Systematic Botany (2005), 30(3): pp. 503–519
q Copyright 2005 by the American Society of Plant Taxonomists
Phylogeny and Systematics of Aponogeton (Aponogetonaceae):
The Australian Species
DONALD H. LES,1,3 MICHAEL L. MOODY,1 and SURREY W. L. JACOBS2
Department of Ecology and Evolutionary Biology, University of Connecticut,
Storrs, Connecticut 06269-3043 USA;
2
Royal Botanic Gardens, Sydney, New South Wales, 2000, Australia
3
Author for correspondence (les@uconn.edu)
1
Communicating Editor: Alan W. Meerow
ABSTRACT. Aponogeton is an important genus whose species are cultivated widely as ornamental aquatic plants. Although
a fairly recent monograph has been published, the genus remains poorly studied systematically. We conducted a phylogenetic
survey of Aponogeton that focused on relationships among the nine native Australian species as well as their relationship to
other members of the genus. Our analyses included a phylogenetic assessment of morphological characters and molecular
data obtained both from chloroplast (trnK 59 intron, matK) and nuclear DNA (nrITS) loci. Molecular data provided evidence
of hybridization and polyploidy as well as an informative overview of interspecific relationships in the genus. Two potentially
new Australian species also were identified by the molecular data. Combined molecular data produced a well-resolved
cladogram that enabled us to evaluate previous phylogenetic hypotheses based on non-explicit methods as well as the
soundness of the existing classification of the genus. We conclude that Aponogetonaceae originated in Australia and subsequently radiated into Africa, Madagascar, and Asia, from which a secondary Australian diversification occurred resulting
in a biphyletic origin of the native Australian species. A pattern of morphological distinctiveness coupled with low molecular
divergence indicates relatively recent and rapid speciation of Aponogeton in Australia. Our results also demonstrate that in
this group, morphological data are extremely unreliable taxonomically due to their extensive homoplasy. The phylogenetic
relationships elucidated by this study provide evidence to support the establishment of two additional sections, Flavida and
Viridis, which are described.
The freshwater monocotyledon genus Aponogeton
L.f. (Aponogetonaceae) comprises approximately 50
species of obligate aquatic plants that are distributed
mainly in tropical or subtropical regions of the Old
World (Cook 1996; Hellquist and Jacobs 1998; Bruggen
1985). Analysis of rbcL sequence data indicates that the
monotypic Aponogetonaceae occupy a relatively basal
position near the families Juncaginaceae and Scheuchzeriaceae in one of two major clades that subdivide
subclass Alismatidae (Les et al. 1997).
Although the genus is not a dominant element of
any flora, Aponogeton is important economically as it
long has been regarded as a valuable source of species
suitable for use as aquarium plants. In particular, the
unusual fenestrate-leaved A. madagascariensis (Mirb.)
H. Bruggen (known as ‘‘Madagascar lace plant’’) has
been in cultivation since 1855 (Tricker 1897) and continues to rank among the most popular ornamental
freshwater aquarium plants. Aponogeton distachyos L.f.
(‘‘water hawthorne’’) has been cultivated for more than
two centuries, but mainly as an outdoor pond or watergarden ornamental (Bruggen 1985). It has been introduced to Victoria, Australia where it is regarded as an
invasive weed (Gunasekera 2003). Another 15 species
from Asia, Australia, and Madagascar are grown either widely or at least to a minor degree as ornamental
aquarium plants (Kasselmann 1995). Aponogeton distachyos recently has become a popular food plant in
South Africa where it is now cultivated intensively
(Gunasekera 2003; Pemberton 2000). Several other species (A. capuronii H. Bruggen, A. crispus Thunb., A.
elongatus F. Muell. ex Benth., A. euryspermus Hellq. & S.
W. L. Jacobs, A. madagascariensis, A. natans (L.) Engl. &
K. Krause, A. queenslandicus H. Bruggen, A. vanbruggenii Hellq. & S. W. L. Jacobs) have edible tubers, but
are important only locally as food plants (Bruggen
1985; Cowie et al. 2000).
There are no Aponogetonaceae native to the New
World. Over half of the species occur on the African
continent (17 species) and in Madagascar (11 species)
(Bruggen 1985; Kasselmann 1995). Ten Aponogeton species grow in India and Southeast Asia (Bruggen 1985;
Cook 1996) with a subset of five in Sri Lanka (Thabrew
and Thabrew 1983). Two species (both endemic) are
known from New Guinea (Leach and Osborne 1985)
and the remaining species are Australian. The South
African A. distachyos has been introduced to Australia
(New South Wales, Victoria, South Australia), Europe
(England, France), New Zealand, North America (California), and South America (Argentina, Peru).
Aponogeton is poorly understood systematically. In
Aponogeton, as in many aquatic plants, taxonomic study
has been hampered by the similar, often convergent
vegetative morphology of most species (sterile plants
are notoriously difficult to identify), extensive phenotypic plasticity (Bruggen 1985; Hellquist and Jacobs
1998), and highly simplified reproductive structures.
Consequently, there are few morphological characters
that are useful for making taxonomic distinctions or
that might serve as reliable phylogenetic markers
(Bruggen 1985).
In the early 19th century, Aponogeton was subdivided
503
504
SYSTEMATIC BOTANY
into two genera with Ouvirandra segregated to include
plants with caducous tepals and large plumules. However, the latter genus was ill-defined and could not be
maintained as circumscribed (Bruggen 1985). Several
classifications of Aponogeton have emphasized the taxonomic importance of seed-coat (testa) number and inflorescence morphology but have treated infrageneric
groups only informally. The most recent formal classification of Aponogeton was given by Camus (1923)
who divided the genus into two sections: Aponogeton
(flowers omnilateral) and Pleuranthus (flowers secund).
Each section was divided further into two subsections
that separated plants with simple versus forked inflorescences. The fact that this classification subdivides a
group of nearly 50 species by incorporating only two
characters attests to the difficulty in finding characters
appropriate for systematic applications.
With admitted reluctance, Bruggen (1985) followed
the classification of Camus (1923) in his monograph of
42 Aponogeton species even though he was not confident that the classification succeeded in depicting natural groups. Bruggen also believed that Aponogeton
was unsuitable for cladistic analysis because of what
he perceived as many reticulate relationships attributable to polyploidy. Nevertheless, within the broader
classification, Bruggen identified several groups of species that he believed to be closely related. However,
these proposed interspecific relationships essentially
remain untested.
Thanikaimoni (1985, p. 11) evaluated phylogenetic
relationships in Aponogeton by presenting a ‘‘scheme
depicting the morphological diversifications’’ that he
believed to indicate ‘‘evolutionary trends’’ in the genus
(Fig. 1). Technically, this diagram is not a cladogram,
but represents a phylogenetic hypothesis based upon
Thanikaimoni’s perception of interspecific relationships as indicated by transitional morphological series.
By these phylogenetic relationships, the sections and
subsections recognized by Camus (1923) all represent
polyphyletic groups (Fig. 1) with the exception of
Aponogeton sect. Pleuranthus subsect. Monostachys,
which is monotypic. Thanikaimoni’s phylogenetic perspective placed the Malagasy A. longiplumulosus H.
Bruggen as basal in the genus, which led him to hypothesize that Aponogetonaceae originated in Madagascar. However, the soundness of the existing classification and these phylogenetic hypotheses cannot be
ascertained until more empirical analyses have been
undertaken.
Studies made during the past 35 years have provided evidence that Australia is an important center of
diversity for Aponogeton. Although early workers recognized Aponogeton elongatus as the only native Australian species (Krause and Engler 1906), Bruggen’s (1969)
revision of Australian Aponogeton added three new
species to yield a total of four native (A. bullosus H.
[Volume 30
Bruggen, A. elongatus, A. hexatepalus H. Bruggen, A.
queenslandicus) and one nonindigenous species (A. distachyos) in the flora. Among these, A. hexatepalus was
so distinctive by its forked inflorescence and flowers
with six tepals that Bruggen (1969) doubted whether
it shared a close relationship with any living Aponogeton species. Aston (1973) recognized the same five
species in Australia, but also commented on an unnamed, ‘‘proliferous’’ taxon (i.e., producing vegetative
plantlets in lieu of flowers in the inflorescence) in
northern Queensland. Most recently, Hellquist and Jacobs (1998) have reevaluated the Australian Aponogetons thoroughly and described six new taxa including five new species: A. kimberleyensis Hellq. & S. W. L.
Jacobs, A. euryspermus, A. vanbruggenii, A. lancesmithii
Hellq. & S. W. L. Jacobs, and A. proliferus Hellq. & S.
W. L. Jacobs (the latter corresponding to Aston’s unnamed proliferous species).
In this study, we investigate relationships among
Australian Aponogeton species in detail using an explicit, phylogenetic approach. Our work represents the
first empirical phylogenetic analysis to be undertaken
for any portion of Aponogeton. It provides the first test
of various systematic hypotheses, including a phylogenetic appraisal of the classification developed by Camus (1923). We evaluate the use of several types of
characters for phylogenetic analysis of Aponogeton, including both morphological features and molecular
data.
In a taxonomically difficult genus such as Aponogeton, where reliable morphological characters are scarce,
the analysis of molecular markers provides one alternative means of obtaining a relatively large number of
characters suitable for phylogenetic analysis. However,
even though the genetic basis of directly sequenced
DNA regions might be perceived as unambiguous, the
homology of molecular data also is subject to misinterpretations due to parallel substitutions and paralogous loci especially in polyploid species where gene
duplications are prevalent (Page and Holmes 1998).
This observation is pertinent because Aponogeton is
highly polyploid. Reported counts indicate a chromosomal base number for the genus of x 5 8, and an
assortment of chromosome numbers ranging from 2n
516 to 2n 5 100 occurs among various species (Arends 1985). Indeed, A. elongatus, the only Australian
Aponogeton for which a chromosome number has been
reported, is polyploid (2n 5 40).
Although Bruggen (1985) did not believe that Aponogeton formed natural hybrids, he observed several instances of apomixis (agamospermy), which often is associated with hybridization and polyploidy (Grant
1981); thus the potential for hybridization certainly exists in the genus. To safeguard against misleading results that might arise from analyses involving reticulate
relationships such as those associated with hybridiza-
2005]
LES ET AL.: AUSTRALIAN APONOGETON
505
FIG. 1. Interspecific relationships in Aponogeton as indicated by morphological trends [redrawn in tree form from the diagram
in Thanikaimoni (1985)]. Epithets of the species included in the present study are highlighted in bold. Native geographical
distributions of the species are shown at right.
506
SYSTEMATIC BOTANY
[Volume 30
TABLE 1. List of taxa investigated, distributions, vouchers, specimens and Genbank accession numbers. Multiple accessions of a species
are indicated by numbers in parentheses. The taxon A. ‘‘indet.’’ comprised vegetative material that could not be identified to species
confidently. Two taxa (identified as A. ‘‘species 1’’ and A. ‘‘species 2’’) emerged as new species that remain unnamed at present.
Cycnogeton procerum and Tetroncium magellanicum constituted the outgroup used in all molecular analyses. Specimens designed by ‘‘cult.’’
were obtained from sources of material grown in cultivation (see text). Citations with mutiple numbers (e.g., Jacobs 8572 & Les 595)
reflect different personal numbers assigned to a single specimen gathered jointly by the collectors. The order of Genbank accession
numbers for each taxon is: matK, trnK 59 intron, nrITS (bracketed numbers [] 5 multiple cloned sequences recovered; NA 5 sequence
not determined).
Aponogeton bullosus H. Bruggen, Australia (Queensland), Jacobs 8572 & Les 595 (CONN, NSW); AY926279, AY926344, AY926318;
A. crispus Thumb., SE Asia (cult.), Jacobs 8537 & Les 563 (CONN, NSW); AY926263, AY926328, AY926288; A. distachyos L.f., S
Africa (cult.), Les s.n. (CONN); AY926281, AY926346, AY926320; A. elongatus F.Muell. ex Bentham, (1), Australia (Queensland),
Jacobs 8525 & Les 551 (CONN, NSW); AY926266, AY926331, AY926296; A. elongatus (2), Australia (Queensland; cult.), Les s.n.
(CONN); AY926267, AY926332, AY926297; A. elongatus (3), Australia (New South Wales), Jacobs 9075 (NSW); AY926266, AY926331,
AY926294; A. elongatus (4), Australia (New South Wales), Jacobs 9074 (NSW); AY926266, AY926331, AY926295; A. euryspermus
Hellq. & S.W.L.Jacobs (1), Australia (N Territory; cult.), Jacobs 8532 & Les 558 (CONN, NSW); AY926273, AY926338, AY926308;
A. euryspermus (2), Australia (W Australia), Jacobs 8839 (NSW); AY926275, AY926340, AY926310; A. hexatepalus H.Bruggen,
Australia (W Australia), Sainty NSW434337 (NSW); AY926282, AY926347, AY926321; A. ‘‘indet.’’, Australia (Queensland), Jacobs
8571 & Les 594 (CONN, NSW); AY926278, AY926343, AY926317; A. kimberleyensis Hellq. & S.W.L.Jacobs, Australia (W Australia),
Jacobs 8831 (NSW); AY926274, AY926339, AY926309; A. lancesmithii Hellq. & S.W.L.Jacobs, Australia (Queensland), Jacobs 8567 &
Les 590 (CONN, NSW); AY926277, AY926342, AY926316; A. longiplumulosus H.Bruggen, Madagascar (cult.), Jacobs 8534 & Les
560 (CONN, NSW); AY926260, AY926325, AY926284; A. madagascariensis (Mirbel) H.Bruggen (1), Madagascar (cult.), Jacobs 8535
& Les 561 (CONN, NSW); AY926261, AY926325, AY926286; A. madagascariensis (2), Madagascar (cult.), Jacobs 8536 & Les 562
(CONN, NSW); AY926261, AY926325, AY926285; A. proliferus Hellq. & S.W.L.Jacobs, Australia (Queensland; cult.), Jacobs 8523
& Les 549 (CONN, NSW); AY926276, AY926341, AY926315; A. queenslandicus H.Bruggen (1), Australia (Queensland; cult.), Jacobs
8524 & Les 550 (CONN, NSW); AY926265, AY926330, AY926293; A. queenslandicus (2), Australia (Queensland; cult.), Jacobs 8526
& Les 552 (CONN, NSW); AY926265, AY926330, AY926289 [2.1], AY926290 [2.2], AY926292 [2.3]; A. queenslandicus (3), Australia
(Queensland; cult.), Jacobs 8541 & Les 567 (CONN, NSW); AY926265, AY926330, AY926298; A. rigidifolius H.Bruggen (1), Sri
Lanka (cult.), Jacobs 8529 & Les 555 (CONN, NSW); AY926262, AY926327, AY926287; A. rigidifolius (2), Sri Lanka (cult.), Jacobs
8530 & Les 556 (CONN, NSW); AY926262, AY926327, AY926287; A. robinsonii A. Camus, Vietnam (cult.), Jacobs 8806 (NSW);
AY926280, AY926345, AY926319; A. stachyosporus de Wit, India (cult.), Jacobs 8538 & Les 564 (CONN, NSW); AY926272, AY926337,
AY926303 [1.1], AY926304 [1.2], AY926305 [1.3], AY926306 [1.4]; A. ‘‘ulvaceus’’ Baker (1), Madagascar (cult.), Jacobs 8546 & Les
572 (CONN, NSW); AY926259, AY926324, AY926307 [1.1], AY926312 [1.2], AY926313 [1.3]; A. ulvaceus (2), Madagascar (cult.),
Jacobs 8543 & Les 569 (CONN, NSW); AY926259, AY926324, AY926283; A. undulatus Roxb. India (cult.), Jacobs 8539 & Les 565
(CONN, NSW); AY926271, AY926336, AY926302; A. vanbruggenii Hellq. & S.W.L.Jacobs (1), Australia (N Territory; cult.), Jacobs
8542 & Les 568 (CONN, NSW); AY926269, AY926334, AY926300 [1.1], AY926311 [1.2], AY926314 [1.3]; A. vanbruggenii (2),
Australia (N Territory; cult.), Jacobs 8533 & Les 559 (CONN, NSW); AY926268, AY926333, AY926299; A. ‘‘species 1’’, Australia
(Queensland; cult.), Jacobs 8528 & Les 554 (CONN, NSW); AY926264, AY926329, AY926291; A. ‘‘species 2’’, Australia (N Territory),
Jacobs 8801 (NSW); AY926270, AY926335, AY926301; Cycnogeton procerum Buchenau, Australia, Beesley 449 (CBG); NA, AY926349,
AY926323; Tetroncium magellanicum Willd., Chile, Alvarez s.n. (CONN); NA, AY926348, AY926322.
tion and polyploidy, we included both maternally-inherited (trnK 59 intron; matK) and biparentallyinherited (nrITS) molecular markers in our study. In addition, we employed a molecular cloning strategy as a
means of evaluating potentially paralogous loci. Using
this approach we were able to make a preliminary assessment of the existing classification, achieve a reasonable phylogenetic assessment of Australian Aponogeton,
and elucidate further details on the relationships of the
Australian species to other species in the genus.
MATERIALS
AND
METHODS
Taxon Sampling. Thirty-three accessions of 23 taxa (21 Aponogetonaceae; 2 Juncaginaceae) were evaluated (Table 1). This sample included all nine species currently recognized as native to Australia, one African species (introduced to Australia), three Malagasy species and five Asian species. We examined multiple accessions for seven species (Table 1). Identification of all cultivated
material was verified by the authors. The cultivated accessions of
Australian species were collected originally from sites where species determinations had already been made (and prior vouchers
collected) by coauthor SWLJ.
Morphological Analyses. To serve as an initial hypothesis of
phylogenetic relationships in Aponogeton, we reconstructed, in tree
format, the diagram presented by Thanikaimoni (1985) that purportedly shows interspecific relationships as inferred from the pattern of morphological diversification that he elucidated in the genus. For empirical analysis, a total of 19 morphological characters
(5 vegetative, 14 reproductive) was scored for 17 Aponogeton species (Tables 2, 3). Following Bruggen (1985), we did not distinguish
A. stachyosporus de Wit from A. undulatus Roxb. in the morphological analysis as it would have been scored with identical character
states. Morphological characters were selected from those emphasized taxonomically by Bruggen (1969, 1985) and Hellquist and
Jacobs (1998). We excluded characters that would have been autapomorphic (i.e., those varying in only one of the species). Several
other characters (submersed leaf margin undulation; seed number
and length; seed coat adherence; pericarp texture) were included
in initial analyses, but were excluded when it became apparent
that their high degree of homoplasy resulted in a nearly complete
loss of resolution in resulting trees.
Morphological data were analyzed phylogenetically using unweighted maximum parsimony as implemented by the program
PAUP* (Swofford 1998). Searches were conducted using the
branch-and-bound algorithm (furthest addition sequence; MulTrees options) with all character states treated as unordered. We
were unable to include members of the outgroup (Juncaginaceae,
see below) in the morphological analyses due to our inability to
score homologous states confidently between the two families. We
relied on results from the molecular analyses (below), which clear-
2005]
507
LES ET AL.: AUSTRALIAN APONOGETON
TABLE 2. Morphological characters and character states used in phylogenetic analysis of Aponogeton (compiled from Bruggen, 1969;
1985; Hellquist & Jacobs, 1998).
Vegetative: 1. habit (0 5 floating leaves only; 1 5 leaves floating & submersed; 2 5 leaves all submersed); 2. maximum tuber
length (0 5 $4 cm; 1 5 ,3 cm); 3. submersed leaf blade surface (0 5 flat; 1 5 bullate); 4. maximum submersed leaf
width (0 5 .2.5 cm; 1 5 #2.5 cm); 5. floating leaf base morphology (0 5 never cordate; 1 5 commonly to rarely cordate).
Reproductive: 6. inflorescence habit (0 5 emergent; 1 5 emergent or floating; 2 5 non-emergent); 7. inflorescence morphology (0 5 branched; 1 5 unforked/rarely branched; 2 5 unforked); 8. peduncle (0 5 not proliferous; 1 5 proliferous); 9.
peduncle diameter (0 5 equal to inflorescence rachis; 1 5 . inflorescence rachis); 10. spathe duration (0 5 persistent; 1 5
caducous); 11. maximum spathe length (0 5 long, . 1.5 cm; 1 5 short, #1.5 cm); 12. flower arrangement (0 5 all around
axis; 1 5 secund in 2 rows); 13. flower spacing (0 5 loose; 1 5 dense; 2 5 dense or loose); 14. tepal number (0 5 6; 1 5
2; 2 5 1); 15. tepal color (0 5 white/pink; 1 5 green; 2 5 yellow); 16. number of tepal nerves (0 5 13; 1 5 one); 17.
stamen number (0 5 8-16; 1 5 6; 2 5 4); 18. testa number (0 5 1; 1 5 2); 19. plumule (0 5 absent; 1 5 present).
genera of Juncaginaceae (Cycnogeton, Tetroncium) to function as
outgroups in accordance with the rbcL survey by Les et al. (1997),
which showed that family to be closely related to Aponogetonaceae.
Routine procedures as described in Moody and Les (2002) were
followed for the extraction, amplification, and automated sequencing of ITS (ITS-1 and ITS-2 regions including the 5.8s rRNA gene)
and cpDNA (trnK 59 intron with an adjacent 59 portion of the matK
coding region). In addition it was necessary to develop two new
sequencing primers: ApotrnKR (59ATAATTTTGTTGATACAT)
and Apo340F (59ACGAGCTTATGTTCTTA). Nevertheless, we
were unable to obtain complete trnK 59 intron/matK sequences for
Cycnogeton and Tetroncium. Sequencing was performed using an
ABI 3100 automated sequencer. All sequences used in our analyses
were newly generated and have been deposited in the GenBank
database (Table 1).
Four Australian accessions showed numerous polymorphisms in
their ITS sequence chromatograms, which indicated that they comprised mixed pools of similarly sized ITS fragments. To isolate
these sequence variants we subcloned the polymorphic PCR amplification products into plasmids using TOPO TA cloning protocol (Invitrogen, Carlsbad, CA) as described in Les et al. (2004). All
monomorphic sequences resulting from the cloned PCR products
initially were added to the analyses as separate OTUs identified
by their specimen of origin. We also observed several polymorphisms in the sequences derived from three Malagasy species (A.
longiplumulosus, A. madagascariensis, A. ulvaceus Baker). Because
subsequent phylogenetic analyses indicated that all sequences occurred within a single clade, and because this group was not our
ly indicated Aponogeton hexatepalus as sister to the rest of the genus,
and used this species for ‘‘ingroup’’ rooting of the trees. Results
were depicted by retrieving the strict consensus tree to which
bootstrap values were added (1,000 replicates; same search options
as described previously) to indicate the degree of internal support
for each resolved branch. Missing or inapplicable data constituted
4.3% of morphological data cells and were treated as missing in
all analyses. Character state distributions were examined for each
morphological character using both ACCTRAN and DELTRAN
optimizations on the tree derived from combined molecular data
(see below).
Molecular Analyses. Specimens for molecular analysis were
collected either in the field (12 accessions) or from material grown
in cultivation (21 accessions). A large number of specimens (18
accessions) was obtained from Lance Smith (Kelso, Queensland,
Australia) an aquatic plant propagator who maintains and preserves many Aponogeton species in pond culture. Voucher specimens were prepared for all specimens examined (Table1). Our selection of material provided for the analysis of all previously
known Australian taxa (nine species) and other members of the
genus to yield taxonomic coverage of both sections and three of
four subsections as defined by Camus (1923). The monotypic subsection Monostachys (A. vallisnerioides Baker) was not included. We
supplemented the material with multiple accessions for several
species that possessed unusual morphologies (e.g., ‘‘coarseleaved’’ and ‘‘fine-leaved’’ variants of A. madagascariensis, obtuse
and acute leaf apex variants of A. rigidifolius H. Bruggen and several plants of Australian origin that we could not identify to species confidently using morphological characters). We included two
TABLE 3. Matrix of morphological character states (from Table 2) used in a phylogenetic analysis of Aponogeton. (—) 5 data not
applicable; (?) 5 data missing.
Character number
A.
A.
A.
A.
A.
A.
A.
A.
A.
A.
A.
A.
A.
A.
A.
A.
A.
bullosus
crispus
distachyos
elongatus
euryspermus
hexatepalus
kimberleyensis
lancesmithii
longiplumulosus
madagascariensis
proliferus
queenslandicus
rigidifolius
robinsonii
ulvaceus
undulatus
vanbruggenii
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
2
1
0
1
1
0
2
2
2
2
2
1
2
1
2
1
1
1
0
0
0
1
0
1
0
1
1
1
0
0
1
1
1
0
1
0
—
0
0
—
0
1
0
0
0
0
0
0
0
0
0
1
0
—
0
1
—
1
0
0
0
1
0
0
0
0
0
0
—
1
0
1
0
0
—
—
—
—
—
1
—
1
—
1
1
2
0
0
1
1
0
0
1
0
0
0
1
0
0
0
0
1
2
2
0
2
1
0
2
1
0
0
1
2
2
0
0
2
2
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
1
0
0
1
1
1
1
0
1
1
1
1
1
1
0
1
1
1
1
0
1
1
1
1
0
1
1
1
1
0
1
1
1
1
0
1
0
0
0
1
1
0
0
1
0
0
1
1
0
0
1
0
1
0
0
1
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
2
0
0
1
0
0
1
1
1
0
1
1
0
1
0
1
1
1
2
1
1
0
1
1
1
1
1
1
1
1
1
1
1
2
0
0
2
2
1
2
2
0
0
2
2
0
0
0
0
2
1
1
0
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
0
1
1
1
1
1
1
1
2
1
1
1
1
1
1
0
0
0
0
1
0
0
1
0
0
?
1
0
0
1
0
1
0
1
1
1
0
1
0
0
1
1
?
0
1
1
0
1
0
508
SYSTEMATIC BOTANY
main group of interest, we did not subclone that material for further clarification. Cloning was unnecessary for the cpDNA data,
which yielded monomorphic sequences for all species surveyed.
Sequences were aligned manually and analyzed for polymorphisms and/or variable sites using Sequencher (Gene Code Corp.)
and MacClade 4 (Maddison and Maddison 2000). Phylogenetic
analyses of molecular data were performed under maximum parsimony using PAUP* (Swofford 1998) (heuristic search; random
taxon addition; TBR; characters unordered and weighted equally).
Indels were treated as missing data, but five gaps in the trnK 59
intron were included in the analysis as additional binary-coded
characters (presence/absence of gap). The degree of internal support for recovered clades was indicated by the results of bootstrap
values obtained from 500 replicates (same parameters as described
for parsimony analysis). All results of analyses generating multiple, equally parsimonious trees were evaluated using strict consensus trees. The consensus trees were output as tree files to facilitate the representation of relative branch lengths. The nuclear
(ITS) and cpDNA (trnK 59 intron/matK) data initially were partitioned to enable their separate analysis. With these preliminary
analyses yielding similar results, a combined analysis of all molecular data was carried out.
Molecular data were not combined with morphological data, because the latter could not be scored fully for the specimens used
for molecular analyses, many having been collected in vegetative
condition. However, ‘‘species level’’ morphological character state
distributions were estimated by adding and mapping the morphological data to the combined molecular tree. This analysis was
accomplished by providing the identical morphological character
states for each specimen identified as conspecific in the molecular
analysis.
Sequence homology for those OTUs characterized by multiple
ITS alleles was evaluated by examining the distribution of clones
on the cladogram resulting from phylogenetic analysis of the ITS
data. In the case of A. queenslandicus and A. vanbruggenii, homologous alleles were determined by matching them to sequences obtained from monomorphic conspecific accessions. Matches of any
other alleles to other taxa were interpreted as evidence of hybridization and they, along with other divergent alleles, were removed
prior to performing the combined molecular data analysis. For A.
stachyosporus, which showed several paralogous ITS alleles, we
identified as homologs two identical clones that grouped with A.
undulatus, a species with which it has been merged in past taxonomic treatments. The other clones formed an isolated cluster that
did not associate closely with any species in the analysis and likely
represent paralogous polyploid duplications (see discussion). We
could not determine a homolog for one accession yielding several
ITS alleles that did not associate with any taxon. We designated
this accession as ‘‘A. ulvaceus’’ because its morphology resembled
that species but also was uncharacteristic in several respects.
These, as well as other anomalous sequences, were excluded from
the combined molecular analysis.
All data used in phylogenetic analyses have been submitted to
the TreeBASE database (study accession number: S1242; matrix
accession number: M2166).
RESULTS
Morphological Analyses. Maximum parsimony
analysis of our morphological data set recovered six
equal-length trees (48 steps) characterized by fairly
high homoplasy (CI 5 0.54, CI(exc) 5 0.49, RI 5 0.68).
The strict consensus tree (Fig. 2) was poorly-resolved
and characterized by low internal support (16–48%
bootstrap values). Twelve of the characters (evaluated
on the six maximum parsimony tree variants) had a
consistency index (CI) less than or equal to 0.50. Higher consistency (0.67–1.00) was observed for seven characters (#1, 6, 13–17) (Table 2). The best-supported clade
[Volume 30
separated A. hexatepalus and A. distachyos from all other
species. Other Australian species (excluding A. hexatepalus) formed two distinct subclades, but resolution
was inadequate to establish their monophyly as a single clade. Although these results were supported only
weakly, they agreed in some respects with Thanikaimoni (1985) who hypothesized the lack of a close relationship between A. hexatepalus and other Australian
species. Although poorly supported, the monophyletic
association of Malagasy species (Fig. 2) conflicted with
Thanikaimoni’s scheme, which showed all three species as distantly related (Fig. 1). Overall, the topology
of the morphological cladogram showed higher compatibility than Thanikaimoni’s hypothesis with respect
to both the Camus classification and with geographical
regions, although the low level of resolution rendered
these assessments equivocal.
Molecular Analyses. Alignment of ITS data provided 930 aligned nucleotide sites for phylogenetic
analysis (Fig. 3). Of these, 532 sites were constant and
214 were parsimony-informative. Under maximum
parsimony we recovered 719 minimal-length trees (715
steps) characterized by moderate homoplasy (CI 5
0.78; CI(exc) 5 0.68; RI 5 0.82). Resolved nodes were
relatively well-supported (bootstrap values 5 59–
100%) with 14 of the nodes (56%) supported above
90%. The strict consensus tree (rooted by the two Juncaginaceae sequences) placed A. hexatepalus as sister to
the rest of the genus (64% bootstrap support), succeeded in position by a metaphyletic group comprising
the African A. distachyos and Asian A. robinsonii A. Camus (Fig. 3). The Malagasy species were resolved as
monophyletic (bootstrap 5 100%) as were the remaining Asian (bootstrap 5 94%) and Australian species
(bootstrap 5 99%).
Because minor sequence variation occurs regularly
in cloned DNA (Les et al. 2004), we identified as homologous those alleles differing only by a few base
pairs from other accessions. The accession A. queenslandicus (2) yielded multiple cloned ITS alleles, including both an A. queenslandicus homolog and an A. rigidifolius homolog. This accession also produced a substantially divergent sequence clone that differed by 24
steps from the next closest sequence (A. crispus). These
results indicate that the A. queenslandicus (2) accession
represents a hybrid involving A. queenslandicus and A.
rigidifolius whereas the divergent sequence probably
indicates a paralog resulting from a polyploid duplication. Similarly, the accession A. vanbruggenii (1)
yielded one cloned homolog [identical to A. vanbruggenii (2)], one homolog that allied with sequences of A.
bullosus, A. lancesmithii, and A. proliferus (all identical),
and a divergent sequence that differed from the nearest homolog by 14 steps, again probably a paralog due
to polyploidy. Thus it appears that the accession A.
vanbruggenii (1) represents a hybrid involving A. van-
2005]
LES ET AL.: AUSTRALIAN APONOGETON
509
FIG. 2. Strict consensus of six maximum parsimony trees depicting the phylogenetic distribution of states for 19 morphological characters (Tables 2, 3). Branch lengths (tree steps) are indicated above nodes; bootstrap support (%) is indicated beneath
nodes. Bracketed abbreviations after species names indicate their assignment to the classification proposed by Camus (1923):
AA 5 sect. Aponogeton subsect. Aponogeton; PO 5 sect. Aponogeton subsect. Polystachys; PL 5 sect. Pleuranthus subsect. Pleuranthus;
(?) designates newly described species not yet classified. Geographical distributions (for species grouped in boxes) are abbreviated as AF (continental Africa), AS (Asia), AU (Australia). Ingroup rooting using A. hexatepalus was performed in lieu of an
outgroup (see text).
bruggenii as one parent and a member of the ‘‘A. bullosus, A. lancesmithii, A. proliferus clade’’ as the other.
The cloned accession identified as ‘‘A. ulvaceus’’ yielded three divergent sequences that occurred within the
Australian clade, but differed from the most similar
sequences (A. elongatus) by 13–15 steps. No sequence
clones were recovered that appeared to be homologous
to any surveyed species, indicating probable hybridization with one of the Australian species. The relatively high degree of divergence of these ITS copies
could indicate paralogous polyploid loci or chimeric
sequences (produced subsequent to hybridization or as
altered cloning artifacts) and our inability to recover
alleles identical to either parent involved in the cross.
The cpDNA of this accession (see below) matched that
of A. rigidifolius, indicating that this unusual accession
was not A. ulvaceus but a hybrid involving A. rigidifolius. Aponogeton stachyosporus produced two cloned
identical homologs (differing from A. undulatus by seven steps) and two considerably divergent cloned sequences (presumably paralogs) differing from the presumed homologs by 60–70 steps and from each other
by 13 steps. All cloned sequences from A. stachyosporus
occurred within the well-supported Asian clade. There
was no evidence of hybridization in this species, with
the divergent sequences likely indicating paralogous
polyploid duplications.
Two accessions of the distinctive A. madagascariensis
differed by five steps and did not resolve as a clade
(Fig. 3); however, the sequences contained a number of
polymorphic sites and we did not subclone the PCR
amplifications to isolate individual alleles. One accession [A. madagascariensis (1)] associated closely with A.
ulvaceus (differing by 2 bps); whereas, the A. madagascariensis (2) accession was basal in the clade (Fig. 3).
None of the Malagasy species surveyed differed by
more than nine steps in the ITS tree. The ITS sequences
of A. stachyosporus and A. undulatus were fairly distinct
(differing by 7 bps). Both A. crispus and A. rigidifolius
possessed distinct ITS sequences.
ITS data did not distinguish A. bullosus, A. lancesmithii or A. proliferus, despite their well-marked morphological differences as reported by Hellquist and Jacobs (1998). The ITS data also indicated that a speci-
510
SYSTEMATIC BOTANY
[Volume 30
FIG. 3. Strict consensus of 719 maximum parsimony trees resulting from analysis of ITS data (40 Aponogeton sequences and
two outgroup sequences). Branch lengths (tree steps) are indicated above nodes; bootstrap support (%) is indicated beneath
nodes. Names followed by a bracketed number represent multiple accessions (see Table 1); those marked with an asterisk (*)
are cloned sequences recovered from the same accession. Homologous sequences are indicated in bold type except where they
occur in hybrids (so indicated and marked by light type). Paralogous sequences also are marked in light type (see text).
Geographical distributions (for species grouped in boxes) are abbreviated as in Fig. 1.
men collected in vegetative condition that we found
particularly difficult to identify (A. ‘‘indet.’’; suspected
as being A. elongatus) either was A. bullosus, A. lancesmithii, or A. proliferus rather than A. elongatus, from
which its sequence differed by 11–12 steps (Fig. 3). ITS
data corroborated the distinctness of A. elongatus, A.
queenslandicus, and A. vanbruggenii as well as a close
relationship between A. euryspermus and A. kimberleyensis. However, further details of interspecific relation-
ships among the Australian species could not be resolved by ITS data alone.
The possible existence of two new Australian Aponogeton species was indicated by the recovery of distinctive, monomorphic ITS sequences that did not associate closely with any of the species described previously. One taxon (A. ‘‘species 1’’) was similar to A.
queenslandicus but differed by 3–4 steps from all three
accessions of that species surveyed. The sequence of a
2005]
LES ET AL.: AUSTRALIAN APONOGETON
511
FIG. 4. Single most parsimonious tree resulting from analysis of cpDNA (trnk 59 intron, matK) sequence data from 30
Aponogeton and two outgroup accessions. Branch lengths (tree steps) are indicated above nodes; bootstrap support (%) is
indicated beneath nodes. Names followed by a bracketed number represent multiple accessions (see Table 1). Geographical
distributions (for species grouped in boxes) are abbreviated as in Fig. 1.
second taxon (A. ‘‘species 2’’) was most similar to
those of A. euryspermus and A. kimberleyensis, but differed from them by 6–7 steps.
Alignment of trnK 59 intron/matK data provided 915
aligned nucleotide sites and five indel characters for
phylogenetic analysis (Fig. 4). Of these, 803 sites were
constant and 63 were parsimony-informative. Maximum parsimony analysis produced a single minimallength tree (136 steps) characterized by low homoplasy
(CI 5 0.90; CI(exc) 5 0.84; RI 5 0.91). There was less
resolution among closely-related species than we observed in the ITS analysis, but bootstrap support for
nodes was similar, ranging from 61–100% with six
nodes (50%) supported above 90% (Fig. 4). As with
ITS data, analysis of cpDNA sequences also placed A.
hexatepalus as basal but with even higher bootstrap
support (97%). The cpDNA data further resolved the
positions of A. distachyos (the sister group to the remaining Aponogeton species excluding A. hexatepalus)
and A. robinsonii (the sister group to the remaining
Aponogeton species excluding A. distachyos and A. hexatepalus) with high levels of support (Fig. 4). Likewise,
analysis of cpDNA provided support for a clade comprising the three Malagasy species (95% support) although no finer resolution was achieved within that
clade. The Asian species formed a polytomous, paraphyletic grade (in a basal position to a clade comprising all Australian species except A. hexatepalus) rather
than a clade (Fig. 4). cpDNA provided moderate support (78% bootstrap) for an Australian clade (exclud-
512
[Volume 30
SYSTEMATIC BOTANY
ing A. hexatepalus), as congruent with the ITS data
(Figs. 3, 4). Further details of relationships within this
Australian clade could not be ascertained due to the
virtual lack of resolution or support provided by the
cpDNA sequences. However, both taxa identified as
possible new species by ITS data also possessed distinct cpDNA sequences (Fig. 4).
The combined (ITS, cpDNA) molecular data set
comprised 1845 aligned nucleotide sites and five indel
characters of which 1335 (72%) were constant, 238
(13%) were variable but uninformative, and 277 (15%)
were informative phylogenetically. Approximately
24% of the final molecular data matrix included missing data cells due mainly to the large number of gaps
required for sequence alignment. Maximum parsimony analysis generated 18 minimal-length trees (769
steps) characterized by moderate homoplasy (CI 5
0.83; CI(exc) 5 0.73; RI 5 0.84). The combined data strict
consensus tree was more highly-resolved than either
of those resulting from the independent analyses.
Bootstrap support for nodes was similar, ranging from
,50% (one node) to 100%; however, a slightly larger
proportion (13 nodes; 57%) was supported above 90%
(Fig. 5).
The combined molecular cladogram positioned A.
hexatepalus as sister to the rest of the genus (97% bootstrap support) succeeded in position first by A. distachyos (82%), and then by A. robinsonii (100%; Fig. 5).
Three major species groups were resolved as clades:
the Malagasy species (100%), the ‘‘Asian’’ species (excluding A. robinsonii; 98%), and the ‘‘Australian’’ species (excluding A. hexatepalus; 100%). The arrangement
of species within the Malagasy clade was identical to
the ITS result given that cpDNA data provided no additional resolution within the group. Two subclades
were resolved in the ‘‘Asian’’ clade; one comprising A.
undulatus and A. stachyosporus (100%) and one comprising A. rigidifolius and A. crispus (60%). Aponogeton
stachyosporus and A. undulatus were differentiated by
seven steps in the tree. Within the ‘‘Australian’’ clade,
two major subclades were resolved, but neither was
particularly well-supported by bootstrap values. One
subclade (72%) comprised A. queenslandicus, an undescribed taxon (A. ‘‘species 1’’), and A. vanbruggenii (Fig.
5). The other subclade (supported at less than 50%)
was itself subdivided into three well-supported (88–
100%) subclades that mirrored the ITS results. The second undescribed taxon (A. ‘‘species 2’’) occurred within the same subclade as A. euryspermus and A. kimberleyensis, but was distinguished from both species by
11–12 and 8 steps respectively, in the combined data
tree (Fig. 5).
When mapped on the combined molecular data
cladogram (ACCTRAN and DELTRAN reconstructions), the character states for many morphological
characters showed extensive homoplasy. Only tepal
color (character #15; Table 2) correlated unambiguously with the monophyly of the ‘‘Australian’’ clade whose
representatives all uniquely possess yellow tepals (Fig.
5). Other morphological characters deemed to be highly informative taxonomically (e.g., testa number, character #15; Table 2) were homoplasious and showed numerous instances of multiple origins throughout the
genus (Fig. 5). With one exception (A. hexatepalus), the
classification developed by Camus (1923) was highly
compatible with the cladogram topology generated using the combined molecular data (Fig. 5). The tree topology also depicted clades that, with the exception of
A. hexatepalus and A. robinsonii, correlated well with
distinct geographical regions.
DISCUSSION
Our study of Aponogeton represents the first cladistic
evaluation of these poorly-understood aquatic plants
and also assesses the systematic utility of morphological and molecular data for phylogenetic reconstruction
in the genus. Prior taxonomic studies have been based
exclusively on morphology despite the confounding
level of phenotypic variation encountered in many species (Bruggen 1985). Our incorporation of molecular
characters provided a means of evaluating morphologically-based taxonomic hypotheses using an independent source of data. Such an approach is particularly
important in Aponogeton where the influence of polyploidy and hybridization on morphology is yet to be
determined. Our incorporation of molecular regions
that are inherited both maternally (cpDNA) and biparentally (ITS) has been particularly useful as a
means of providing genetic markers capable of identifying potential hybrids (Moody and Les 2002; Les et
al. 2004).
Morphological Variation. An effective taxonomic
assessment of Aponogeton based on morphology has
proven to be notoriously difficult. Bruggen (1985)
stressed the importance of studying plants extensively
under cultivation in order to fully understand the
range of phenotypic variation that can be encountered
within a species. Leaf characters that appear to be distinctive, such as undulate margins or bullate laminas,
can either appear or disappear when grown under different conditions. Similarly, some species vary widely
in their ability to produce either submersed leaves (often absent from herbarium material) or floating leaves,
making it particularly difficult to identify vegetative
specimens that may lack one or the other leaf type,
especially where such a character has been regarded
as an important distinguishing feature. Understandably, there has been greater confidence in the taxonomic value of reproductive characters that, by virtue of
their aerial disposition, are less influenced by the variable conditions of aquatic habitats. However, some
species flower rarely in culture, making it difficult to
2005]
LES ET AL.: AUSTRALIAN APONOGETON
513
FIG. 5. Strict consensus of 18 maximum parsimony trees recovered from combined analysis of ITS and cpDNA sequence
data obtained for 30 Aponogeton and two outgroup accessions. Branch lengths (tree steps) are indicated above nodes; bootstrap
support (%) is indicated beneath nodes. Names followed by a bracketed number represent multiple accessions (see Table 1).
Two-letter abbreviations after species names indicate their assignment to the classification proposed by Camus (1923): AA 5
sect. Aponogeton subsect. Aponogeton; PO 5 sect. Aponogeton subsect. Polystachys; PL 5 sect. Pleuranthus subsect. Pleuranthus; (N)
designates accessions representing taxa that have not yet been classified. Geographical distributions (for species grouped in
boxes) are abbreviated as in Fig. 1. The circled numbers outside the right edge of the boxes denote species having two testas
(the unmarked accessions have one testa), a character state that is highly homoplasious. The character state of yellow tepals is
unique to and occurs in all representatives of the Australian clade (designated by the arrow).
study the highly simplified reproductive characters
and vegetative characters simultaneously. The relatively unstructured pattern of morphological variation in
Aponogeton has resulted in a fairly small number of
infrageneric classifications that mainly have defined
only a few groups using few characters (e.g., Camus
1923).
The only phylogenetic assessment of Aponogeton
made previously was by Thanikaimoni (1985) who elucidated relationships using a non-cladistic method that
placed species in an inferred network based upon mor-
phological trends observed in the genus (Fig. 1). By
this approach, Thanikaimoni (1985) hypothesized a
basal position for A. longiplumulosus, leading him to
suggest a Malagasy origin of the genus (Fig. 1). However, the interspecific relationships suggested by Thanikaimoni are highly inconsistent with groups in the
classification of Camus (1923) and show poor geographical correlation overall. Species assigned to section Pleuranthus are dispersed broadly across the genus
and are embedded among different groups of species
placed in section Aponogeton. Species of section Apon-
514
SYSTEMATIC BOTANY
ogeton subsection Polystachys are not monophyletic but
occur among different groups of species in subsection
Aponogeton. The monotypic subsection Monostachys
(section Pleuranthus) is not distinct but is embedded
among representatives from subsection Pleuranthus as
well as section Aponogeton. Geographically, the Malagasy species occur throughout the diagram and are
interspersed among species from continental Africa
and India (Fig. 1). The diagram depicts a biphyletic
Australian element with A. hexatepalus in a near-basal
position with the remainder of the Australian (and
New Guinea) species forming a terminal clade derived
from Asian species (Fig. 1). The Asian species are associated relatively closely and, with the exception of
one species (A. satarensis), comprise a paraphyletic
grade. From these examples, we perceive two fundamentally different perspectives of relationships in the
genus: one (Thanikaimoni) inferring that relationships
are best indicated by a plethora of morphological
trends; and another (Camus) where only a few morphological characters define relatively large groups of
related species. Geographical integrity among species
is greater overall in the latter.
We assessed these hypotheses first by conducting a
cladistic analysis of data (Tables 2, 3) comprising those
morphological characters stressed most often in taxonomic treatments of Aponogeton (Bruggen 1969, 1985;
Hellquist and Jacobs 1998). When analyzed phylogenetically (Fig. 2), we found that these data provided a
poor assessment of clades within the genus by offering
little resolution and poor support for all resolved
nodes. Furthermore, some species believed to be closely related (e.g., A. bullosus, A. lancesmithii) were separated quite widely in the cladogram. However, the
morphological data did indicate the distinctness of the
Western Australian A. hexatepalus from all other Australian species, a result in agreement with Bruggen
(1969; p. 136) who concluded that its ‘‘. . . close relationship with any other species of Aponogeton does not
seem probable.’’ The clade of Malagasy species resolved by morphological data was supported weakly,
but (with exception of A. hexatepalus) did unite species
classified by Camus (1923) within sect. Aponogeton subsect. Polystachys (Fig. 2). Although the morphological
cladogram was poorly resolved, the geographical
groupings of species were consistent with results obtained by the molecular analyses (Figs. 3–5).
This analysis indicated that morphological data
alone were insufficient for adequately resolving relationships in Aponogeton. The extensive homoplasy associated with the morphological character states also
may explain why Thanikaimoni (1985) assumed numerous instances of parallel evolution and why Bruggen (1985) found it so frustrating to elucidate a satisfactory classification of the genus based on the bewil-
[Volume 30
dering, mosaic array of morphological features in the
genus.
Molecular Anomalies. Overall, the interrelationships indicated by our phylogenetic analysis of molecular data were far more compelling than those indicated using morphological data. However, we initially
had to clarify several anomalies that appeared in the
molecular analyses, notably those inherent to the biparentally-inherited ITS sequences.
We first had to distinguish the homologous ITS sequences (Fig. 3) from a number of divergent sequences
(see Materials and Methods) that were recovered from
several species (A. queenslandicus, A. stachyosporus, A.
rigidifolius [as A. ‘‘ulvaceus’’], A. vanbruggenii). We believe that these sequences are either paralogs resulting
from polyploidy (probably the case in A. stachyosporus
as the related A. undulatus is highly polyploid), result
from incomplete gene conversion of divergent homologous loci subsequent to hybridization (possibly the
case in A. queenslandicus (2), A. ‘‘ulvaceus’’ and A. vanbruggenii (1) which are hybrids but also could be polyploid) or are generated otherwise as cloning artifacts.
Similarly divergent chimeric ITS sequences have been
recovered from F1 hybrids of other aquatic plants
(Moody and Les 2002). Chromosome counts of these
species would be helpful in evaluating these interpretations but as yet are unavailable.
Hybridization. The presence in a single specimen
of homologous ITS sequences originating from two
distinct species indicated that some accessions clearly
were of hybrid origin. This approach revealed that
Australian species are capable of hybridization with
other Australian species [A. vanbruggenii (1) 5 A. vanbruggenii 3 A. bullosus, A. lancesmithii or A. proliferus]
as well as Asian species [A. queenslandicus (2) 5 A.
queenslandicus 3 A. rigidifolius]. Also compelling was
the recovery of divergent sequences cloned from an
accession of the Sri Lankan A. rigidifolius (that we erroneously had identified tentatively as A. ‘‘ulvaceus’’)
from within the Australian clade (Fig. 3), which indicated the hybridization of A. rigidifolius with some
Australian species (probably occurring while in cultivation). However, our failure to recover homologous
clones from this accession precluded a more definitive
assignment of parentage to this accession other than
to A. rigidifolius with which it shared the maternallyinherited cpDNA markers. These examples indicate
that intrinsic barriers to hybridization in Aponogeton
may be weak even among species originating from distant geographical regions. Furthermore, we observed
that hybrids generally resembled their maternal parent
(as indicated by cpDNA sequences) to which the accessions were assigned taxonomically at the time of
collection.
Bruggen (1985) believed that natural hybridization
did not occur in Aponogeton despite the ability to cross
2005]
LES ET AL.: AUSTRALIAN APONOGETON
a number of species artificially. Unfortunately, we cannot evaluate this possibility further given that the hybrid accessions that we examined all originated from
ponds in cultivation where the hybridization undoubtedly took place. Thus, we can demonstrate only that
hybridization among these taxa is possible, but not that
such hybrids could persist in nature. However, Bruggen (1985) may have overly relied on the ability of
some morphological characters to estimate proclivity
for hybridization. For example, he assumed that hybridization was ‘‘highly improbable’’ between species
that differed by their testa number (Bruggen 1985).
However, when mapped on the combined molecular
cladogram, the distribution of species having single vs.
double testa is highly homoplasious (Fig. 5). Moreover,
there are a number of species in our analysis that
group as sister species, yet differ by their testa number.
Thus, hybridization between sister species of different
testa number predictably would be much more likely
than between distantly related species having the same
testa number. Testa number is a good example of how
homoplasious morphological data have provided misleading indications of phylogenetic relationships in
Aponogeton.
Major Phylogenetic Features of Aponogeton. Once
we had accounted for the unusual paralogous, hybrid
or otherwise divergent ITS sequences, the results from
phylogenetic analyses of ITS and cpDNA yielded similar results. No data supported the recognition of a
segregate genus Ouviranda as being distinct from Aponogeton. Several species proposed for inclusion within
Ouviranda (e.g., A. crispus, A. madagascariensis, A. undulatus) did not form a clade but were embedded
among other Aponogeton species (Figs. 2–5). Bruggen
(1985) previously had dismissed the segregate Ouviranda because the type (A. madagascarensis) possesses
characters that are discordant with those that allegedly
define the genus.
Both molecular data sets (individually and combined) placed the W. Australian A. hexatepalus sister to
the rest of the genus (Figs. 3–5) with moderate to
strong internal support. This result differs from Thanikaimoni (1985) who regarded A. longiplumulosus as the
ancestral Aponogeton species, but whose phylogenetic
scheme placed A. hexatepalus as relatively closely related (Fig. 1). By specifically identifying A. hexatepalus as
sister to the rest of Aponogeton (and A. longiplumulosus
as comparatively derived), our results indicate that the
genus is more likely to have originated in Australia
and not in Madagascar as Thanikaimoni concluded.
The placement of the African A. distachyos in a position between A. hexatepalus and A. robinsonii also is
well-supported by combined molecular data analysis
(Fig. 5) and also varies considerably from Thanikaimoni (1985) whose phylogenetic scheme placed A. distachyos in a fairly derived position (Fig. 1). A survey of
515
additional African species that includes both secund
and omnilaterally-flowered species would be highly informative and would be necessary to estimate whether
African species are monophyletic or reflect multiple
colonizations. More in accord with Thanikaimoni’s
scheme was the phylogenetic placement of A. robinsonii
in the combined molecular analysis (Fig. 5), which
strongly supported an isolated position of the species
between A. distachyos and the remaining species. Similarly, Thanikaimoni (1985) placed A. robinsonii in a relatively basal position near A. hexatepalus (Fig. 1). The
Vietnamese A. robinsonii is unusual among other Asian
Aponogeton by its secund flowers and paired spikes for
which Camus assigned it to sect. Pleuranthus (otherwise comprising four African and one Malagasy species). Thus, it is not surprising that this species associated closely with the African A. distachyos (also sect.
Pleuranthus) but distantly from the other Asian species
surveyed (all sect. Aponogeton) in both molecular analyses.
Beyond these three relatively isolated, basal species,
our combined molecular analysis resolved three major,
highly-supported clades consisting of 1) Malagasy species, 2) remaining Asian species, and 3) remaining
Australian species with the latter two clades constituting a sister group (Fig. 5). With one exception (A. satarensis Sundararagh., A. R. Kulk. & S. R. Yadav, which
we did not survey), Thanikaimoni’s (1985) scheme positioned the Asian species as a paraphyletic grade that
gave rise ultimately to the Australian/New Guinea
species (Fig. 1), a result not differing in essence from
that depicted by our combined molecular cladogram.
However, Thanikaimoni’s (1985) placement of the Malagasy species that we surveyed differed substantially
(see below).
Malagasy Species. All four accessions that we surveyed (A. longiplumulosus, A. madagascariensis (1, 2), A.
ulvaceus) showed polymorphic ITS sequences, indicating that they might be of polyploid or hybrid origin.
However, we did not clone any of these accessions for
further clarification because they were not the focus of
our study and because their sequences formed a single,
well-supported clade (Fig. 3), thus providing adequate
resolution of their phylogenetic position in the family.
Notably, the two accessions of the distinctive ‘‘lace
plant’’ (A. madagascariensis) did not associate together;
the ITS sequence of one specimen with fine leaf fenestration [A. madagascariensis (1)] was more similar to
that of A. ulvaceus than it was to a more coarsely fenestrate, conspecific accession [A. madagascariensis (2)].
Although the possibility that our accessions included
a hybrid between A. madagascariensis and A. ulvaceus
cannot be ruled out, further study and cloning would
be necessary to clarify this question. Both species have
been synthetically hybridized successfully (Bruggen
1985) and all three species appear to be closely related
516
SYSTEMATIC BOTANY
by virtue of their identical cpDNA sequences (Fig. 4)
and low level of ITS sequence divergence (Fig. 3).
The three Malagasy species included in our analysis
formed a single clade, whether generated using morphological (Fig. 2), ITS (Fig. 3), or cpDNA (Fig. 4) data.
Combined molecular data provided 100% bootstrap
support for this clade (Fig. 5). This result is strongly
at odds with Thanikaimoni’s (1985) phylogenetic
scheme, which depicted the three species as unrelated
and dispersed widely across the genus (Fig. 1). Although we did not include a large sample of Malagasy
species in our analysis, our results indicate that at least
these species are much more closely related to each
other than previously had been thought. Furthermore,
our result is congruent with the classification of Camus
(1923) who assigned all three species to subsect. Polystachys.
Asian Species. Bruggen (1985) treated A. undulatus
and the narrower-leaved A. stachyosporus as conspecific, assuming that wide variability of leaf morphology
existed in A. undulatus. This conclusion would be supported by our cpDNA sequences, which were identical
in the two taxa (Fig. 4); however, we found their ITS
sequences to differ by seven substitutions (Fig. 3), a
level comparable to (or exceeding) the degree of ITS
divergence that we observed between a number of other Aponogeton species. The ITS data would indicate that
the distinctness of these two taxa should be reconsidered. We also compared two accessions of A. rigidifolius
that differed conspicuously by their leaf-apex morphology (obtuse vs. acute). Despite this morphological
difference, the ITS and cpDNA sequences of both accessions were identical, thus providing no evidence
that these accessions might represent taxa worthy of
nomenclatural distinction. Bruggen’s (1985, p. 47) presumption that A. rigidifolius: ‘‘. . . is, no doubt, closely
related to A. crispus’’ is corroborated by our combined
molecular cladogram, which grouped the pair as sister
species (Fig. 5). Thanikaimoni (1985) also depicted a
close relationship between these species (Fig. 1). Considering the five species that we have now evaluated,
the Asian Aponogetons appear to be biphyletic, with
A. robinsonii clearly distantly related to the other species. A survey of additional Asian species (especially
the secund-flowered A. satarensis) could readily determine whether additional clades exist among them.
Australian Species. Despite the opportunity that
this study has given us to address other issues, our
primary interest was to evaluate phylogenetic relationships among the Australian Aponogetons, a group in
which many species have been described only recently.
Because we had access to material of all the known
Australian species, we believed that a comprehensive
assessment was possible.
Our analyses of Aponogeton have provided conclusive evidence for the biphyletic origin of extant native
[Volume 30
TABLE 4. Refined classification of 18 Aponogeton species proposed as a result of phylogenetic analyses.
Aponogetonaceae J. Agardh
1. Aponogeton L.f.
1. Section Aponogeton
1. Subsection Aponogeton
1. A. crispus Thunb.
2. A. rigidifolius H.Bruggen
3. A. stachyosporus de Wit
4. A. undulatus Roxb.
2. Subsection Polystachys A.Camus
5. A. longiplumulosus H.Bruggen
6. A. madagascariensis (Mirbel) H.Bruggen
7. A. ulvaceus Baker
2. Section Flavida Les, S.W.L.Jacobs & M.Moody
3. Subsection Flavida
8. A. bullosus H.Bruggen
9. A. elongatus F.Muell. ex Bentham
10. A. euryspermus Hellq. & S.W.L.Jacobs
11. A. kimberleyensis Hellq. & S.W.L.Jacobs
12. A. lancesmithii Hellq. & S.W.L.Jacobs
13. A. proliferus Hellq. & S.W.L.Jacobs
14. A. queenslandicus H.Bruggen
15. A. vanbruggenii Hellq. & S.W.L.Jacobs
3. Section Pleuranthus A.Camus
4. Subsection Pleuranthus
16. A. distachyos L.f.
17. A. robinsonii A.Camus
4. Section Viridis Les, S.W.L.Jacobs & M.Moody
5. Subsection Viridis
18. A. hexatepalus H.Bruggen
Australian species with A. hexatepalus clearly not closely related to the other Australian taxa (see above). The
sister position of A. hexatepalus to the rest of the genus
not only points to a possible Australian origin for the
family, but further indicates that Aponogetonaceae
conceivably originated under temperate climatic conditions (both A. hexatepalus and A. distachyos presently
inhabit temperate areas) and have radiated subsequently into the tropics. Camus (1923) classified A. hexatepalus within sect. Aponogeton (subsect. Polystachys)
because of its branched, omnilaterally-flowered spikes.
However, A. hexatepalus is distinct phylogenetically
from other surveyed members of subsect. Polystachys
that constituted a distinct clade (Fig. 5). The existing
classification of Aponogeton would be improved by
placing A. hexatepalus within a distinct section that better reflects its isolated position (Table 4). Aponogeton
hexatepalus can be distinguished morphologically from
other species in the genus by its six tepals that are
green in color (Tables 2, 3).
All other Australian Aponogeton species occur within
a clade (the ‘‘Australian clade’’) that is well-supported
by ITS and cpDNA data (Figs. 3, 4). The combined
molecular cladogram supports this group at 100%
2005]
LES ET AL.: AUSTRALIAN APONOGETON
(Fig. 5). Thanikaimoni’s (1985) phylogenetic scheme
also portrayed the group as monophyletic with the inclusion of the two endemic New Guinea species (Fig.
1). Although our morphological analysis failed to resolve the Australian clade (Fig. 2), we found one character state (yellow tepal color) that uniquely defined
the group (Tables 2, 3; Fig. 5). Bruggen (1969) commented that the yellow tepal color of the four native
Australian species with which he was familiar was ‘‘remarkable’’ and uncommon elsewhere in the genus.
Our studies confirm that all eight species within the
Australian clade possess yellow tepals (Fig. 5); this
morphological character state does indeed appear to
represent a reliable taxonomic marker for the group.
Of the New Guinea species (which we did not survey),
A. loriae Martelli possesses yellow tepals; they are
whitish in A. womersleyi H. Bruggen (Leach and Osborne 1985). It would be informative to include these
species in subsequent analyses to determine whether
they both are derived from the Australian clade as
Thanikaimoni (1985) suggested.
All three species of the Australian clade that were
known to Bruggen (1985) were placed by him within
sect. Aponogeton subsect. Aponogeton, but Hellquist and
Jacobs (1998) did not classify any of the five new species that they named in this group. From our results,
it would be reasonable to classify all eight species of
the Australian clade within sect. Aponogeton subsect.
Aponogeton. However, the recognition of this clade as a
distinct section of Apongeton also is justified and would
further enhance the information content of the existing
classification. Because the type species (A. natans) is
not part of the Australian clade, the establishment of
a new section name is necessary (Table 4).
Thanikaimoni (1985) concluded that species in the
Australian clade were derived from a Malesian
‘‘source,’’ which itself was derived from Indian
‘‘stock.’’ Results from ITS and cpDNA analysis corroborate this hypothesis by placing the Australian and
Asian species within a single clade (Figs. 3, 4). Internal
support for this clade was high (95%) in the combined
molecular analysis (Fig. 5). Although results of our
morphological analysis were compatible (Fig. 2), the
extremely poor resolution precluded a definitive assessment and provided no indication of morphological
characters that might define this association.
Prior to this study, phylogenetic relationships among
only four of the nine native Australian species had
been hypothesized specifically, and these were estimated using non-cladistic methods (Fig. 1). Other authors (Aston 1973; Bruggen 1969, 1985; Hellquist and
Jacobs 1998) had commented on possible relationships
of the Australian species but again without the use of
cladistic methods.
Bruggen (1985), who recognized only three species
in what we have called the Australian clade, indicated
517
that A. bullosus, A. elongatus, and A. queenslandicus were
not particularly closely related, a result supported by
our molecular analyses (Fig. 5). He remarked that the
submersed leaves of A. queenslandicus sometimes resembled A. bullosus yet the species were ‘‘easily distinguished’’ by their reproductive characters. Although
Bruggen (1985) observed that A. elongatus was ‘‘impossible to distinguish’’ vegetatively from some forms
of A. queenslandicus, he believed that A. queenslandicus
was related more closely to several Asian species (A.
lakhonensis, A. natans) than to any of the Australian
species (Bruggen 1985). Most likely, Bruggen based his
conclusion on the presence of a double testa in these
three species (cf. single in A. elongatus and A. bullosus),
a character state he regarded as diagnostic taxonomically, but one that we have demonstrated as being
highly homoplasious. With moderate support (72%),
our combined molecular cladogram (Fig. 5) grouped
A. queenslandicus as the sister species to A. vanbruggenii,
a taxon not known to Bruggen but specimens of which
he included in his concept of A. elongatus. Contrary to
Bruggen’s assessment, both species are fairly closely
related to others in the Australian clade as indicated
by the low level of molecular divergence among them
(Fig. 5). Interestingly, Hellquist and Jacobs (1998) commented on a specimen of A. vanbruggenii from the Atherton Tablelands that greatly resembled A. queenslandicus vegetatively. Our phylogenetic analysis of morphological data (Fig. 2) also indicated a close relationship between the species. Within the A. queenslandicus/
A. vanbruggenii clade was one accession (A. ‘‘species
1’’) that was difficult to identify conclusively. This taxon associates with, but is distinct from, A. queenslandicus at the molecular level (Fig. 5). We believe this
taxon (for which we have not yet seen fruiting material) to represent an undescribed species and currently
have initiated a more thorough study of it.
Bruggen (1969) presumed a close relationship between A. bullosus and A. elongatus, which Aston (1973)
later remarked were ‘‘not always easy to distinguish.’’
Our results (Fig. 5) indicate that A. bullosus is related
most closely to A. lancesmithii and A. proliferus, from
which it cannot be distinguished by any of the molecular data that we evaluated. However, because the latter two species were not recognized by Bruggen at the
time of his work (which included their material in A.
elongatus), his hypothesized relationship between A.
bullosus and A. elongatus is accurate given that the latter
represents the next closest species phylogenetically
(Fig. 5). Furthermore, because Bruggen actually had
included material of what eventually was transferred
to A. lancesmithii within A. bullosus, their close relationship as evidenced by molecular data is not surprising. Aponogeton lancesmithii resembles A. bullosus
by its ability to produce bullate laminas (a trait unknown in other Australian species), but differs from A.
518
SYSTEMATIC BOTANY
bullosus by possessing a double testa and extremely
long-emergent inflorescences with chasmogamous
flowers (A. bullosus has short usually submerged inflorescences with cleistogamous flowers). The species are
identical for the molecular data that we surveyed. Here
is yet another example of the potentially misleading
homoplasy associated with testa number, which varies
even between what appear to be very closely related
species. Hellquist and Jacobs (1998) believed initially
that A. lancesmithii was a hybrid between A. bullosus
and A. elongatus but ultimately concluded that it was a
distinct species. Our molecular analyses provided no
evidence of hybridization between these species.
Aponogeton proliferus is one of only two proliferous
species in the genus, the other being the Asian A. undulatus (Bruggen 1985). Despite the uniqueness of this
feature, there is no question that proliferous shoots
arose independently in the genus as evidenced by the
vastly different placement of these two species phylogenetically (Figs. 3–5). Morphological data indicate a
close relationship between A. proliferus and A. bullosus
but not with A. lancesmithii, which is placed among a
group of species possessing a double testa (Fig. 2).
However, the close relationship of A. proliferus to both
A. bullosus and A. lancesmithii is supported strongly by
molecular data (Figs. 3, 5) and again calls into question
the phylogenetic utility of testa number. Hellquist and
Jacobs (1998) surmised that A. proliferus was most
closely related to A. elongatus; however, the present results show these species to be considerably more distantly related.
Hellquist and Jacobs (1998) suggested a close relationship between A. kimberleyensis and A. euryspermus
and also between the latter and A. elongatus. Morphological data separate these species phylogenetically
(Fig. 2), but mainly by virtue of their different testa
numbers. Molecular data (Figs. 3, 5) group A. kimberleyensis and A. euryspermus as sister species with a high
degree of bootstrap support (95%), a result that raises
further doubt on the taxonomic reliability of testa
number. The association of these two sister species
with A. elongatus (as proposed by Hellquist and Jacobs)
is legitimate given their proximity to the latter species
phylogenetically (Fig. 5). However, there is a confounding issue regarding one accession (A. ‘‘species 2’’) that
we could not identify confidently to species, and which
groups as a sister to the clade comprising A. kimberleyensis and A. euryspermus (Fig. 5). Tentatively, we consider this material to represent a second undescribed
species whose taxonomic status we have deferred
pending further study.
Lastly we consider the relationships of A. elongatus,
a species that has been allied variously to A. bullosus
(Bruggen 1969), A. euryspermus, A. lancesmithii, and A.
proliferus (Hellquist and Jacobs 1998). Indeed, A. elongatus is closely related to all of these species (Fig. 5)
[Volume 30
and groups centrally in the Australian clade by its
morphology (Fig. 2). Unfortunately, the limited resolution of our morphological and molecular cladograms
precludes a more precise estimation of relationship for
A. elongatus. Hellquist and Jacobs (1998) recognized
two distinct subspecies of A. elongatus whose evaluation would require the use of genetic markers having
finer resolution than those that we surveyed. However,
our observation of minor DNA sequence divergence
among the four accessions of A. elongatus that we surveyed indicates that this species possesses at least a
moderate degree of interpopulational genetic variation.
Overall, the elucidation of phylogenetic relationships
on the basis of perceived morphological trends (e.g.,
Thanikaimoni 1985) has depicted species relationships
successfully in some instances but has failed badly in
others. On the other hand, the classification developed
by Camus (1923) incorporated only a few characters
but appears to be reasonably compatible with our phylogenetic assessment of the genus, requiring only slight
modifications (Table 4).
Molecular data have proven to be far more reliable
than morphology for elucidating phylogenetic relationships in Aponogeton, yielding cladograms with relatively high resolution and nodal support that greatly
facilitate study of the genus. The comparison of biparentally-inherited ITS sequences has provided evidence
of hybridization in several instances and detected paralogous loci that reflect the extensive polyploidy that
pervades the genus. Certainly, a more comprehensive
assessment of chromosome numbers for Aponogeton
would provide a useful adjunct to phylogenetic studies
of the group. Intrinsic barriers to hybridization do not
appear to be developed strongly in Aponogeton despite
the apparent lack of natural hybrids, which may be
due to the appropriate species rarely growing sympatrically. Molecular data also have indicated the possible existence of two undescribed Aponogeton species
from Australia, the distinctness of which, had gone unnoticed previously from material that had been evaluated only morphologically, and indicates that there
may be further benefit from studying populations still
referred to A. elongatus.
Our analyses indicate that Aponogetonaceae conceivably originated in Australia and experienced early
radiations into Africa and Asia. Subsequent diversification in Aponogeton yielded relatively discrete groups
that radiated in Madagascar, Asia and ultimately back
to Australia where considerable speciation has occurred since. Australian Aponogeton species represent
an actively evolving group in which new species continue to be discovered. The recent origin of several species (e.g., A. bullosus, A. lancesmithii, A. proliferus) is evidenced by their distinct morphology coupled with a
virtual lack of detectable molecular divergence in the
genes sequenced.
2005]
LES ET AL.: AUSTRALIAN APONOGETON
The study of additional species in a similar fashion
will be necessary before a comprehensive, phylogenetically defensible classification of Aponogeton can be
achieved. However, this study has made some progress
towards this objective by providing evidence in support of establishing two new sections (Table 4) that we
describe below.
TAXONOMIC TREATMENT
1. Aponogeton L.f. sect. Flavida Les, M. Moody & S.
W. L. Jacobs, sect. nov.—TYPE SPECIES: A. queenslandicus H. Bruggen
Tepalis tribus, inflorescentia omnino sulphurea a
congeneribus differt.
Differing from other related sections in having three
tepals and a completely sulphur-yellow inflorescence.
The section contains 8–10 species distributed throughout tropical regions of Australia.
2. Aponogeton L.f. sect. Viridis Les, M. Moody & S.
W. L. Jacobs, sect. nov.—TYPE SPECIES: A. hexatepalus H. Bruggen
Tepalis sex vice tribus, semper viridibus, floribus
circum axem patentibus, inflorescentia ramosa viridique, a congeneribus differt. Hieme florens; in terram
hieme inundatam.
Differing from other related sections in having six
always green tepals instead of three, flowers spread
around the axis, and the inflorescence branched and
green. Flowering in winter; growing in winter-flooded
habitats. The section comprises one species (A. hexatepalus), which is distributed in temperate regions of
Western Australia.
ACKNOWLEDGEMENTS. We thank Geoff Sainty, Lance Smith,
and Dave Wilson for providing many specimens critical to our
project, and to Karen Wilson for providing the Latin diagnoses.
Portions of this work were funded by an NSF grant (DEB-9806537)
and Fulbright award to DHL and support to SWLJ from the RBG,
Sydney.
LITERATURE CITED
ARENDS, I. J. C. 1985. Somatic chromosome numbers. Pp. 9–10 in
Monograph of the genus Aponogeton (Aponogetonaceae), ed. H. W.
E. Van Bruggen. Bibliotheca Botanica Heft 137. Stuttgart: E.
Schweizerbart’sche Verlagsbuchhandlung (Nägele u. Obermiller).
ASTON, H. I. 1973. Aquatic plants of Australia. Carleton: Melbourne
University Press.
519
BRUGGEN, H. W. E. VAN. 1969. Revision of the genus Aponogeton
(Aponogetonaceae): III. The species of Australia. Blumea 17:
121–137.
——— 1985. Monograph of the genus Aponogeton (Aponogetonaceae).
Bibliotheca Botanica Heft 137. Stuttgart: E. Schweizerbart’sche Verlagsbuchhandlung (Nägele u. Obermiller). 97 pp.
CAMUS, A. 1923. Le genre Aponogeton L.f. Bulletin de la Société Botanique de France 70: 670–676.
COOK, C. D. K. 1996. Aquatic and wetland plants of India: a reference
book and identification manual for the vascular plants found in permanent or seasonal fresh water in the subcontinent of India south
of the Himalayas. New York: Oxford University Press.
COWIE, I. D., P. S. SHORT, and M. OSTERKAMP MADSEN. 2000. Floodplain flora—A flora of the coastal floodplains of the Northern Territory, Australia. Flora of Australia Supplementary Series No.
10. Canberra: Australian Biological Resources Study.
GRANT, V. 1981. Plant speciation. New York: Columbia University
Press.
GUNASEKERA, L. 2003. Cape pond lily (Aponogeton distachyos):
South African food plant—emerging aquatic weed in Victoria. Weedwatch 2: 6.
HELLQUIST, C. B. and S. W. L. JACOBS. 1998. Aponogetonaceae of
Australia with descriptions of six new taxa. Telopea 8: 7–19.
KASSELMANN, C. 1995. Aquarienpflanzen. Stuttgart: Verlag Eugen
Ulmer. 472 pp.
KRAUSE, K. and A. ENGLER. 1906. IV. 13. Aponogetonaceae. Pp. 1–
22 in Das Pflanzenreich. Regni vegetabilis conspectus, Heft 24,
ed. A. Engler. Leipzig: W. Engelmann.
LEACH, G. J. and P. L. OSBORNE. 1985. Freshwater plants of Papua
New Guinea. Papua New Guinea: The University of Papua
New Guinea Press.
LES, D. H., M. A. CLELAND, and M. WAYCOTT. 1997. Phylogenetic
studies in Alismatidae, II: evolution of marine angiosperms
(‘seagrasses’) and hydrophily. Systematic Botany 22: 443–463.
———, M. L. MOODY, and A. S. DORAN. 2004. A genetically confirmed intersubgeneric hybrid in Nymphaea L. (Nymphaeaceae Salisb.). HortScience 39: 219–222.
MADDISON, D. R. and W. P. MADDISON. 2000. MacClade 4: analysis
of phylogeny and character evolution. Sunderland: Sinauer
Associates.
MOODY, M. L. and D. H. LES. 2002. Evidence of hybridity in invasive water milfoil populations. Proceedings of the National
Academy of Sciences, USA 99: 14867–14871.
PAGE, R. D. M. and E. C. HOLMES. 1998. Molecular evolution: a phylogenetic approach. Oxford: Blackwell Science.
PEMBERTON, R. W. 2000. Aponogeton distachyos, a recently domesticated aquatic food crop in Cape South Africa with unusual
origins. Economic Botany 54: 145–149.
SWOFFORD, D. L. 1998. PAUP*: phylogenetic analysis using parsimony (*and other methods). Version 4.0. Sunderland: Sinauer
Associates.
THABREW, W. V. DE. and W. V. DE THABREW. 1983. Water plants of
Sri Lanka. New Malden: Suhada Press.
THANIKAIMONI, G. 1985. Palynology and phylogeny. Pp. 11–14 in
Monograph of the genus Aponogeton (Aponogetonaceae), ed. H. W.
E. Van Bruggen. Bibliotheca Botanica Heft 137. Stuttgart: E.
Schweizerbart’sche Verlagsbuchhandlung (Nägele u. Obermiller).
TRICKER, W. 1897. The water garden. New York: A. T. De La Mare
Printing and Publishing Co., Ltd.