 species of Ilex,
dispersed globally, and recognizing several subgeneric ranks
although under an unconventional hierarchical arrangement.
Since then, several authors have recognized some inaccuracies
in Loesener's system and proposed emendations in the system-
atics of the genus (Rehder, 1927; Hu, 1949, 1950; Krtissmann,
1962; Baas, 1975; Lobreau-Callen, 1975; Martin, 1977). Spe-
cies delimitation in Ilex, based on overall morphological sim-
ilarity, is complicated due to the various concepts used by
different authors to define the taxa and to the lack of complete
information, particularly for evergreen species (Galle, 1997).
In order to investigate the relationships among the species
of Ilex found in sSA, we implemented the amplified fragment
length polymorphism (AFLP) technique (Vos et al., 1995). In
addition, we studied the phylogenetic relationships of the spe-
cies using ribosomal internal transcribed spacer (ITS) se-
quence data alone and in combination with AFLP data, taking
into account the possible existence of paralogous sequences
and the influence of alignment parameters.
' Manuscript received 11 December 2003; revision accepted 17 September
2004.
The authors are indebted to L. D. Belingheri, S. D. Prat Kricun, and R. M.
Mayol from EE. INTA Cerro Azul (Misiones, Argentina) who kindly provided
specimens from living collections; to N. Jouv6 and P. Rubio of University of
Alcald (Madrid, Spain) who supported the sequencing, to M. E. Rol6n from
the "Establecimiento Las Marias" (Corrientes, Argentina) who facilitated
commercial lines of "mate," to S. A. Catalano and V. A. Confalonieri for
their invaluable suggestions. We also want to thank three anonymous review-
ers whose suggestions helped to improve this manuscript. This work was
supported from grants of the "Agencia Nacional de Investigaciones Cientifi-
cas y Tecnicas" (Argentina) (PICT 01-4443) and from CONICET (PIP 0559/
98).
SAuthor for correspondence. E-mail: gottlieb@bg.fcen.uba.ar.
352
This content downloaded from 157.92.6.34 on Tue, 18 Dec 2018 17:13:49 UTC
All use subject to https://about.jstor.org/terms
February 2005] GOTTLIEB ET AL.-MOLECULAR ANALYSES OF SOUTH AMERICAN ILEX 353
Beside the general use of AFLP markers in the estimation
of genetic diversity and differentiation among individuals,
populations, and plant species, they have also been success-
fully used in phylogenetic analysis between closely related
species (Cervera et al., 1998; Kardolus et al., 1998; Mace et
al., 1999; Negi et al., 2000; Ren and Timko, 2000; Koopman
et al., 2001). Because of the rapidity with which reliable high
resolution markers can be generated, the AFLP technique is
considered a powerful tool for molecular studies (Mueller and
Wolfenbarger, 1999). In addition, the high ratio of polymor-
phism generated per PCR experiment (the multiplex ratio)
(Powell et al., 1996) and the random distribution of most
AFLP bands across the genome (Zhu et al., 1998) are consid-
ered key features of this technique. Several drawbacks have
been detected for AFLPs (Robinson and Harris, 1999), which
introduce bias into the estimates, whether analyzed by phe-
netics or cladistics.
A fundamental requirement in phylogenetic studies based
on nucleic acid or protein sequences is that the genes com-
pared are orthologous (originating from organismal cladogen-
esis) (Alvarez and Wendel, 2003). Plant rDNA consists of
thousands of copies or paralogues (derived from gene dupli-
cation events) spanning several loci, which evolve through
gene conversion, unequal crossing over, and perhaps repeat
amplification (Baldwin et al., 1995). ITS sequences have prov-
en useful in species-level phylogenetic studies of a wide range
of taxa (Alice and Campbell, 1999; Downie et al., 2000; Lia
et al., 2001, among others). Within-individual rDNA poly-
morphisms may occur when concerted evolution is not fast
enough to homogenize repeats in the face of high rates of
mutation and/or recent interspecific hybridization (Campbell
et al., 1997). Intraspecific variation concerning ITS sequences
has been detected in several plant groups (Baldwin et al., 1995
and references therein; Buckler and Holtsford, 1996a, b; Buck-
ler et al., 1997; Campbell et al., 1997; Hartmann et al., 2001;
Mayol and Rosell6, 2001). Active ITS regions have functional
constraints that can help discriminate between functional and
nonfunctional paralogous (i.e., pseudogene) sequences (Buck-
ler et al., 1997). For instance, nonfunctional ITS paralogues
could accumulate random substitutions, which will destabilize
hairpins and reduce secondary structure stability. When non-
functional paralogous ITS sequences are unknowingly includ-
ed in the phylogenetic analyses (to the exclusion of appropriate
orthologous comparisons), the resulting gene tree may con-
found organismal divergence events with a tracking of the his-
tory of gene duplication (Alvarez and Wendel, 2003). Erro-
neous assessment of orthology and paralogy will lead to phy-
logenetic incongruence. Hence, discrimination of inactive par-
alogues and pseudogenes becomes crucial.
Another critical step in molecular phylogenetic studies is
the alignment of sequence data, in which primary homology
statements are being established. However, it is typically per-
formed without considering the effects of analytical parame-
ters, such as indel and base transformation values, on the phy-
logenetic inference. In the case of multiple sequence align-
ments, different parameters may yield alternative alignments,
and these primary homology hypotheses may support distinct
topologies (Doyle and Davies, 1998). Indeed, Morrison and
Ellis (1997) showed that different multiple sequence alignment
methods were responsible for the variation encountered among
resultant phylogenetic trees topologies, and this variation was
more important than the one found when different methods of
phylogenetic reconstruction were applied to the same align-
ments. Because parameter choice is arbitrary but unavoidable
when doing algorithmic DNA sequence comparisons, its ex-
ploration becomes essential (Giribet and Wheeler, 1999; Gi-
ribet and Ribera, 2000). Herein, we used ITS sequences to
study the phylogenetic relationships of sSA taxa and to explore
the influence of applying different insertion/deletion (indel or
gap) and substitution models (sensitivity analysis sensu
Wheeler, 1995) on phylogenetic inference. Phylogenetic anal-
yses were accomplished through direct optimization of DNA
sequences (Wheeler, 1996). The sensitivity analysis is a way
to explore the data and to discern between stable relationships
(those supported throughout the parameters range) and unsta-
ble relationships (those appearing only under particular con-
ditions), allowing the formulation of more robust hypotheses
(Giribet and Ribera, 2000) than in customary phylogenetic
analyses. Direct optimization is a method that does not dis-
connect the alignment step from the tree-building step and that
generalizes phylogenetic character analysis to include indel
events (Giribet, 2001, 2003). By doing this, indels appear as
transformations (not as states) linking ancestral and descendent
nucleotide sequences (Giribet and Wheeler, 1999).
MATERIALS AND METHODS
Specimens studied-The list of sSA flex taxa employed, together with their
accession numbers, geographical origin and GenBank accession numbers of
ITS sequences obtained in this study are presented in Appendix 1 (see Sup-
plemental Data accompanying the online version of this article). Plant material
was principally obtained from the Estaci6n Experimental INTA Cerro Azul
(EEINTACA) Germplasm Bank (Misiones, Argentina). Leaf samples were
also collected from natural populations from Northern Argentina; voucher
specimens of these samples were deposited at the BACP Herbarium (Centro
de Estudios Farmacol6gicos y Botfinicos, CEFyBO, Buenos Aires, Argentina).
Furthermore, leaf samples of commercial lines of mate were obtained from
the Establecimiento Las Marias (Corrientes, Argentina). Young leaf samples
were collected from healthy plants (without evidence of fungal or insect in-
fection) and were silica-gel preserved. Additional 42 ITS sequences were
downloaded from GenBank (http://www.ncbi.nlm.nih.gov); for their geo-
graphical origin and accession numbers see Appendix 2 (Data Supplement
accompanying the online version of this article). Helwingia japonica (C. P.
Thunberg ex A. Murray) Dietrich and Helwingia chinensis Batalin were cho-
sen as the outgroup. According to several molecular studies (Cu6noud et al.,
2000; Powell et al., 2000; Savolainen et al., 2000; Albach et al., 2001; Bremer
et al., 2002), Helwingia and Ilex are closely related members of Aquifoliales.
Loeseners' (1891, 1901, 1908, 1942) original dispositions of subgeneric ranks
were used in this work (Fig. 19). The species flex microdonta, I. pseudobuxus,
and I. taubertiana, were only included in the sequence analyses because ac-
cessions of these taxa were unavailable in sufficient number for AFLP anal-
ysis. Only two sSA taxa remain to be studied, namely flex affinis and I.
chamaedryfolia, because these have not been found in the field since the
1980s (G. Giberti, Instituto de Quifmica y Metabolismo del Firmaco, IQUI-
MEFA, personal communication). Attempts to obtain good quality DNA from
herbarium/air-dried specimens were unsuccessful.
DNA extraction-Leaves were ground to a fine powder in liquid nitrogen
and then placed in a microtube. The DNeasy Plant kit (QIAGEN Inc., Valen-
cia, California, USA) was used for DNA extraction following the manufac-
turer's instructions. DNA was kept at 40C until use and then stored at-200C.
AFLP technique--The AFLP methodology was carried out using the AFLP
Analysis System I Starter Primer kit (Gibco BRL, Life Technologies, Carls-
bad, California, USA) as described in the instruction manual. Eight selective
primers were combined as follows: E + AGG, M + CTT; E + ACT, M +
CAC; E + AGC, M + CAT; and E + ACC, M + CAG. Other primers
combinations assayed were discarded because they generated profiles in which
This content downloaded from 157.92.6.34 on Tue, 18 Dec 2018 17:13:49 UTC
All use subject to https://about.jstor.org/terms
354 AMERICAN JOURNAL OF BOTANY [Vol. 92
the amplification products were too dense to allow reliable scoring, or alter-
natively, generated too few amplification products. Selective amplification
products were mixed with an equal volume of dye reagent (98% [v/v] form-
amide, 10 mM EDTA, 0.025% [w/v] bromophenol blue and 0.025% [w/v]
xylene cyanol). Five microlitres were separated by electrophoresis in a Model
S2 apparatus (Gibco BRL Sequencing System, Life Technologies) through
6% (w/v) polyacrylamide gels containing 5 M urea, in 1 X TBE buffer (89
mM Tris, 89 mM boric acid, 2 mM EDTA, pH 8). A 30-330 bp AFLP DNA
Ladder (Gibco BRL, Life Technologies) size marker was included twice or
three times in each electrophoresis run. Thus, the size of AFLP bands scored
ranged from 50 to 330 bp. Gels were stained with silver nitrate (Bassam et
al., 1991).
ITS amplification-Amplification of ITS 1, 5.8S gene, and ITS 2 was car-
ried out using primers ILEXFP (5'-AACAAGGTTTCCGTAGGTGA-3',
Powell et al., 2000) and ITS-4 (White et al., 1990). Amplifications were per-
formed in 50 [pL using 0.225 mM of each dNTP (Promega Corp., Madison,
Wisconsin, USA), 20-25 pmoles of each primer, 5 .L of Thermophilic DNA
Polymerase 10 X Buffer (Promega Corp.), 1.5 mM MgC12, 1.25 units of Taq
DNA polymerase (Promega Corp.), I fxL of genomic DNA template, and
sterile double-distilled water. Conditions for PCR amplification were those of
Vilgalys and Hester (1990). Amplifications were carried out in an Eppendorf
Mastercycler (Perkin-Elmer Corp., Foster City, California, USA). Control
samples without DNA template were included in each PCR run. PCR products
were checked by electrophoresis of 1-[LL samples in 1% agarose gels in 1 X
TAE buffer (0.04 M Tris, 0.114% glacial acetic acid, 1 mM EDTA, pH 8.0).
Quantification of PCR products was estimated by comparison with known
amounts of a DNA molecular-size marker (Lambda EcoRI/HindIII, Promega
Corp.) included in duplicate in all gels. Gels were stained with ethidium bro-
mide for 15-20 min and photographed under UV light. In some cases, best
amplification results were achieved by adding 6% bovine serum albumin
(BSA, Promega Corp.) to the PCR reaction mix. PCR products were purified
using a QIAGEN Gel Extraction kit (QIAGEN Inc.). Both strands of the
complete ITS region (ITS 1-5.8S-ITS 2) were sequenced in Perkin Elmer/
Applied Biosystems automatic sequencing units (ABI Prism 310 or 377) at
the Molecular Biology Facility of the Universidad de Alcaldi (Madrid, Spain).
Electropherograms were proofread with the software BioEdit Sequence Align-
ment Editor (Hall, 1999). Boundaries of the coding and spacer regions were
determined by comparison with published sequences of Ilex. In general, when
two or more orthologous ITS sequences were available per sSA taxon (see
Appendix 1; Supplemental Data accompanying the online version of this ar-
ticle), these were used to produce consensus sequences in BioEdit (Hall, 1999)
to speed-up computational times of phylogenetic analyses.
Data analyses-AFLP-Air-dried gels were scanned and banding patterns
were registered with the Gel-Pro 4 Analyzer Program (Media Cybernetics,
Maryland, USA). Each AFLP band, regardless of its relative intensity, was
considered as a dominant allele at a unique locus. In some cases, fragments
were scored as missing data because character states could not be interpreted
unambiguously. The data were extracted as a table as either present (1) or
absent (0). Monomorphic bands (bands present in all individuals of a species)
and diagnostic bands (monomorphic bands exclusively present in one species)
were discriminated within each species and across the entire data set. Box
plots of mean values; SE and SD were calculated and plotted with the program
STATISTICA '99 edition (Kernel release 5.5 A, StatSoft, 1999). For these
calculations, accessions with missing data in one or more primer combinations
were excluded. Pairwise genetic distance (D) between all individuals was
estimated according to the complementary value of Nei and Li's (1979) sim-
ilarity coefficient, implemented in PAUP* (Swofford, 1998): D = 1 - (2n,/1
[n; + ni]), where n,i is the number of shared fragments between two individ-
uals i and j, and n, and n, are the total number of fragments in individual i
and j, respectively. From these values, mean intra- and interspecific distance
values were obtained. Furthermore, the AFLP data set for 120 accessions
belonging to eight taxa was analyzed using maximum parsimony in PAUP*.
Initial trees in the heuristic search by tree-bisection-reconnection (TBR)
branch-swapping algorithm were built stepwise with 500 random addition
sequence replicates. Characters were equally weighted and unordered, and all
other settings were defaults. The reliability of the clades was evaluated by
bootstrap analysis (Felsenstein, 1985) in PAUP* with 1000 replications and
all other settings as described earlier. Trees were inspected with the program
TreeView (Page, 1998). The unrooted resulting strict consensus tree was root-
ed using I. paraguariensis to be consistent with ITS trees.
ITS sequence studies-We characterized and identified ITS paralogues in
sSA Ilex species by examining their G + C content, secondary structure
stabilities, and nucleotide substitution patterns and using the tree-based ap-
proach advocated by Bailey et al. (2003). To discriminate the presence of
presumed active sequences from paralogues and putative pseudogenes, we
first followed the procedure proposed by Mayol and Rosell6 (2001). There-
fore, pairwise nucleotide divergences were calculated between the 21 ITS
sequences obtained herein and the 10 sequences retrieved from GenBank that
belong to the same sSA Ilex taxa. The values were calculated using the Ki-
mura two-parameter model (Kimura, 1980) with DNAdist 3.5c program im-
plemented in BioEdit (Hall, 1999). Additionally, ITS sequences were screened
using BioEdit (Hall, 1999) for length variation, G + C content, and the pres-
ence of a structural motif in ITS 1 (i.e., GGCRY(4 to 7 n)GYGYCAAGGAA;
Liu and Shardl, 1994). This conserved motif is often associated with part of
the hairpin and loop structures that may serve as a critical signal in the en-
zymatic processing of the ribosomal RNA (Liu and Shardl, 1994). Further-
more, optimal and suboptimal folding structures, and associated free energy
values (AG) in Kcal/mol, of ITS 1 and ITS 2 were recorded. Fold predictions
were made at the Quickfold web server (Zuker, 2003) of the MFOLD pro-
gram, version 3.1. In all cases, foldings were conducted at 370C by use of a
search within 5% of the thermodynamic optimality set. Because each se-
quence could adopt more than one folding structure, maximum and minimum
AG values were recorded for each sequence, together with the number of all
possible (optimal and suboptimal foldings) secondary structures. However, no
attempts were made to study the RNA structures per se. These screenings
were extended to the remaining representatives of Ilex and outgroup.
Phylogenetic and sensitivity analyses-Parsimony was chosen as the op-
timality criterion, as implemented in the program for Direct Optimization
POY, version 3.0.11 (Gladstein and Wheeler, 1997; Wheeler and Gladstein,
2000; De Laet and Wheeler, 2003). To save computation time, as suggested
in the POY user's manual and in Giribet (2001), the sequence data set was
partitioned. The first partition comprises 257 to 265 bp, and the second
partition has ca. 340 bp. Approximately 45 positions of the 5.8S gene, rep-
resenting missing data in several sequences downloaded from the GenBank,
were excluded from the analyses. A parameter space of two analytical var-
iables was examined, i.e., gap cost and transversions/transitions (Tv/Ts) ra-
tio cost. Precisely, we used gap costs = 1, 2, 4, 8 and 10, and Tv/Ts costs
= 1 (equal weights), 2 (transversions receive twice the weight than transi-
tions) and 4. All combinations of transformation costs used were in accor-
dance with the triangle inequality (Wheeler, 1993, 1995). Each particular
tree search was performed as follows: 50 independent heuristic searches
starting from the best of five independent Wagner trees were conducted via
subtree-pruning and regrafting (SPR) followed by TBR, retaining two trees
per step, as specified in the following commands: "-norandomizeoutgroup
-buildsperreplicate 5 -replicates 50 -maxtrees 2 -slop 2 -checkslop 5".
Commands "-slop" and "-checkslop" were employed to reduce error de-
rived from the heuristic operations (Wheeler, 2002). Step matrices (i.e.,
transformation matrices) were specified in POY using the command "-mo-
lecularmatrix" with an argument for the given step matrix. Strict consensus
trees were calculated using the command: "-poystrictconsensustreefile".
An implied alignment was generated in each case by the command:
"-impliedalignment". Approximate Bremer support values (Bremer, 1988)
were calculated with a heuristic approximation procedure implemented in
POY (command: "-bremer"). Finally, strict consensus trees and single trees
were used to obtain 50% majority rule consensus tree. Implied alignments
and trees were deposited at TreeBase (S996). We specifically used nodal
stability as defined by Giribet (2003), i.e., the degree to which nodes are
affected by variation in the analytical conditions. Robustness was used here-
This content downloaded from 157.92.6.34 on Tue, 18 Dec 2018 17:13:49 UTC
All use subject to https://about.jstor.org/terms
February 2005] GOTTLIEB ET AL.-MOLECULAR ANALYSES OF SOUTH AMERICAN ILEX 355
TABLE 1. Number and relative abundance of AFLP monomorphic (m) TABLE 2. Number of AFLP monomorphic (m) and diagnostic (d)
and diagnostic (d) bands obtained per primer combination for se- bands recognized per Ilex species.
lected Ilex taxa.
d/m X
Total m/total d/m X Species m d 100 (%)
Primer combination bands m x 100 (%) d 100 (%)
Ilex argentina 58 19 33
E + AGG/M + CTT 224 28 12.5 12 43 L brasiliensis 60 13 22
E + ACT/M + CCA 210 57 27 30 53 I. brevicuspis 40 19 47.5
E + AGC/M + CAT 193 42 22 21 50 L dumosa 13 8 61.5
E + ACC/M + CAG 209 58 28 31 53 L integerrima 74 13 17.5
Total 836 185 22 94 51 L paraguariensis 38 17 44.7
L theezans 69 5 7
in as a measure of stability of nodes to fluctuation in parameters values
across an analytical landscape (Ogden and Whiting, 2003).
Two sets of analyses were performed. First, we undertook a phylogenetic
analysis of all ITS sequences, including those tentatively determined as par-
alogues and pseudogenes. To explore the influence of alignment parameters
on the relative placement of these sequences, five different combinations of
the two variables were employed. In particular, these combinations were: 111
(i.e., gap cost = 1, Tv = 1 and Ts 1), 241, 421, 811 and 1021. Then, we
performed a broader sensitivity analysis using only presumably active ITS
sequences under 15 analytical conditions (namely, 111, 121, 141, 211, 221,
241, 411, 421, 441, 811, 821, 841, 1011, 1021 and 1041). Each of the 15
tree searches was performed as indicated earlier.
Combined analysis-The combined analysis involved AFLP data and ITS
sequences. The sensitivity analysis was performed in POY as indicated pre-
viously, with the exception that a third file consisting of AFLP data per species
was included in all calculations. The new data matrix for species was extracted
from the original AFLP matrix for individuals and consisted of eight taxa and
578 characters. A band was considered present in a taxon (1) if it was re-
corded in more than 80% of the individuals, whereas it was considered absent
in a taxon (0) if it was recorded in less than 20% of the individuals. Otherwise,
bands were considered as polymorphic (0,1). The weights of AFLP changes
were set so that nucleotide characters were never upweighted relative to AFLP
data. The computer program POY allows nonsequence character changes to
be simultaneously optimized with molecular data on multiple trees.
RESULTS
AFLP analysis-The four primer combinations generated
836 scoreable bands, of which 22% were monomorphic (Table
1). All 120 individuals tested shared a single constant band.
Primer combinations E + ACT/M + CCA and E + ACC/M
+ CAG yielded a nearly identical number of monomorphic
and diagnostic bands. Among the seven species included in
this study, 94 diagnostic bands were distinguished, which ac-
counted for more than 50% of the monomorphic fragments
recorded. The proportion of monomorphic and diagnostic
bands detected varied among the species (Table 2). Ilex inte-
gerrima had the highest number of monomorphic bands,
whereas the highest number of diagnostic bands was obtained
for I. argentina and I. brevicuspis. In the box plot in Fig. 1,
I. argentina gave the highest mean number of AFLP bands
(mean = 175, SD = 29.9), whereas the lowest number of
bands was found in I. brevicuspis (mean = 124, SD = 6.1).
Intra- and interspecific distance values derived from AFLP
data are presented in Table 3. Excluding I. dumosa, for which
the highest intraspecific distance value was obtained (0.1223,
SD = 0.0408), remaining taxa had similar values. When in-
terspecific mean distances were compared, I. integerrima and
I. theezans appeared to be the least distant pair of taxa (0.1014,
Table 3).
220
2 00 .. ..... .. ....... . ..... ........... ............
180.
0
CO
o
F--
120 . . . . . . Mean+SD
Mean-SD
m Mean+SE
100 Mean-SE
I. argentina I. brevicuspis I integerrima I theezans
I. brasiliensis I. dumosa I. paraguariensis O Mean
Ilex species
Fig. 1. Comparison of mean number of AFLP bands obtained for selected Ilex species.
This content downloaded from 157.92.6.34 on Tue, 18 Dec 2018 17:13:49 UTC
All use subject to https://about.jstor.org/terms