| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
National Herbarium of New South Wales, Botanic Gardens Trust, Mrs Macquaries Road, Sydney 2000 Australia
Received for publication June 8, 2005. Accepted for publication May 29, 2006.
ABSTRACT
To better understand the historical biogeography of the southern hemisphere and evolutionary responses of plants to aridity, we undertook a detailed phylogenetic study of the predominantly southern family Elaeocarpaceae sensu lato (including Tremandraceae). Plastid trnL-trnF and nuclear ITS sequence data were analyzed using parsimony and Bayesian methods and molecular evolutionary rates calibrated using the Oligocene fossil record of Elaeocarpus mesocarps to estimate the minimum divergence dates. The results indicate the monophyly of all recognized genera and a placement for the former Tremandraceae (three genera and about 49 species of shrubby, dry-adapted Australian plants) within the widespread predominantly rainforest tree family Elaeocarpaceae (nine genera, over 500 species). The former Tremandraceae clade diverged from its sister (Aceratium + Elaeocarpus + Sericolea) during the Paleocene, after which it underwent a marked acceleration in evolutionary rate. Furthermore, this lineage diversified during the late Miocene, coincident with widespread aridification in Australian environments and extensive radiations of several sclerophyllous groups. The role of dispersal in explaining the current geographical distribution of Elaeocarpaceae is illustrated by Aristotelia. This genus, whose distribution was previously thought to reflect Gondwanan vicariance, is shown to have arrived in New Zealand from Australia at least 6–7 million yr ago.
Key Words: Australia • dispersal • Elaeocarpaceae • historical biogeography • molecular clock • molecular phylogeny • Tremandraceae
Australia's flora is highly diverse and has high levels of endemism, a product of both its long period of geographic isolation and great climatic change during the Tertiary. The separation of Australia from Antarctica in the early Oligocene established the circum-Antarctic current. This current contributed to significant changes in weather patterns, resulting in increased aridity in Australia from the Miocene onward. In response, the rainforests, which covered much of the continent, retreated toward the coasts, and certain groups radiated into the newly arid interior. Collision with the Southeast Asian plate allowed significant biological exchange between the Australian and Asian floras in the late Tertiary. Today, the species composition of Australia's rainforests reflects both an east Gondwanan heritage and an Asian geographical neighborhood. This complex history makes Australian rainforests a unique model for the investigation of plant evolutionary patterns.
Understanding the patterns and processes underlying the adaptation of Australia's plants to the changing environment requires knowledge of their evolutionary history. Molecular phylogenies can often provide a robust and well-resolved estimate of relationships among organisms, which, in the absence of a complete fossil record, are a most useful source of information on the evolution and historical biogeography of the organisms. Much has been learned already from molecular phylogenies about the relationships and evolution of conspicuous members of the Australian flora such as Myrtaceae (Gadek et al., 1996
; O'Brien et al., 2000
; Steane et al., 2002
), Ericaceae subfamily Styphelioideae (Crayn and Quinn, 2000
; Kron et al., 2002
; Quinn et al., 2003
), Proteaceae (Hoot and Douglas, 1998
; Mast and Givnish, 2002
) and Fabaceae (Wojciechowski, 2002
; Crisp and Cook, 2003
; Miller et al., 2003
). This knowledge has provided valuable new perspectives on the evolution of Australia's vegetation. Recent work with limited sampling suggests that the endemic Australian family Tremandraceae is phylogenetically nested within the Elaeocarpaceae (Savolainen et al., 2000
; Soltis et al., 2000
; Bradford and Barnes, 2001
). Some recent classifications have formally sunk Tremandraceae (APG, 2003
; Coode, 2004
). This relationship presents an enticing opportunity to study a continental-scale radiation of small, dry-adapted shrubs from precursors that belong to a widespread group of predominantly rainforest trees. Such studies can provide useful data for investigating the processes by which plants adapt to increasing aridity.
Elaeocarpaceae are trees, sometimes of towering stature or less often shrubs, and usually occur in rainforests or relatively moist environments although a few species have colonized drier habitats. The more than 500 species in nine genera are distributed mainly in South America, Australasia, and Southeast Asia with outliers in Madagascar. The family is absent from continental Africa, Europe, and North America. This distribution has been cited as evidence of a southern origin of the family (Raven and Axelrod, 1974
). Fossil wood from the Tertiary of India (Prakash and Dayal, 1964
) and the Palaeocene of Patagonia (Petriella, 1972
) has been attributed to Elaeocarpaceae. Eocene leaf fossils of Elaeocarpaceae are abundant in southern Australia, suggesting the family was widespread and diverse in the early Tertiary on that continent (Christophel and Greenwood, 1987
; Christophel et al., 1987
), but the identification of Elaeocarpaceae leaves is problematic (Pole, 1996
; Barnes et al., 2001
). Australasia supports the greatest taxonomic diversity of extant Elaeocarpaceae having seven of the nine genera, including the endemic genera Aceratium, Dubouzetia, Peripentadenia, and Sericolea. The small genera Crinodendron and Vallea are endemic to South America.
The former Tremandraceae are endemic to Australia, and the group comprises three genera and c. 54 species (R. Butcher, Western Australian Herbarium, personal communication). They are wiry shrubs, often lax or trailing, sometimes leafless, and typically inhabit relatively dry habitats ranging from sclerophyll forests to rock outcrops in the arid zone. Platytheca and Tremandra (each with two species) are endemic to southwest Western Australia, whereas Tetratheca (c. 50 species) is distributed across southern Australia (Thompson, 1976
; Alford, 1995
).
Here we used plastid and nuclear DNA sequences to further explore in detail phylogenetic relationships within Elaeocarpaceae sensu lato (s.l., i.e., including Tremandraceae). This study aims to (1) determine the relationships of the former Tremandraceae within Elaeocarpaceae by using a much denser sampling of taxa in the two groups than has hitherto been undertaken, (2) test the monophyly of genera within Elaeocarpaceae s.l., and (3) use the molecular data together with the data from the palaeobotanical literature to infer divergence dates and the biogeographic history of the major clades within the groups.
MATERIALS AND METHODS
Taxa
Ingroup taxa were sampled on the basis of the availability of material and the capacity to best represent the known diversity within and among groups in Elaeocarpaceae s.l. For several taxa, fresh or silica-gel-dried material was not available for DNA extraction, so recently collected herbarium specimens were used. Representatives of each of the 12 currently recognized genera (Thompson, 1976
; Coode, 1987
; Coode 2004
) were obtained: five species of Aceratium (of c. 20 species; Coode, 2004
; distributed from Celebes and Moluccas, New Guinea, New Hebrides to northeastern Australia; taxa sampled from northeastern Australia and New Guinea), all five extant species of Aristotelia (southeastern Australia, New Zealand, Chile, Argentina; Coode, 1985
), two species of Crinodendron (four species; Bricker, 1991
; central Bolivia, central and southern Chile, southeastern Brazil, northwestern Argentina; taxa sampled from northern and southern Chile), four species of Dubouzetia (c. 11 species; Coode, 1987
; Moluccas, New Guinea, New Caledonia, N. Australia; taxa sampled from New Guinea, New Caledonia and N. Australia), 13 species of Elaeocarpus (no recent monograph available, but the number of species is probably at least 300; M. J. E. Coode, Royal Botanic Gardens Kew, personal communication; widespread in Southeast Asia and distributed west to India, Madagascar and Mauritius, north to Japan, east to Hawaii and south to Australia and New Zealand; taxa sampled from Australia, New Guinea, and New Zealand), both species of Peripentadenia (northeastern Australia; Coode, 1987
), one of the two species of Platytheca (Thompson, 1976
; southwest Western Australia), three species of Sericolea (c. 11 species; Coode, 1981
; New Guinea), six species of Sloanea (c. 120 species; Earle-Smith, 1954
; Coode, 1983
; Smith, 1996
; India, Indochina, South China, Malesia, Australia, New Caledonia, Madagascar, tropical Americas; taxa sampled from Australia, New Guinea, and the Caribbean), six species of Tetratheca (c. 50 species; Thompson, 1976
; R. Butcher, Western Australian Herbarium, personal communication; southern Australia; taxa sampled from across the range), both species of Tremandra (Thompson, 1976
; southwest Western Australia), and one of the two species of Vallea (Jaramillo Azanza, 1988
; Chile, Ecuador) (Appendix).
Outgroups were selected on the basis of previous studies demonstrating that the closest relatives of Elaeocarpaceae include Brunelliaceae, Cephalotaceae, Connaraceae, and Cunoniaceae s.l. (i.e., including Baueraceae, Davidsoniaceae, and Eucryphiaceae: Savolainen et al., 2000
; Soltis et al., 2000
; Bradford and Barnes, 2001
). Sequences of the trnL intron and the trnL-trnF spacer of 14 representatives of these families were obtained from GenBank (one representative from each of the first three families and the remainder from Cunoniaceae).
Sequences of the ITS region (comprising ITS1, ITS2, and 5.8S) of the nuclear ribosomal DNA were obtained for 35 of the same ingroup taxa to provide an independent molecular estimate of the phylogeny, to identify possible evolutionary reticulation events, and to better resolve relationships within groups that showed low trnL-trnF region sequence divergence.
Amplification and sequencing
DNA was extracted with DNEasy Plant Mini kits (Qiagen, Hilden, Germany) by following the manufacturer's instructions. PCR amplifications were performed in a CP2-03 Thermal Cycler (Corbett Research, Mortlake, Australia). Each 25-µL reaction contained 200 µM of each primer, 200 µM of each dNTP, 2.5 µL 10x buffer (Bioline, Luckenwalde, Germany), 1.5–3.0 mmol MgCl2, and 0.5 units BIOTAQ DNA polymerase (Bioline). Primers c and f of Taberlet et al. (1991)
were used to amplify the trnL-trnF region, and reaction mixtures were incubated at 94°C for 3 min, then subjected to 25 cycles of the following profile: denaturation for 30 s at 94°C, annealing for 30 s at 52 or 55°C, and extension for 30 s at 72°C. After cycling, a final 5-min incubation at 72°C was performed. The primers GN1 (Scott and Playford, 1996
) and ITS4 (White et al., 1990
) were used to amplify the ITS region. The cycling profile differed from that used for trnL-trnF amplifications only in the use of a 55°C annealing temperature and a 1-min extension step.
Two distinct ITS amplification products were obtained for Elaeocarpus sp. ‘Rocky Creek'. Products from one individual from each of the two known populations of this species were cloned by using the pGEM–T Easy Vector System (Promega, Madison, Wisconsin, USA) in accordance with the manufacturer's protocol. Four bacterial colonies from each individual were selected for amplification of the inserted DNA via PCR by using the M13F and M13R primers supplied (Promega).
PCR products were purified by using Jetquick columns (Genomed, Bad Oeynhausen, Germany) in accordance with the manufacturer's protocol. Each 20-µL sequencing reaction mixture contained approximately 60 ng of purified PCR product, 1 µL of BigDye version 3.1 (ABI, Foster City, California, USA), 3.5 µL of CSA buffer (ABI) and purified water (to 20 µL) and was subjected to 25 cycles of 10 s at 96°C, 5 s at 50°C, and 4 min at 60°C, after which a final 1 min incubation at 60°C was performed. Sequencing products were purified by ethanol precipitation and visualized using model 3730 capillary sequencing machines (ABI). Electropherograms were viewed and edited, and consensus sequences for each sample were assembled using BioEdit 7.0.1 (Hall, 1999
).
Sequence alignment and phylogenetic analysis
Sequence alignment of the trnL-trnF region was relatively straightforward. Manual alignment of the sequences was performed with the program PAUP* version 4.0b10 (Swofford, 2000
). Gaps were positioned to maximize conformity to known indel types, such as simple and inverted duplications of adjacent sequences (Levinson and Gutman, 1987
; Golenberg et al., 1993
). Inferred parsimony-informative indels were scored as unordered two-state characters. Indels of the same length and position and showing minor differences in nucleotide sequence were scored as the same state (Simmons and Ochoterena, 2000
). The gap states in the sequence matrices were treated as missing data in the phylogenetic analyses.
ITS region sequences showed extensive length variation among the ingroup taxa. Alignment was performed using the program ClustalW (Thompson et al., 1994
) as implemented in the program BioEdit 7.0.1 with subsequent manual improvement based on the "simple" method of Simmons and Ochoterena (2000)
. Several areas of the alignment, however, showed considerable ambiguity, and these areas were excluded from phylogenetic analysis.
Parsimony analyses were performed using PAUP* 4b10 (Swofford, 2000
). All nucleotide positions were treated as independent, unordered characters of equal weight (Fitch, 1971
). Heuristic searches were conducted using 1000 random taxon-addition-order replicates with tree-bisection-reconnection (TBR) branch swapping, and only 100 best trees per replicate were saved. Branch lengths were derived under ACCTRAN optimization. Clade support was assessed by bootstrap analysis (Felsenstein, 1985
) with 1000 heuristic search pseudoreplicates using the same search parameters as those in the parsimony analysis. The complete trnL-trnF analysis was limited to 300 trees saved per replicate due to memory overrun.
An alternative estimate of the phylogeny, the posterior probability, was obtained using the Markov chain Monte Carlo (MCMC) implemented in MrBayes 3.0b4 (Huelsenbeck and Ronquist, 2001
). Four Markov chains were started simultaneously from random trees. The most general likelihood model was used: the number of substitution types (nst) was set to 6, and among-site rate variation was modelled by the gamma distribution (rates=gamma) with four rate categories (ngammacat=4). All other priors for the analysis were set flat (i.e., as Dirichlet priors). One million generations were performed with a tree sampled every 100 generations. Trees generated prior to the four Markov chains reaching stationarity (the burn-in) were discarded. The remaining trees were used to construct a 50% majority rule consensus tree in PAUP*. Although studies have shown Bayesian methods can overestimate support for nodes (Suzuki et al., 2002
; Cummings et al., 2003
), we considered such an analysis useful here to provide a model-based estimate of the phylogeny for a data set too large to feasibly conduct a standard maximum likelihood analysis.
Molecular evolutionary rates and divergence time estimation
Minimum rates of molecular evolution across ingroup lineages were investigated. First, we determined whether nucleotide substitutions were accumulating at a rate that did not diverge significantly from constancy, by using the likelihood ratio test implemented in the program r8s version 1.70 (Sanderson, 2003
). Log-likelihood scores were generated for one of the optimal trees (using the model parameters determined by Modeltest analysis) with both the molecular clock enforced and unenforced. The test statistic, defined as twice the difference in the negative of the log likelihoods, is distributed as the 
2, with the degrees of freedom equal to the number of taxa minus two. Tests were conducted on both the entire data set and with the former Tremandraceae removed. Because the data exhibited rate constancy once the former Tremandraceae were removed, estimates of absolute divergence times of lineages on the phylogeny were obtained by using the Langley–Fitch "local molecular clock" method in r8s (Sanderson, 2003
). This method allows different evolutionary rates to be applied to different lineages, which is appropriate if departure from a global molecular clock can be localized to specific lineages. The number of rate categories was set to two (the command nrates=2 was used), one rate for the descendents of the node (and its subtending stem: stem=YES) defining the MRCA (most recent common ancestor) of the former Tremandraceae and the other for the remaining taxa. Among-site rate variation was modelled by the gamma distribution with shape parameter value of one (rates=GAMMA, shape=1.0), and multiple starts from different conditions were undertaken to minimize the likelihood of entrapment in local optima (num_time_guesses=10, num_restarts=10). The gradient of the solution of the objective function was evaluated (checkgradient=YES). A cross-validation analysis (Sanderson, 2002
) was performed (crossv=YES) to assess the suitability of the analytical model and parameters.
External calibration was based on estimated dates for the origin of Elaeocarpaceae and the Elaeocarpus-Platytheca split, as determined by nonparametric rate smoothing (NPRS) analysis of a combined rbcL+atpB+18S sequence data set for 560 species of angiosperms (Wikström et al., 2001
). Using a single, fossil-based calibration point (84 million yr ago [Ma] for the Fagales-Cucurbitales split) Wikström et al. (2001)
estimated the age of Elaeocarpaceae to be 59 Ma under ACCTRAN optimization, 56 Ma under DELTRAN optimization, and 57 Ma from a maximum likelihood (ML) estimate, with the standard error estimated by bootstrap at ±4 Ma. Elaeocarpus and Platytheca were estimated to have diverged 32 (ACCTRAN), 42 (DELTRAN), and 41 (ML) Ma (±4 Ma). Although there are sound reasons to avoid using divergence dates derived by molecular dating as calibration points for new studies (Magallón, 2004
), we have employed such dates as calibrations only to provide a comparison to internal, fossil-based calibration.
Internal calibration was conducted by using a fossil-based estimate of 30 Ma for the minimum age of the genus Elaeocarpus. Fossil mesocarps (previously referred to in the literature as endocarps; Dettmann and Clifford, 2001
) of Elaeocarpus are distinctive (woody, loculicidally dehiscent, with a smooth or sculptured surface, carpels 2–9, locules 1-seeded, locules fertile and sterile [often compressed], ovules anatropous, pendulous with a ventral raphe etc.; Dettman and Clifford, 2001), abundant and widespread in Tertiary floras of eastern Australia, beginning in the early part of the Oligocene (33.9–23.0 Ma; time scale from Gradstein and Ogg, 2004
). Three or five types of fossil mesocarps are recognized representing up to 17 species (Rozefelds and Christophel, 1996b
; Dettmann and Clifford, 2001
). Four of the five types defined by Dettmann and Clifford (2001)
are recorded from the early Oligocene: types 1 and 3 from Queensland and Tasmania, and types 1, 2, and 5 from South Australia. Representatives of all four of these groups that contain extant species (groups 1–4) were included in the molecular trees used for divergence time analysis. Because these fossil mesocarps exhibit the putative synapomorphies for Elaeocarpus and can be assigned to subgroups within the genus on the basis of mesocarp morphology (see Dettman and Clifford, 2001), they can be confidently ascribed to the crown group of Elaeocarpus (rather than the stem group) and their ages used as minimum estimates of the age of the crown group node (Magallón, 2004
). The ages of the sediments that contain Elaeocarpus mesocarps cannot be determined more precisely than early Oligocene; therefore, it seemed reasonable to use 30 Ma as a conservative estimate of the minimum age of the crown group. Older, Eocene-aged fossil leaves have also been assigned to Elaeocarpus (Ettingshausen, 1883
, 1886
; Chapman, 1935
), or at least compared with it (Christophel and Greenwood, 1987
; Christophel et al., 1987
; O'Dowd et al., 1991
; Pole, 1993
), but they cannot be used with confidence because leaves of certain Cunoniaceae and Elaeocarpaceae are difficult to distinguish (Pole, 1996
; Barnes et al., 2001
). Fossil pollen referable to, or compared with Elaeocarpus or Elaeocarpaceae have been reported from the Late Eocene (Macphail et al., 1994
) onward (Blackburn and Sluiter, 1994
; Kershaw et al., 1994
), but these assignments are also doubtful (Dettmann and Clifford, 2001
; Rozefelds and Christophel, 2002
).
The 95% confidence limits associated with the divergence dates were determined using a bootstrap procedure (Sanderson, 2003
). Divergence time analyses were conducted on trees of fixed topology but with branch lengths estimated using each of 100 bootstrapped data sets generated with SEQBOOT (Felsenstein, 2004
). Much of the subsequent file merging was facilitated by scripts written by T. Eriksson (available from the web site http://www.bergianska.se/index_forskning.php).
Additional analyses were conducted in which the maximum age for the family was set at 124 Ma (in addition to the 30 Ma fixed calibration point for Elaeocarpus). This corresponds to the earliest record of tricolpate pollen grains in the Barremian of southern England (Hughes and McDougall, 1990), which is probably a good estimate of the age of the eudicots (Crane et al., 1989
; Anderson et al., 2005
), within which Elaeocarpaceae are nested.
RESULTS
trnL-trnF analysis
The aligned trnL-trnF data set comprised 1327 positions, of which 189 were potentially parsimony-informative. Missing data mostly occurred near the 3' end of the 3' trnL exon and the adjacent part of the trnL-trnF spacer in the outgroup taxa (these data were not available from GenBank). The sequence for Connarus conchocarpus was 52% complete, because the trnL-trnF spacer was not sequenced, and Cephalotus follicularis was 79% complete. The other outgroup taxa were at least 87% complete. Among the ingroup, Sloanea australis was 91% complete and Aceratium ferrugineum 94% complete, whereas other ingroup sequences were at least 99% complete. Alignment required 76 gaps, of which 36 were potentially parsimony-informative and added as binary characters to the matrix. The mean base frequencies across the data set were A = 0.36 (range 0.35–0.41), C = 0.17 (0.15–0.18), G = 0.17 (0.15–0.18), T = 0.30 (0.26–0.32).
Parsimony analysis of the trnL-trnF data including indel characters yielded 481 trees of length (L) 457 steps (consistency index [CI] excluding uninformative characters = 0.70, retention index [RI] = 0.92 and rescaled consistency index [RC] = 0.72). The four Markov chains reached stationarity after approximately 22 000 generations (data not shown). The strict consensus of the parsimony trees and posterior probabilities shows strong support for monophyly of the Elaeocarpaceae (bootstrap percentage, BP = 88%; posterior probability, PP = 99), including a strongly supported Tremandraceae clade (100; 100), which is placed in a derived position within Elaeocarpaceae, sister (96; 100) to a clade comprising Aceratium, Elaeocarpus, and Sericolea (the Elae-Trem clade, 86; <50; Fig. 1). Most currently recognized genera of Elaeocarpaceae s.l. for which more than one species was sampled are supported albeit some only weakly: Aceratium (63; 83), Aristotelia (99; 100), Crinodendron (88; 100), Dubouzetia (67; <50), Peripentadenia (94; 100), Sericolea (85; 100), Sloanea (99; 100), Tetratheca (100; 100), and Tremandra (100; 100). The largest genus, Elaeocarpus, is not supported: most sampled taxa form a large weakly supported (59; <50) polytomy with Aceratium. One unnamed species from northern New South Wales (Australia) tentatively assigned to Elaeocarpus (E. sp. ‘Rocky Creek') is placed outside of the Aceratium-Elaeocarpus clade in the parsimony analysis. Bayesian analysis does not distinguish this taxon from a cluster comprising all sampled species of Aceratium, Elaeocarpus, and Sericolea.
|
ITS analysis
The length of the ITS region sequences ranged from 518 bp (Elaeocarpus hookerianus) to 633 bp (Sericolea gaultheria) in Elaeocarpaceae s.l. The majority of the sequences were complete, with only Sloanea woollsii and Sericolea calophylla subsp. grossiserrata incomplete at the 5' end (by approximately 51 and 56 bases, respectively) and E. hookerianus and E. angustifolius incomplete at the 3' end (by 98 and 81 bases respectively). Initial analyses indicated that the two paralogues of E. sp. ‘Rocky Creek' branched sequentially as sisters to a clade comprising members of Aceratium, Elaeocarpus, Sericolea, and former Tremandraceae (results not shown). The branch lengths of the two paralogues differed by a factor of about four (ACCTRAN optimization). The sequence with the shorter branch length was selected to represent the species in the analyses because its rate of molecular evolution is comparable to that across the remainder of the ingroup. Only a single ITS product was amplified in all other samples, and none had branch lengths that were vastly disproportionate to the remainder, with the exception of the former Tremandraceae as a group, which had a similar pattern for the single-copy plastid locus trnL-trnF. Therefore we are confident that ITS homologues were sequenced in all taxa.
The complete aligned data set comprised 849 positions, of which 228 were parsimony-informative. The mean base frequencies across the data set were A = 0.17 (range 0.15–0.20), C = 0.29 (0.25–0. 31), G = 0.33 (0.27– 0.36), T = 0.21 (0.19–0.25). Numerous gaps were required and of these, 10 were of sufficiently certain homology to be coded as characters. Parsimony analysis that included the indel characters yielded 16 trees (L = 918, CI = 0.54, RI = 0.75, RC = 0.49). The strict consensus tree (Fig. 2) is consistent with the results of the trnL-trnF analysis in resolving all genera as monophyletic and separating Elaeocarpus sp. ‘Rocky Creek' from Elaeocarpus. Support for the majority of genera is strong (Aceratium [BP = 97, PP = 100], Aristotelia [100; 100], Crinodendron [95; 100], Peripentadenia [98; 100], Sericolea [99; 100], Sloanea [100; 100], Tetratheca [99; 100]), but Dubouzetia and Elaeocarpus are not supported by bootstrap values >50% (albeit these genera receive posterior probabilities of 99 and 87, respectively). The pattern of relationships among the lineages differs from the trnL-trnF analysis in that Crinodendron and Peripentadenia form a clade, as do Aceratium and Sericolea. Neither clade receives bootstrap support; however, Aristotelia and Vallea are sisters (100; 100).
|
Relationships within Elaeocarpus are resolved in both parsimony and Bayesian analyses, but the patterns of relationships from these analyses conflict. The Bayesian analysis strongly supports E. arnhemicus as sister to the clade comprising E. angustifolius and E. williamsianus (not shown), whereas the parsimony analysis weakly places the latter as sister to all other Elaeocarpus (excluding E. sp. ‘Rocky Creek').
Combined analyses
The combined data set including the indel characters comprised 2222 positions of which 333 were parsimony-informative. Only the 35 taxa for which both trnL-trnF and ITS sequences were available were analyzed. Parsimony analysis yielded a single tree (L = 1097, CI = 0.57, RI = 0.80, RC = 0.54; Fig. 3). Relationships in this tree are compatible with those in the ITS consensus tree alone, but with greater resolution and generally stronger support. In particular, the Crinodendron + Peripentadenia clade now receives bootstrap support >50% (56) and Dubouzetia + Elae-Trem clade is robust (99, 100). Relationships among species of Elaeocarpus were not resolved, nor were their relationships to the genera Aceratium and Sericolea.
|
There is a marked difference in the estimates of minimum divergence times obtained using the three calibration points for most nodes between the internal (Elaeocarpus crown group = 30 Ma) and external (Wikström et al., 2001
) calibrations. For example, the age of the MRCA of Crinodendron and Peripentadenia was estimated at 73–81 Ma on the basis of the internal calibration (Table 1) and at 42–46 Ma on the basis of the external calibration (not shown). When both calibrations were used together, the age of this node was estimated to be an intermediate 46–50 Ma (not shown), which is closer to the estimate based on the external calibration. Most other nodes showed a similar difference in the estimates; on the whole, the external calibration gave dates which were about 35–45% younger than the internal calibration for the ITS tree, and 45–55% younger for the combined tree.
|
DISCUSSION
Origin of the former Tremandraceae
The placement of the former Tremandraceae as sister to a derived lineage comprising species of Aceratium, Elaeocarpus, and Sericolea improves upon recent molecular analyses that have suggested a close relationship between the two families (Savolainen et al., 2000
; Soltis et al., 2000
; Bradford and Barnes, 2001
) and have led to the incorporation of Tremandraceae into Elaeocarpaceae (APG, 2003
; Coode, 2004
). Comparative studies of floral morphology in these families and relatives (Matthews and Endress, 2002a
, b) inspired by phylogenetic recircumscription of Oxalidales have identified a number of features shared by Elaeocarpaceae s.l., many of which are associated with the buzz pollination syndrome characterizing its members. These features include flowers pendant, anthers basifixed on short filaments and dehiscing poricidally, and punctiform stigmas (Matthews and Endress, 2002b
). Further evidence for a placement of Tremandraceae in Elaeocarpaceae comes from corolla aestivation. In Tremandraceae, Crinodendron, and Elaeocarpus, the corolla aestivation is induplicately valvate (cochlear in Vallea, cochlear or quincuncial in Aristotelia, variable in Sloanea) with the petals enclosing part of the adjacent stamens in bud (Matthews and Endress, 2002b
).
The former Tremandraceae comprise a distinctly dry-adapted group of sprawling to shrubby plants with small, stiff leaves. Some species of Tetratheca are leafless, a feature that is particularly common among species inhabiting rocky outcrops in the semi-arid zone (Thompson, 1976
; Alford, 1995
). That this lineage has arisen within a group of rainforest trees and shrubs provides an opportunity to study the adaptation of elements of the endemic Australian flora to the Tertiary aridification, which resulted in the development of a scleromorphic fire- and drought-adapted flora and the retreat of the previously extensive rainforests to pockets mostly along the ranges of the eastern margin of the continent. This progressive aridification was probably due to two phenomena: (1) the initiation of the Antarctic circumpolar current (Kemp, 1978
; Truswell and Harris, 1982
), which developed after the separation of Antarctica from Australia and South America (Florindo et al., 2003
) and resulted in a steepening of the latitudinal temperature gradient and cooler conditions in Australia, and (2) the rafting of Australia into the subtropical high-pressure ridge, which blocks rain-bearing lows (Bowler, 1982
; Frakes, 1999
). Our range of minimum divergence time estimates (Fig. 4) support the origin of the former Tremandraceae lineage between about 64 Ma (stem node age, date of divergence of Elaeocarpus and Tremandra) and 37 Ma (crown node age, date of divergence of Tremandra from the remainder). These ages range from the early Palaeocene to the early Oligocene, a time interval during which Australia experienced relatively warm, wet conditions that allowed the development of extensive meso-megathermal rainforests similar to extant lowland tropical rainforests to occur (Christophel and Greenwood, 1988
; Hill, 2004
). Although such conditions would not be expected to favor the evolution of a scleromorphic flora (such as the former Tremandraceae), scleromorphy began to appear in the fossil record in the middle Eocene in Western Australia and may have been a response to oligotrophic soils rather than aridity (Hill et al., 1999
; Hill, 2004
).
|
).
Adaptation to drier conditions among Elaeocarpaceae is not restricted to the former Tremandraceae. Several species of Dubouzetia have colonized relatively dry and exposed habitats, including the New Caledonian radiation comprising the closely related D. acuminata, D. campanulata, D. caudiculata, D. confusa, and D. guillauminii (restricted to maquis vegetation on ultrabasic soils; Tirel, 1982
) and the two Australian cliff-dwelling species D. australiensis and D. saxatilis (Bean and Jessup, 1997
). Several species of Elaeocarpus are also found in relatively dry environments, e.g., E. alaternoides in maquis in New Caledonia (Tirel, 1982
) and E. arnhemicus in dry sclerophyll woodlands in northern Australia and southeastern New Guinea (Coode, 1984
, 2001
).
The former Tremandraceae show markedly elevated molecular evolutionary rates compared to those of their relatives—the alternative interpretation that rates slowed in the other groups is less parsimonious, requiring at least two instances of rate change to explain the observed data. Several explanations for such discrepancies in molecular evolutionary rates exist. Those explanations that have been invoked for observed positive correlations between molecular evolutionary rate and either biologically available energy (higher in the tropics than temperate zones, e.g., Davies et al., 2004
; Wright et al., 2006
) or diversification rate (e.g., Jobson and Albert, 2002
) are not applicable for Elaeocarpaceae s.l. This is because the former Tremandraceae (c. 54 species), a subtropical and temperate group, are considerably less species rich than their sister group, Elaeocarpus + Aceratium + Sericolea, which comprises more than 330 mostly tropical species.
Differences in generation times are a third explanation for discrepancies in molecular evolutionary rates (e.g., Laroche and Bousquet, 1999
; Andreasen and Baldwin, 2001
). Indeed, some Tetratheca species may flower within 9 months of germination post-fire (personal observation, DMC and MR), whereas to our knowledge no species of any other genus of Elaeocarpaceae s.l. achieves a generation time approaching this duration. However, a broad-scale, phylogenetically independent comparative study failed to identify any correlation between generation time and evolutionary rate in angiosperms (Whittle and Johnston, 2003
), and therefore a causal link in Elaeocarpaceae must remain purely speculative.
The disparities between the divergence date estimates based on our internal calibration and the external one (Wikström et al., 2001
) is interesting in view of the results of a detailed analysis of the asterids (Bremer et al., 2004
) using six reference fossils, which showed that Wikström et al. (2001)
underestimated the ages of many asterid nodes by 10–20 Ma. In Elaeocarpaceae, the Wikström et al. (2001)
dates conflict with the fossil evidence. The age of Elaeocarpus is estimated at 15–18 Ma, whereas five types of Elaeocarpus mesocarps are present in Oligocene (33.9–23.0 Ma) sediments in eastern Australia (Dettmann and Clifford, 2001
).
Systematics of Elaeocarpaceae
Our analysis supports the majority of current generic concepts in Elaeocarpaceae s.s. based on previous phylogenetic estimates. Aristotelia and Vallea form a robust group, with Sloanea, which is often considered isolated in Elaeocarpaceae (Coode, 1987
), resolved as its sister. Coode (1985)
proposed a phylogeny of Aristotelia and Vallea based on 11 (mostly reproductive) morphological characters. In a manually constructed cladogram, Aristotelia was monophyletic with the Australian species (A. australasica and A. peduncularis), forming a clade sister to a group comprising the other three species; in the latter group, the New Zealand taxa (A. fruticosa, A. serrata) formed a clade. Our results conflict with this set of relationships in that only the monophyly of A. serrata and A. fruticosa is supported by both studies.
Among the remainder of the ingroup, some support is found in the trnL-trnF analysis for the successive divergence of Crinodendron, Peripentadenia, and Dubouzetia from the rest. The relative order of the first two was not supported by analysis of ITS data (Fig. 2), but the combined analysis strongly suggests that they are sisters (Fig. 3). The idea of a close relationship between these three genera dates back to Miers (1868)
who grouped Crinodendron (as Tricuspidaria) and Dubouzetia in subtribe Tricuspidarieae. Some later workers agreed the two are related (Sprague, 1907
), and Baillon (1873)
suggested combining them. Coode (1987)
was the first to analyze the relationships of Peripentadenia (described by Smith, 1957
) to the other two genera and noted several features shared by all three (namely leaves always penninerved as opposed to three-nerved from the base, seeds fewer than ovules as opposed to about as many as ovules, fruit dehiscent with tough valves, and a persistent columella as opposed to dehiscent with not very woody valves and no columella). Upon cladistic analysis of the morphological database (which included these characters), Coode (1987)
discovered a monophyletic group comprising all three genera (and a sister relationship between Crindodendron and Dubouzetia), but of the shared features only the fruit type is a unique synapomorphy for this clade (Coode, 1987
). A later discovery of two whorls of stamens in these three genera might indicate additional support (Bricker, 1991
), but further occurrences of this rather obscure character state have not been rigorously sought across the rest of the family. The present molecular phylogeny conflicts with the morphological one and suggests that the fruit type and staminal arrangement were present in the ancestor of all Elaeocarpaceae excluding Aristotelia, Sloanea, and Vallea, reversing to the plesiomorphic states in the clade comprising Aceratium, Elaeocarpus, and Sericolea. Comparative studies of fruit morphology of the former Tremandraceae and Crinodendron, Peripentadenia and Dubouzetia would be useful for further evaluating this hypothesis.
Relationships among species of Elaeocarpus were not resolved in this study. There is some information, however, on the systematics of the putative new species of Elaeocarpus, E. sp. ‘Rocky Creek'. All analyses indicate that this species is not closely related to the other species of Elaeocarpus sampled, instead tending to be placed sister to the clade that includes Aceratium, Elaeocarpus, and Sericolea. Elaeocarpus sp. ‘Rocky Creek' bears a striking similarity to a species from New Guinea, E. blepharoceras (Coode, 1984
; Maynard, 2004
). Further work is needed to clarify the affinities of this entity, but if its position is confirmed then two taxonomic options are available: to recognize this species (together with E. blepharoceras, at least) as a new genus of Elaeocarpaceae, or to synonymize Aceratium and Sericolea under Elaeocarpus. Morphologically, there is little to differentiate Elaeocarpus sp. ‘Rocky Creek' and E. blepharoceras from other species in the genus, apart from having unusually dense radial fibers in the mesocarp and deeply furrowed fibrous bark (M. Coode, personal communication; Maynard, 2004
). Interestingly Elaeocarpus sp. ‘Rocky Creek' and E. blepharoceras are widely disjunct geographically: the former inhabits subtropical rainforests in northeastern New South Wales, whereas the latter is found over 3000 km to the north in lowland and upland rainforests in New Guinea (M. Coode, personal communication).
Historical biogeography of Elaeocarpaceae
Coode's (1985) phylogeny of Aristotelia based on morphology has been used as data for event-based biogeographic analysis of the southern hemisphere (Sanmartín and Ronquist, 2004
). In that study Aristotelia was interpreted as an example of the "inverted southern pattern" of biogeographic distribution. This pattern has New Zealand and southern South America as sister areas, with Australia sister to that group, and Africa sister to all of them. In the present work, however, the phylogeny resolved for Aristotelia is an example of the "plant southern pattern." This pattern, which differs from the inverted southern pattern in that the positions of Australia and southern South America are reversed, is consistent with the general plant area cladogram resolved with 19 published phylogenies (Sanmartín and Ronquist, 2004
). For Aristotelia, the minimum estimates of divergence times allow an evaluation of the biogeographic process giving rise to this pattern. A dispersal scenario is more probable than vicariance because the divergence of the South American and New Zealand lineages at 24–27 and 3 Ma respectively, postdates the isolation of their respective landmasses (c. 35–30.5 and c. 80 Ma; McLoughlin, 2001
). An additional divergence time analysis using a single, fixed calibration of 80 Ma on the node separating the New Zealand lineage from the Australian sister species A. australasica yielded an age for the origin of the family of approximately 550 Ma, earlier than the origin of land plants (results not shown). This is strong evidence against a vicariance explanation. This further example of a recent long-distance dispersal to New Zealand adds to a growing body of evidence supporting the notion of dispersal having had a much greater influence on extant distributions in the southern Hemisphere than is often recognized (Pole, 1994
; Winkworth et al., 2002
; Cook and Crisp, 2005
; de Queiroz, 2005
).
Crinodendron comprises five (Coode, 1987
) or four (Bricker, 1991
) species that are allopatric in extratropical South America. The two Chilean species were used in this study, the southern C. hookerianum and the northern C. patagua. The divergence of these species is estimated to have occurred during the Paleo-Eocene (Table 1, Fig. 4), but the origin of the genus is almost certainly older given the position of D. brasiliense (from dwarf cloud forest near the Atlantic coast of Brazil) as sister to the rest of the genus, based on morphological data (Coode, 1987
).
The genus Peripentadenia comprises two species of large, buttressed rainforest trees, which are estimated to have diverged 18–19 Ma (Table 1). Both species are endemic in the Wet Tropics World Heritage Area in northeastern Queensland (Australia) but have different altitude preferences: P. mearsii is most often found above c. 500 m a.s.l. on the Atherton Tableland, whereas P. phelpsii is found in lowland rainforest in the Mossman area.
Peripentadenia and Crinodendron diverged 73–91 Ma. Their respective landmasses, Australia and South America, have been isolated from Antarctica since the early Oligocene (about 35 Ma), well after the lineages diverged. Therefore, vicariance is the most plausible explanation for the biogeographical relationships among the extant taxa.
Dubouzetia comprises 11 species (Coode, 1987
; Bean and Jessup, 1997
) distributed from the Moluccas to New Caledonia and northern Australia. For this study, species from the three main parts of the extant distribution of the genus (New Guinea, Australia, New Caledonia) and two of the three sections (sects. Dubouzetia and Oligovula; Coode, 1987
; Bean and Jessup, 1997
) were sampled. The sister relationship between the New Guinea species (D. kairoi) and the Australian species (D. saxatilis) is not surprising as these two landmasses were connected frequently throughout the Quaternary, doubtless affording repeated opportunities for terrestrial migration between them. However, the estimated date of 23–26 Ma for the divergence of D. kairoi and D. saxatilis predates the mid Miocene collision between Australia and Asia. The divergence of D. caudiculata (New Caledonia) from the New Guinea-Australia clade is estimated to have occurred 40–44 Ma. New Caledonia, as part of Tasmantia, split off from Australia about 80 Ma (McLoughlin, 2001
), indicating that transoceanic dispersal is the most likely explanation for the presence of the genus on New Caledonia and New Guinea-Australia. Transoceanic dispersal is apparently possible for Dubouzetia, because D. elegans (including D. novoguineensis A.C. Sm. at varietal rank) is known from New Guinea and New Caledonia (Tirel, 1982
; Coode, 1987
), probably having its origins on the former landmass. As noted by Coode (1987)
, evidence for this view includes the fact that D. elegans is unique among New Caledonian Dubouzetia species in not being restricted to ultramafic soils and in being cladistically distant from the other New Caledonian species (it is nested in sect. Oligovula sister to the New Guinea species D. dentata). This clade in turn is sister to another New Guinea species, D. kairoi. If dispersal is the favored hypothesis, then the geographic origin of the genus can be sought. However, the current database contains only one species from each of the three main areas of distribution of the genus, and therefore the ancestral state at the root of the area cladogram cannot be determined. Phylogenetic placement and divergence dating of the remaining seven species of Dubouzetia would be required to more fully understand the historical biogeography of this genus.
Divergence time analysis indicates that the E. sp. ‘Rocky Creek' lineage diverged from the most recent common ancestor shared with Aceratium, Elaeocarpus, and Sericolea, between 46–47 Ma (mid Eocene), a time when Australia was still connected to Antarctica and had well-developed meso-megathermal rainforests in the southern part (Hill, 2004
), which were similar to the extant rainforests of the wet tropics of north Queensland (Christophel and Greenwood, 1988
).
Divergence time analysis across multiple independent lineages has the power to elucidate general evolutionary and historical biogeographic patterns. A recent study incorporated divergence time analysis of several prominent Australian sclerophyll groups (Banksia, Allocasuarina, Eucalyptus, Fabaceae tribe Mirbelieae) together with published molecular phylogenies of other plant groups represented in Australia to test correspondence of diversity patterns over time in aseasonal wet biome (rainforest and wet heath), sclerophyll biome and Eremean biome (arid zone) groups with historical climatic changes (Crisp et al., 2004
). The results suggest rapid radiation in these sclerophyllous groups at the time of cooling and drying in the Australian climate (25–10 Ma). At the same time, aseasonal wet biome groups (Araucariaceae, Podocarpaceae, Nothofagus) seem to have failed to radiate or were depleted by extinction, based on phylogenetic (having sister taxa that are species rich either outside Australia or in the sclerophyll biomes) and paleontological (greater species diversity in the past) arguments. However for Elaeocarpus, an aseasonal wet biome group, extant diversity in Australia is not substantially lower than its dry-adapted relatives (the former Tremandraceae). Twenty-five extant species (Coode, 1984
) plus at least five undescribed species (Hyland et al., 2003
; Maynard, 2004
) of Elaeocarpus are known from Australia. If Aceratium is included (which is only doubtfully distinct from Elaeocarpus on morphological grounds and not distinct based on trnL-trnF data), then there are 36 Australian species. About 54 species of former Tremandraceae are known; therefore radiation in the lineage represented by Elaeocarpus and Aceratium does not seem to have been substantially constrained relative to its sclerophyll biome sister. To identify the most influential factors in the radiation of these related but very distinct lineages (e.g., life history traits, dispersal potential), we have initiated in-depth phylogeographic investigations within Elaeocarpus and Tetratheca. The results of the present study provide the necessary broad phylogenetic and historical biogeographic context for these studies.
Fossil species diversity in Elaeocarpus appears to be lower than extant species diversity. Between 11 and 17 species of fossil Elaeocarpus mesocarps are recognized in the mid Cenozoic of southern Australia (Rozefelds and Christophel, 1996a
, b, 2002; Dettmann and Clifford, 2001
), fewer than the 30 extant Elaeocarpus species found in Australia. However, comparing fossil species diversity with extant species diversity is problematic because the fossil flora is undoubtedly not as well sampled as the extant flora (even though the fossil record for Elaeocarpus is relatively good), and fossil morphotypes may represent a number of distinct species. For example, mesocarps of the approximately eight species in the E. reticulatus group are difficult to distinguish (A. Rozefelds, Tasmanian Herbarium, personal communication). Nevertheless, it does not appear that Elaeocarpus has undergone a similarly major decline in species diversity in the aseasonal wet biome in Australia as have other comparable groups (e.g., Nothofagus). This relative success may be a consequence of in situ speciation, but it is possible that there was significant interchange between the Australian and Malesian Elaeocarpus floras, particularly in the light of the high dispersibility of some Elaeocarpus species as determined by analyses of gene flow (Rossetto et al., 2004
; Whitehead, 2004
). Although the fossil record does not identify any major influx of taxa to Australia from Asia after the mid Miocene collision, there are acknowledged biases in this record, notably against rainforest taxa (Truswell et al., 1987
). Divergence time analysis suggests that the extant species of Elaeocarpus had arisen by about 15 Ma, although low taxon sampling in this clade undoubtedly has led to these dates being overestimated. As Crisp et al. (2004)
point out, their sampling of aseasonal wet biome groups is strongly biased toward relictual taxa (Nothofagus, Podocarpaceae, Araucariaceae). Our study of a group with relatively high extant diversity and a good macrofossil record does not support a major decline in species diversity after Miocene rainforest contraction. This highlights the importance of broad sampling across phylogenetic and ecological groups. Clearly, more chronograms of rainforest groups are required before a comprehensive picture of historical biogeography and evolution of the Australian aseasonal wet biome flora can emerge.
APPENDIX
Taxa used in the phylogenetic analyses, family assignments (indicated for the first species of each genus only), GenBank accessions (a dash indicates the genomic region was not sampled), the geographical origin of the sample and voucher details. Voucher specimens are deposited in the following herbaria: BRI = Queensland Herbarium, JCT = James Cook University, Townsville, Queensland, K = Kew, MEL = National Herbarium of Victoria, MO = Missouri Botanical Garden, NCU = University of North Carolina, NSW = National Herbarium of New South Wales, PERTH = Western Australian Herbarium, QRS = CSIRO Atherton, Queensland, UWA = University of Western Australia.
Taxon—Family; GenBank accessions: trnL-trnF, ITS; Geographic origin of the sample; Voucher specimen. Abbreviation: s. l. = sensu lato.
Aceratium concinnum (S. Moore) C.T. White, Elaeocarpaceae s. l., DQ444678, DQ448684, NE Australia, D.M. Crayn 858 et al. (NSW). Aceratium ferrugineum C.T. White, DQ444681, DQ448685, NE Australia, M. Harrington 296 (JCT). Aceratium megalospermum (F.Muell.) Balgooy, DQ444679, —, NE Australia, D.M. Crayn 523 (NSW). Aceratium sericoleopsis Balgooy, DQ444680, —, NE Australia, D.M. Crayn 779 (NSW). Aceratium ledermannii Schltr., DQ444677, DQ448683, New Guinea, D.M. Crayn 534 et al. (NSW). Acsmithia elliptica (Vieill. ex Pamp.) R.D. Hoogland, Cunoniaceae s. l., AF299179 and AF299232, —, New Caledonia, J. Bradford 613 (MO). Aristotelia australasica F. Muell., Elaeocarpaceae s. l., DQ444661, DQ448661, SE Australia, J.M. Allen s.n. (NSW 605725). Aristotelia chilensis Stuntz, DQ444660, DQ448660, Chile, J.M. Allen s.n. (NSW 605486). Aristotelia fruticosa Hook.f., DQ444662, DQ448662, New Zealand, M.W. Chase 781 (K). Aristotelia peduncularis (Labill.) Hook.f., DQ444659, DQ448664, Tasmania, L. Mulcahy s.n. (NSW 606884). Aristotelia serrata Oliv., DQ444663, DQ448663, New Zealand, J.M. Allen s.n. (NSW 605729).
Bauera rubioides Andrews, Cunoniaceae s. l., AF299183 and AF299236, —, SE Australia, J. Bradford 730 (MO). Bauera sessiliflora F. Muell., AF299184 and AF299237, —, SE Australia, J. Bradford 729 (MO). Brunellia colombiana Cuatrec., Brunelliaceae, AF299181 and AF299234, —, Colombia, J. Bradford 753 (MO).
Cephalotus follicularis Labill., Cephalotaceae, AF299193 and AF299246, —, SW Australia, T.D. Macfarlane 2549 (PERTH). Ceratopetalum gummiferum Sm., Cunoniaceae s. l., AF299176 and AF299229, —, SE Australia, J. Bradford 873 (MO). Connarus conchocarpus F.Muell., Connaraceae, AF365035, —, NE Australia, Wannan 1500 (NSW). Crinodendron hookerianum C. Gay, Elaeocarpaceae s. l., DQ444666, DQ448674, S Chile, J.M. Allen s.n. (NSW 605484). Crinodendron patagua Molina, DQ444665, DQ448673, N Chile, J.M. Allen s.n. (NSW 605483). Cunonia capensis L., Cunoniaceae s. l., AF299156 and AF299209, —, S Africa, J. Bradford 735 (MO).
Davidsonia pruriensvar. jerseyana F.M. Bailey, Cunoniaceae s. l., AF299185 and AF299238, —, E Australia, J. Bradford 887 (MO). Dubouzetia campanulata Pancher ex Brongn. & Gris, Elaeocarpaceae s. l., DQ444667, —, New Caledonia, D.M. Crayn 745 (NSW). Dubouzetia caudiculata Sprague, DQ444668, DQ448675, New Caledonia, G. McPherson 3305 (NSW ex MO). Dubouzetia kairoi Coode, DQ444670, DQ448677, New Guinea, D.M. Crayn 578 (NSW). Dubouzetia saxatilis A.R. Bean & L.W. Jessup, DQ444669, DQ448676, NE Australia, D. Silke s.n. (NSW622671, QRS).
Elaeocarpus arnhemicus F. Muell., Elaeocarpaceae s. l., DQ444690, DQ448690, N Australia, P.H. Weston 599 et al. (NSW). Elaeocarpus angustifolius Blume, DQ444689, DQ448689, New Guinea, R.J. Johns 10685 (K). Elaeocarpus bancroftii F. Muell., DQ444685, DQ448687, NE Australia, D.M. Crayn 502 (NSW). Elaeocarpus culminicola Warb., DQ444688, —, NE Australia, D.M. Crayn 499 (NSW). Elaeocarpus eumundi F.M. Bailey, DQ444682, —, E Australia, D.M. Crayn 505 (NSW). Elaeocarpus foveolatus F. Muell., DQ444691, —, NE Australia, P.D. Hind 6265, J. Holland (NSW). Elaeocarpus hookerianus Raoul, DQ444686, DQ448688, New Zealand, J.M. Allen s.n. (NSW 605721). Elaeocarpus kirtonii F.M. Bailey, DQ444687, —, E Australia, D.M. Crayn 501 (NSW). Elaeocarpus largiflorens C.T. White subsp. retinervis B. Hyland & Coode, DQ444684, DQ448686, NE Australia, D.M. Crayn 503 (NSW). Elaeocarpus reticulatus Sm., DQ444683, —, E Australia, J.M. Allen s.n. (NSW 605722). Elaeocarpus sericopetalus F. Muell., DQ444692, —, NE Australia, G. Fensom 401 (NSW). Elaeocarpussp. ‘Rocky Creek', DQ444676, DQ448682, E Australia, D.J. Maynard 2 (NSW). Elaeocarpus williamsianus Guymer, DQ444693, DQ448691, E Australia, D.M. Crayn 513 (NSW). Eucryphia moorei F. Muell., Cunoniaceae s. l., AF299174 and AF299227, —, SE Australia, J. Bradford 860 (MO).
Geissois superba Gillespie, Cunoniaceae s. l., AF299166 and AF299219, —, Fiji, Fortune Hopkins 5019.
Peripentadenia mearsii (C.T. White) L.S. Sm., Elaeocarpaceae s. l., DQ444672, DQ448679, NE Australia, P. Forster PIF 29760 (BRI). Peripentadenia phelpsii B. Hyland & Coode, DQ444671, DQ448678, NE Australia, D.M. Crayn 887 (NSW). Platylophus trifoliatus D. Don, Cunoniaceae s. l., AF299177 and AF299230, —, S Africa, Goldblatt 10888 (MO). Platytheca galioides Steetz, Elaeocarpaceae s. l., DQ444694, —, SW Australia, A.N. Rodd 4973, G. Fensom (NSW).
