| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Systematics and Phytogeography |
2 Department of Biological Science, Florida State University, Tallahassee, Florida 32308 USA 3 Royal Botanic Gardens and Domain Trust, Mrs. Macquaries Road, Sydney, New South Wales, 2000, Australia
Received for publication 21 December 2007. Accepted for publication 29 April 2008.
ABSTRACT
Tribe Macadamieae (91 spp., 16 genera; Proteaceae) is widespread across the southern hemisphere on all major fragments of Gondwana except New Zealand and India. Macadamia is cultivated outside its natural range as a "nut" crop (notably in Hawaii, where it is the principal orchard crop). We sampled seven DNA regions and 53 morphological characters from the tribe to infer its phylogeny and address the common assumption that the distribution of the extant diversity of the tribe arose by the rafting of ancestors on Gondwanan fragments. Macadamia proves to be paraphyletic with respect to the African genus Brabejum, the South American genus Panopsis, and the Australian species Orites megacarpus. We erect two new generic names, Nothorites and Lasjia, to produce monophyly at that rank. The earliest disjunctions in the tribe are inferred to be the result of long-distance dispersal out of Australia (with one possible exception), rather than vicariance. Evolution of tardy fruit dehiscence is correlated with these dispersals, and the onset of the Antarctic Circumpolar Current (ACC) precedes them. We suggest that the ancestors of extant diversity arrived on their respective continents via the ACC, and we recognize that this is a mechanism precluded, rather than facilitated, by Gondwana’s terrestrial continuity.
Key Words: biogeography • fruit evolution • Gondwana • hydrochory • long-distance dispersal • Macadamia • relaxed-clock molecular dating • phylogenetics • Proteaceae • vicariance
"I have reason to suppose that no well informed naturalist will claim that the dispersal of the Proteaceae [and other groups] is in any way recent." —Léon Croizat (1962
, p. 173)
Macadamia F.Muell. and relatives in tribe Macadamieae (91 spp. in 16 genera; Fig. 1) are the most widespread group recognized at tribal rank in the Proteaceae (Macadamia nut family; ca. 1800 spp. in 80 genera). Today, the tribe naturally occurs on all the major landmasses thought to have once been part of the southern supercontinent Gondwana (Fig. 2), with the exception of India and New Zealand (though it can be found in the fossil record of New Zealand; Carpenter, 1994; Pole, 1998). The tribe also can be found on landmasses that are not derived from Gondwana, such as Central America and parts of the Southeast Asian mainland. Macadamia (M. integrifolia Maiden & Betche, M. tetraphylla L.A.S.Johnson, and their hybrids; Fig. 1A) and other members of the tribe [Athertonia diversifolia (C.T.White) L.A.S.Johnson & B.G.Briggs, Gevuina avellana Molina, Hicksbeachia pinnatifolia F.Muell.; Fig. 1B] are cultivated outside of their native range for their edible, fleshy embryos. At the generic rank, the tribe is most diverse in Australia, where six small (1–9 spp. each), narrowly distributed genera occur in the eastern rainforests and adjacent regions. At the specific rank, the tribe is more diverse in South America, where two larger genera (Panopsis Salisb. and Euplassa Salisb.; 25 and 20 spp., respectively) and one small genus (Gevuina Molina; 1 sp.) naturally occur. Fossils attributed to the family extend back to the late Cenomanian (ca. 93 million years before present [Ma BP]; Dettmann and Jarzen, 1998), prior to most fragmentation events in the southern supercontinent Gondwana. The family’s far-flung distribution in the southern hemisphere is generally explained by the rafting of ancestors on those fragments (e.g., Venkata Rao, 1971; Johnson and Briggs, 1975; Weston and Crisp, 1994, 1996; Prance and Plana, 1998; Prance et al., 2007; cf. Barker et al., 2007), as is the distribution of many other organisms (reviewed in Sanmartín and Ronquist, 2004). Here, we infer a phylogeny for the tribe, the areas occupied by ancestors, the dates of biogeographic disjunctions among the ancestors, and the significance of correlations between disjunctions and the evolution of the tribe’s fruits to address the veracity of this explanation for the tribe’s widespread distribution.
|
|
, in which the tribe and four subtribes are circumscribed based on clades resolved in a supertree summary of phylogenies produced for the family to date. This supertree does not resolve the relationships among the subtribes and resolves just two relationships among genera within subtribes: Brabejum L. as sister to Panopsis and Cardwellia F.Muell. as sister to a clade composed of the remaining genera in subtribe Gevuininae. Three of the phylogenies that they used for the supertree were published (Hoot and Douglas, 1998; Mast and Givnish, 2002; Barker et al., 2002), and two were unpublished (one data set from the internal transcribed spacers of the nuclear ribosomal DNA (ITS) and another from the rbcL gene). These previously unpublished ITS data are, in part, published here. In Weston and Barker’s (2006) supertree, clades corresponding to the four subtribes of tribe Macadamieae are part of a polytomy with a 17th genus, Carnarvonia F.Muell., but they consider Carnarvonia to be incorrectly placed based on its morphology. Orthotropous ovules, tardily dehiscent fruits, unwinged seeds, and fleshy cotyledons are usually present in members of tribe Macadamieae, but all are missing from Carnarvonia (and Cardwellia, but molecular results strongly support the position of Cardwellia in the tribe; Fig. 1D). Some or all of these features are also seen in Floydia L.A.S.Johnson & B.G.Briggs, Roupala Aubl., and Lambertia Sm.—three genera that Johnson and Briggs (1975) included in tribe Macadamieae but Weston and Barker (2006) moved to tribe Roupaleae. This character state distribution makes tribes Macadamieae and Roupaleae difficult to diagnose from each other morphologically. Previously published phylogenetic work on the family sampled the atpB gene and atpB-rbcL spacer (Hoot and Douglas, 1998) or these two regions in combination with the rbcL gene (Barker et al., 2007) from one species each of Macadamia, Brabejum, Panopsis, Cardwellia, Euplassa (not in Barker et al., 2007), and Gevuina and resolved these as a monophyletic group. This prior sampling represents two of four subtribes in tribe Macadamieae.
Synapomorphies have been recognized for Weston and Barker’s (2006) subtribes Gevuininae (8 genera), Macadamiinae (3 genera) and Virotiinae (3 genera), but not for subtribe Malagasiinae (2 genera). Subtribe Gevuininae is composed of genera from South America (Euplassa, 20 spp.; Gevuina, 1 sp.), Australia-New Guinea (Hicksbeachia F.Muell., 2 spp.; Bleasdalea F.Muell. ex Domin, 2 spp.; Cardwellia, 1 sp.), New Caledonia (Kermadecia Brongn. & Gris, 4 spp.; Sleumerodendron Virot, 1 sp.), and Fiji and Vanuatu (Turrillia A.C.Sm., 3 spp.; Fig. 2). The orientation of the carpel and the production of floral zygomorphy through curvature of the style and curvature of three of four tepals are synapomorphies for subtribe Gevuininae (Douglas and Tucker, 1996a; Weston and Barker, 2006), though the last two character states have undergone reversals in some members. Subtribe Macadamiinae is composed of Macadamia (9 spp.) from Australia and Sulawesi, Brabejum (1 sp.) from southern Africa, and Panopsis (25 spp.) from South and Central America. A cuplike nectary surrounding the ovary (thought to have arisen independently in Virotia L.A.S.Johnson & B.G.Briggs) and frequent opposite or whorled phyllotaxy are synapomorphies for subtribe Macadamiinae (Johnson and Briggs, 1975; Weston and Barker, 2006). Subtribe Malagasiinae is composed of Malagasia L.A.S.Johnson & B.G.Briggs (1 sp.) from Madagascar and Catalepidia P.H.Weston (1 sp.) from Australia. A morphological synapomorphy for subtribe Malagasiinae has yet to be identified. Subtribe Virotiinae is composed of Heliciopsis Sleumer (14 spp.) from Burma and southeast China to Malesia (northwest of Wallace’s Line), Virotia (6 spp.) from New Caledonia, and Athertonia (1 sp.) from Australia. Distinctive surface sculpturing of the woody inner mesocarp is a synapomorphy for subtribe Virotiinae (Weston and Barker, 2006). Five of the six New Caledonian species that Johnson and Briggs (1975) assigned to Virotia were placed there informally; all six had been previously treated as members of Macadamia (e.g., Virot, 1968). Although they have mostly been treated since 1975 as members of Virotia (e.g., Weston and Crisp, 1996; Weston, 2006; Weston and Barker, 2006), the new combinations for them in that genus have never been validly published. We find that group to be monophyletic (though with poor statistical support), and thus we treat those five New Caledonian species as members of Virotia and make the necessary combinations.
Fossil evidence from the large and widespread subtribe Gevuininae provides a minimum age for the basal split in that subtribe that predates the last continental connections between some, but not all, of the landmasses that it occupies. Fossil cuticle that has been attributed to subtribe Gevuininae appears in the Early Miocene deposits of the Manuherikia Group of Central Otago, New Zealand (Carpenter, 1994; Pole, 1998), and in the Eocene deposits of the Pidinga Formation in the Lefroy paleodrainage of southern Western Australia (Carpenter and Pole, 1995). The fossil cuticle shares derived features of the trichome base ("a thickened, round platform-like base with a crimped margin"; Carpenter and Pole, 1995, p. 1113) with genera other than Cardwellia in the subtribe. However, both fossils occur in locations not occupied by extant Gevuininae. Macphail (personal communication in Carpenter and Pole 1995, p. 1108) considered the samples taken from the Pidinga Formation to have a maximum age of Middle Eocene (lower boundary of 48.6 Ma BP) and a minimum age of Late Eocene (upper boundary of 33.9 Ma BP; we take all absolute geological ages from Gradstein et al., 2004). This time is prior to the end of continental connections between Australia and Antarctica (via Tasmania and the South Tasman Rise) at ca. 33 Ma BP (Veevers, 2000, 2001; J. J. Veevers, Macquarie University, personal communication) and between Antarctica and South America between 33 and 29 Ma BP (Florindo et al., 2003). Fossils attributable to a second subtribe (Virotiinae) are known from this period of Antarctic isolation. Fossil fruits that share the synapomorphy of an ornamented inner mesocarp with extant members of subtribe Virotiinae are known from the Oligocene deposits of the Glencoe locality in central eastern Queensland (Rozefelds, 1992) and the Miocene deposits of the Gulgong locality in New South Wales (Von Mueller, 1879, 1883; Rozefelds, 1992).
While these fossils predate or co-occur with the period of Antarctic isolation, most of the other fragmentation events involving Gondwana predate the fossils by substantial time spans. Continental connections between a combined Madagascar/India and the rest of Gondwana, in this case via eastern Antarctica, ended by ca. 132 Ma BP (Lawver et al., 1992; McLoughlin, 2001), and continental connections between Africa and the remaining contiguous part of Gondwana, in this case via South America, ended by ca. 105 Ma BP (McLoughlin, 2001). Continental connections between Zealandia (New Caledonia, New Zealand, and a number of now submerged of ridges and plateaus) and the remaining contiguous part of Gondwana, in this case via western Antarctica, began separating 84 Ma BP (McLoughlin, 2001) but might have remained contiguous in places until the Early Paleocene (62–66 Ma BP; Ladiges and Cantrill, 2007). Barker et al. (2007) found in their molecular dating study of the family (the study in which they sampled just five of the 16 genera in tribe Macadamieae) that inferred ages for the older disjunctions involving New Zealand or Africa were inconsistent with a vicariance hypothesis, whereas those for younger disjunctions (e.g., between Australia and South America) were consistent with it.
If the tribe and the landmasses that it occupies did indeed "co-speciate" (as in the vicariance framework commonly accepted for the family) and the habitats and suite of available dispersers on the daughter landmasses are analogous, then we would expect that evolution in the tribe’s diaspores will be uncorrelated with the disjunctions. The genera of tribe Macadamieae produce early or tardily dehiscent fruits (dehiscing at fruit maturity or seed germination, respectively) that are mostly >1 cm in size (Filla, 1926; Sleumer, 1955a; Virot, 1968; Venkata Rao, 1971; Johnson and Briggs, 1975; Weston, 1995a–1995d; Weston and Crisp, 1996; Prance and Plana, 1998; Qiu and Weston, 2003; Prance et al., 2007; Fig. 1). With the exception of Cardwellia, these fruits contain 1–2 wingless seeds. Cardwellia produces follicles that each open to release many thin, winged seeds (Hyland, 1995). The fruits are typically inconspicuous—most species have dull fruits that are black, brown, bronze, green, or gray at maturity. However, more apparent red fruits occur in subtribes Malagasiinae (the Australian genus Catalepidia) and Gevuininae (the Australian genus Hicksbeachia). The mesocarp in subtribes Gevuininae (with a few exceptions), Malagasiinae, and Virotiinae is succulent (Fig. 1C), and it is leathery to woody in subtribe Macadamiinae. There is a strongly lignified and bony inner mesocarp in subtribes Gevuininae, Malagasiinae, and Virotiinae, but not in subtribe Macadamiinae. Instead, in some members of Macadamia there is a bony testa—one of the hardest in the angiosperms (Venkata Rao, 1971). Johnson and Briggs (1975) considered the production of indehiscent fruits and large wingless seeds to have arisen independently at least four times in subfamily Grevilleoideae, each time in lineages inhabiting moist closed forests. They contend that in those habitats wind dispersal was lost in favor of dispersal by fruit-eating mammals (including fruit bats) and birds (cf., Venkata Rao, 1971). Weston and Crisp (1996) consider the fruits of Virotia, Kermadecia, Turrillia, and Sleumerodendron to be typical bat fruits, with their dull color, sour or mildly sweet odor, lack of a protective rind, and possession of large hard parts.
In this study, we infer a phylogeny for Macadamia and relatives in the widespread tribe Macadamieae using molecular and morphological data. Then, we infer the geographic distributions of ancestors, the age of biogeographic disjunctions, and the significance of correlations between disjunctions and features of the diaspores. We use the results to test the hypotheses that (1) Macadamia and other genera, including Virotia, are monophyletic; (2) tribe Macadamieae and the component subtribes of Weston and Barker (2006) are monophyletic; (3) the most recent common ancestor of extant members of tribe Macadamieae occupied Australia, the area of greatest extant generic diversity; (4) biogeographic disjunctions involving Gondwanan fragments date back to the time of the last continental connection between the landmasses involved; and (5) evolution of the tribe’s diaspores are uncorrelated with biogeographic disjunctions. We also use the results to make nomenclatural changes to maintain the monophyly of genera.
MATERIALS AND METHODS
Sampling
We sampled three chloroplast DNA (cpDNA: matK, atpB, and ndhF genes) and three nuclear DNA (nDNA; waxy loci 1 and 2 and PHYA genes) regions from 22 taxa of tribe Macadamieae and 10 taxa from outside the tribe (Appendix 1). The 22 taxa include 7 (of 9) species of Macadamia and 1 species from each of the remaining genera in the tribe (Appendix 1). The 10 extratribal taxa include 8 other representatives of subfamily Grevilleoideae: Carnarvonia araliifolia F.Muell., Banksia serrata L.f., Floydia praealta (F.Muell.) L.A.S.Johnson & B.G.Briggs, Grevillea caleyi R.Br., Lambertia formosa Sm., Orites diversifolius R.Br., O. megacarpus A.S.George & B. Hyland, and Roupala montana. Johnson and Briggs (1975) considered tribe Macadamieae to include Floydia, Lambertia, and Roupala Aubl., but the more recent classification of Weston and Barker (2006) excluded these genera from the tribe. Orites diversifolius is the type species of genus Orites R.Br., and it and O. megacarpus were included in this study because a parallel study (A. Mast and P. Weston, unpublished data) in which all species of Orites were sampled suggested that O. megacarpus (and only this species from that genus) might be nested in tribe Macadamieae. All four of the tribes of subfamily Grevilleoideae sensu Weston and Barker (2006) are represented in this sampling. The topology is shown with Nelumbo lutea Willd. (Nelumbonaceae) as sister to everything else, including the remaining taxon, Platanus occidentalis L. (Platanaceae). This topoglogy is based on previous molecular phylogenetic results (e.g., Angiosperm Phylogeny Group, 2003), which are supported by morphological synapomorphies for a clade composed of Proteaceae and Platanaceae (Hoot and Douglas, 1998).
We sampled the two internal transcribed spacers of the nuclear ribosomal DNA region and the 5.8S gene (ITS) from 31 taxa of tribe Macadamieae, two accessions of O. megacarpus, and Carnarvonia (Appendix 1). The 31 taxa included 8 (of 9) species of Macadamia, 4 (of 25) species of Panopsis, 3 (of 20) species of Euplassa, 3 (of 4) species of Kermadecia, 2 (of 6) species of Virotia, and 1 species from each of the remaining genera. All but one of the 22 species of tribe Macadamieae from which the non-ITS DNA regions were sampled are represented in the ITS data set. In the case of the remaining species, Heliciopsis lanceolata (Koord. & Valeton) Sleumer, it was replaced by its congener, H. lobata (Merr.) Sleumer, in the ITS data due to sequencing difficulties. Seven of 21 species represented in both the non-ITS and ITS data sets are represented by a different specimen (Appendix 1). This agglomeration of data from different specimens occurred because most of the ITS data set was generated several years before the non-ITS data sets were generated, and leaf tissue and DNA from some specimens had been used up. Carnarvonia is shown sister to the other taxa sampled in the ITS data set because this topology is consistent with the results from the other data sets. Specimen voucher information for the DNA samples is in Appendix 1.
We sampled morphological characters from 41 taxa of tribe Macadamieae and the same eight extratribal taxa sampled from subfamily Grevilleoideae for the non-ITS DNA data (Appendix 1). The 41 taxa included 9 (of 9) species of Macadamia, 5 (of 25) species of Panopsis, 2 (of 20) species of Euplassa, 4 (of 4) species of Kermadecia, 6 (of 6) species of Virotia, 3 (of 3) species of Turrillia, and 1 species from each of the remaining genera. All the taxa of tribe Macadamieae from which DNA data were sampled are represented in the morphological data set.
DNA extraction, amplification, cloning, and sequencing
We extracted the DNA used in the non-ITS DNA sequencing and in a small number of ITS sequencing reactions with a DNeasy Plant Mini Kit (Qiagen, Valencia, California, USA) following the manufacturer’s instructions. We extracted DNA used in most of the ITS sequencing using the CTAB-diatomite protocol of Gilmore et al. (1993; a typographical error in Appendix 1B of that reference resulted in all "µL" abbreviations being replaced with "mL" at steps 4, 9, 11, and 12).
We amplified the target regions using the polymerase chain reaction (PCR; Mullis et al., 1986). To amplify the matK, atpB, and waxy genes, we used the primers provided in Mast et al. (2005). We amplified the ndhF gene in two reactions: the 5' end with primer –52pro (a modified version of an unpublished primer of K.-J. Kim and R. K. Jansen; 5' AGG TAA GAT CCG GTG AAT CGG AAA C 3') and 972Rpro (a modified version of the 972R primer of Olmstead and Sweere, 1994; 5' CAT AGT ATA ACC CAA TTG AGA C 3') and the 3' end with 803Fpro (a modified version of the 803F primer of Olmstead and Sweere, 1994; 5' CTA TGG TAG CAG CGG GAA TTT TTC 3') and 2100R (Olmstead and Sweere, 1994) or 972Fgre (a modified version of the 972F primer of Olmstead and Sweere, 1994; 5' TAC AAT GTC TCA ATT GGG TTA TAT TAT G 3') and 2100Rgre (a new primer; 5' CTT GTA ACA CCA ATA CCA TTC GTA ATT C 3'). We amplified the PHYA gene with the new primers 34F (5' CTC CAA TCA TAC AAA CTT GCT GCC AAG G 3') and 1159R (5' CCT TCC AAG GTA AAC TCC TTG TCT TAA C 3'). We amplified the ITS region using the PCR primers of Barker et al. (2002) or Leu 1 (sequence provided in Mast, 1998) and 4 (White et al., 1990). The cpDNA and ITS primers amplify the complete extent of their respective regions. The waxy primers amplify parts of exons 7 and 10 and the intervening introns and exons; the PHYA primers amplify part of exon 1.
We used Platinum PCR Supermix (Invitrogen, Carlsbad, California, USA) to amplify the cpDNA regions, the diagnostic region of waxy (described later), and ITS (from a few of the taxa). We used Platinum PCR Supermix High Fidelity (Invitrogen; a mix that includes a polymerase with proofreading 3' to 5' exonuclease activity) to amplify the nDNA regions that we cloned. We used nonproofreading Taq polymerase (Promega, Annandale, New South Wales, Australia and Boehringer Mannheim, Castle Hill, New South Wales, Australia) to amplify ITS from the majority of the taxa. We separated amplified PCR products on an agarose gel and imaged them using a Bio-Rad Transilluminator with Bio-Rad Quantity One version 4.1.1 Geldoc software (Bio-Rad, Hercules, California, USA) or a Vilber Lourmat TF-20M Transilluminator (Vilber Lourmat, Marne-la-Vallée, France) to determine their concentrations for cloning (in the case of the nDNA regions) and sequencing. We enzymatically cleaned the non-ITS and a few ITS PCR products produced for sequencing with the ExoSAP-IT protocol (Qiagen) following the manufacturer’s instructions. We cleaned the majority of ITS PCR products using Wizard DNA Clean-up or PCR Preps Systems (Promega) using the manufacturer’s instructions.
We cloned two PCR products that represented three nDNA regions. We amplified waxy 1 and2 in the same PCR, and then identified the locus inserted in each plasmid of screened colonies as locus 1 or 2 using the diagnostic PCR reaction described in Mast et al. (2005). We cloned PCR product for the PHYA and waxy regions with a TOPO TA cloning kit (Invitrogen), PCR 2.1 Topo vector, and One Shot TOP 10 chemically competent E. coli following manufacturer’s instructions. We spread 38 µL of X-gal (20 ng/µL, 760 ng) on Luria–Bertani agar plates supplemented with ampicillin (100 µg/mL) prior to plating the cells on the agar. The culture plates incubated at 37°C overnight for 14–16 h. We picked cells from two colonies per plate for PHYA and five colonies per plate for waxy and transferred the cells to PCR reagents that included the m13F and m13R primers (Invitrogen; located on the plasmid adjacent to the point of insertion of the original PCR product). We enzymatically cleaned one PHYA insert, one waxy 1 insert, and one waxy 2 insert of the correct size for sequencing.
The Florida State University’s Department of Biological Science DNA sequencing facility prepared the cycle-sequencing reactions for the non-ITS and a few of the ITS amplifications with the ABI PRISM Big Dye Terminators v3.1 Cycle Sequencing Ready Reaction Kit (Applied Biosystems, Foster City, California, USA). The primers that the facility used to sequence matK, atpB, and the 2 waxy genes are the same as those used in Mast et al. (2005). The facility used the same primers that we used for the initial amplification of ndhF to sequence that product, and it used the m13F and m13R primers to sequence the cloned PHYA product. The facility precipitated each sequenced product in ethanol and EDTA to remove excess dye terminators before running each sequence on an ABI Prism 3100 Genetic Analyzer (Applied Biosystems). We cycle sequenced the majority of the ITS amplifications using the ABI PrismTM DyeDeoxy Cycle Sequencing System (Applied Biosystems) using the same primers for sequencing as were used in Barker et al. (2002). We sent these reactions to the Sydney University and Prince Alfred Molecular Analysis Centre and Westmead Hospital for electrophoresis on an ABI 373 or 377 (Applied Biosystems). To detect mistakes and correct any uncertainties in the computer-generated sequence, we compared aligned trace-files in Sequencher versions 4.6 and earlier (Gene Codes, Ann Arbor, Michigan, USA).
Defining substitution characters
To determine the boundaries of each region, we compared the cpDNA sequences to the complete cpDNA sequence of Nicotiana tabacum L. (GenBank accession NC 001879), the waxy sequences to the complete waxy sequence of Solanum tuberosum L. (GenBank accession X58453), and the ITS sequences to the annotated sequences in Baldwin (1992). We aligned the coding regions to maintain the reading frame when the indels were in multiples of 3. We imported the aligned sequence data sets into the program PAUP* version 4.0b10 (Swofford, 2003) where they were formatted in the NEXUS format for phylogenetic inference in PAUP* and MrBayes version 3.1.2 (Ronquist and Huelsenbeck, 2003).
Coding morphological characters
We coded the morphological characters using the literature and specimens. Important literature sources included Bosser and Rabevohitra (1991), Cooper and Cooper (2004), Douglas (1996), Douglas and Tucker (1996a, 1996b), George (1981), Johnson and Briggs (1963, 1975), flora treatments in McCarthy (1995), McDonald and Ismail (1995), Makinson (2000), Prance et al. (2007), Sleumer (1955a, 1955b), Smith and Haas (1975), Van Steenis (1952), Strohschen (1986a, 1986b), Venkata Rao (1971), Virot (1968), De Vogel (1980), and Weston (2006). Specimens that served as sources for morphological data are given in Appendix 2. PAUP* calculated the consistency index (CI; the minimum number of state changes divided by the observed number of state changes on the tree; Kluge and Farris, 1969) for each character on the tree inferred with the combined molecular and morphological data. This tree does not include all of the taxa for which morphological data were coded because some of these are not represented in the molecular data sets.
Inference of phylogeny
We used the Akaike information criterion (AIC) in MrModelTest version 1.1b (available from J. A. A. Nylander, Uppsala University, Uppsala, Sweden) to select an adequately parameter-rich model of nucleotide substitution for each of the DNA regions. For the morphological data, we used the standard discrete model of MrBayes 3.1.2—a model based on the ideas of Lewis (2001). These models were then used for their respective partitions in the Bayesian analyses in MrBayes 3.1.2 (Ronquist and Huelsenbeck, 2003). We unlinked the sampling of state frequencies, substitution rates, the gamma shape parameter, and the proportion of invariant sites for each DNA partition; we linked branch lengths. MrBayes 3.1.2 spawns two Markov chain Monte Carlo (MCMC) runs, and we ran each for 5 x 106 generations (most runs) or 50 x 106 generations (only when all data were combined, sampling every 1000th generation. We determined that a burn-in of 1.25 x 106 generations was sufficient to get the average standard deviation of split frequencies below 0.01 (a threshold recommended by the authors of MrBayes) for the two runs of each analysis run to 5 x 106 generations. These numbers of MCMC generations also resulted in potential scale reduction factors (Gelman and Rubin, 1992) for all parameters within at least 2 units of 1 (typically within 0.1 of 1), as determined in MrBayes. We constructed a majority-rule consensus of the trees sampled after the burn-in period in PAUP* 4.0b10 (Swofford, 2003) to produce a tree where the posterior probabilities were 
95% for internal nodes. When the majority rule consensus trees differed between the two runs in the number of branches resolved, the less resolved tree is presented and discussed. We inferred the phylogeny with each DNA region individually, with the three cpDNA regions combined, with the six non-ITS DNA regions combined, with the morphological data alone, and with the seven DNA regions and the morphological data combined.
We also used PAUP* 4.0b10 (Swofford, 2003) to do parsimony bootstrapping of each data set individually and in the combinations mentioned and to find the most parsimonious trees for the morphological data. PAUP* searched for the most parsimonious trees for the original morphological data and for the resampled data using a full heuristic search, tree-bisection-reconnection branch swapping, 10 random addition replicates, and, for the bootstrapping only, the maximum number of saved trees set at 10000. The frequency that branches appeared in most parsimonious trees of resampled data were calculated from 100 replicates.
Inference of the origin of disjunctions
We used two optimization methods to determine where biogeographic disjunctions arose on the phylogeny. The first, Fitch parsimony (i.e., unordered parsimony), is a method that minimizes dispersal events among regions and does not reconstruct widespread ancestral lineages that would imply the maintenance of genetic interchange across multiple areas (and water barriers, if the interchange comes at a time when the areas are separated by these). We interpret the inferred dispersal events to co-occur with cladogenetic events, because inferring these to occur along the branches implies additional events (subsequent extinction of each source population’s lineage on the original continent). The Ancestral State Reconstruction Packages version 1.06 (Maddison and Maddison, 2005) in Mesquite version 2.0 build i71 (Maddison and Maddison, 2007) inferred dispersal events on the tree. The second method that we used to determine where biogeographic disjunctions might have occurred is dispersal–vicariance analysis (DIVA; Ronquist, 1997). This method permits widespread ancestors and infers them prior to vicariance events to explain allopatric daughter lineages. The software DIVA version 1.1a (Ronquist, 1997) inferred dispersal and vicariance events using this method. We treated South and Central America as one area, given the young age of the Isthmus of Panama (ca. 3 Ma; Coates and Obando, 1996) and thus its inability to play a part in dispersal scenarios over most of the timeframe considered here. Panopsis, the genus found in Central America, is more diverse in South America and has only two (of 25) species endemic to Central America. We used the tree inferred with the seven DNA regions and the morphological data for the biogeographic reconstructions.
Inference of ancestral states and test of correlations between morphological change and the origin of disjunctions
We used two optimization methods to determine where morphological characters changed state on the phylogeny. The Ancestral State Reconstruction Packages for Mesquite (Maddison and Maddison, 2005) in Mesquite (Maddison and Maddison, 2007) inferred character state changes using Fitch parsimony on the tree topology inferred with the seven DNA regions and the morphological data. We explored the alternative resolutions of polytomies, when the alternative resolutions could affect the results. The Ancestral State Reconstruction Packages also calculated the proportional likelihood of the data with alternative states at each node on the chronogram inferred with the six non-ITS DNA regions. We had Mesquite use a one-parameter model of equal gain and loss (Lewis, 2001), unless the asymmetrical two-parameter model of gain and loss resulted in a significantly better likelihood, as determined using a likelihood ratio test (Goldman, 1993) with one degree of freedom. We compared the proportional likelihoods for the data with alternative states at each node using a threshold of significance set at 2, equivalent to the threshold of a log likelihood ratio of 7.4:1 advocated by Edwards (1972).
We determined the statistical significance of correlations between the origins of disjunctions and the following diaspore characters: lateral position of the conflorescence (cauliflorous or not; a modified version of character 16), exocarp color (red or not; a modified version of 41), fruit dehiscence (42), outer mesocarp texture (43), inner mesocarp texture (45), testa thickness and texture (50), and seed wing (present or not; a modified version of 51). The Correl Package version 0.1 (Midford and Maddison, 2006) in Mesquite calculated the statistical significance of correlations between dispersal events and each of these characters, in turn, on the chronogram inferred with the six non-ITS DNA regions. The Correl Package is an implementation of the likelihood method proposed by Pagel (1994). Correl calculated each p-value using 1000 simulations and the likelihoods using 10 iterations. For the purpose of this analysis the distribution of the extant species were coded as in Australia or not. This modification and those noted for characters 41 and 51 make characters with >2 states binary, so that the analysis can be performed. The modification of character 16 reduces the number of uncertainties for that character.
Inference of ages
We used Thorne’s Bayesian method (Thorne et al., 1998; Thorne and Kishino, 2002) to infer the age of divergence events using the six non-ITS DNA regions. Thorne’s approach, like the nonparametric rate smoothing and penalized likelihood approaches of S and erson (1997, 2002), relaxes the assumption of a constant rate of nucleotide substitution over time by modeling rate change with temporal autocorrelation (Gillespie, 1991). The rate in each branch is drawn from a lognormal distribution with a mean equal to the rate of its parent branch and a variance equal to the time between the midpoints of the two branches multiplied by a rate change parameter, 
(Thorne et al., 1998). Setting to 0 assumes constant rate change (i.e., a molecular clock). The method requires the user to specify a prior distribution for the rate at the root node, since this node does not have a parent branch. We explored the robustness of our results to alternative prior distributions for (described by brownmean and brownsd) and the rate at the root node (described by rtrate and rtratesd), as described later.
Yang’s (1997) baseml program in PAML version 3.13 generated estimates of parameters for the F84 nucleotide substitution model (Felsenstein, 1984) with a discretized gamma approximation of rate variation among sites for each of the six regions, in turn, using the tree topology inferred for the respective region. Thorne’s estbranches program estimated branch lengths and a variance-covariance matrix of branch lengths for each of the six regions, in turn, using the parameter values estimated in the previous step and the tree topology inferred in the combined analysis of all six regions. Thorne’s multidivtime program used a MCMC to approximate the posterior distribution of substitution rates and divergence ages from the branch lengths and variance–covariance matrix generated for each region in the previous step and the tree topology inferred in the combined analysis of all six regions. We ran the MCMC 106 generations, sampling every 100th after a quarter (2.5 x 105) of the generations. We ran the MCMC twice for each analysis to determine whether the parameter values appeared to have reached stationarity after the burn-in.
We used a mean of 1.15 for the prior distribution of time separating the most recent common ancestor (MRCA) of the ingroup (for this purpose, the MRCA of Platanaceae and Proteaceae) from the present (rttm). This value is the mean of two age estimates for that node in a broad-scale dating study of the eudicots by Anderson et al. (2005). The standard deviation of this prior distribution (rttsd) was set to 0.15, a span that includes the earliest fossils that can be assigned to the stem of the Platanaceae (described later). We used a mean of 0.039 for the prior distribution of the rate of molecular evolution at the ingroup root node (rtrate). This value is one-half of the median distance between Platanus L. and each taxon in the sister group (0.0888), as determined using the uncorrected-p parameter in PAUP* 4.0b10 (Swofford, 2003), divided by our prior estimate of time between the root of the ingroup and today (1.15). The standard deviation of this prior distribution (rtratesd) was set to 0.039 (equal to the mean; see Wiegmann et al., 2003 for a justification to making the standard deviation of this prior large relative to the mean). The "bigtime" parameter, a parameter that specifies what we consider to be a maximum age for the ingroup, was set at 1.3. This value is equivalent to 130 Ma, which is older than the first appearance of the eudicots in the fossil record (Magallón et al., 1999). The mean (brownmean) and standard deviation (brownsd) for the Brownian motion constant "nu" were both set to 1.0, as was the prior for the times of the interior nodes given the time of the root, following recommendations made by Thorne in the multidivtime readme file. We left other parameters dealing with proposals in the MCMC at their default values (newk = 0.1, othk = 0.5, thek = 0.5).
We used three constraints in the analyses. The maximum age for the MRCA of Proteaceae and Platanaceae was set at 125 Ma. This age corresponds to the approximate time of appearance in the fossil record of the tricoplate pollen of the eudicots (Magallón et al., 1999), of which Proteales is an early-diverging branch. It has been used elsewhere as a calibration point for the MRCA of the eudicots (e.g., Bell et al., 2005), and we think that this is a conservative approach given the hypotheses to be tested (i.e., more recent divergences falsify them). The minimum age for the MRCA of Proteaceae and Platanaceae was set at 100 Ma. This value is the upper boundary of the Albian (rounded up from 99.6 Ma; Gradstein et al., 2004), the earliest period from which fossil inflorescences have been recovered that can be assigned to Platanaceae (e.g., Crane et al., 1993) using shared derived features (reviewed by Crepet et al., 2004; Anderson et al., 2005). For most analyses, the minimum age for the MRCA of subtribe Gevuininae was set to 34, rounded up from 33.9, the upper boundaries of the Eocene (Gradstein et al., 2004). This age corresponds to a period to which fossils from one of the two clades arising with the basal split in Gevuininae have been assigned using shared derived features (Carpenter and Pole, 1995). We used the occurrence of subtribe Virotiinae in the Oligocene (23–33.9 Ma BP), as determined using the derived surface sculpturing of the inner mesocarp (Rozefelds, 1992; Weston and Barker, 2006) as a means of assessing the veracity of the results.
We explored the sensitivity of the inference of ages to assumptions made in the dating analysis by systematically varying the assumptions in three ways. First, we varied the age constraints by setting the minimum age of the third constraint (for the crown age of subtribe Gevuininae) to the maximum age estimate for the formation in which the fossil was found (Middle Eocene according to Macphail personal communication in Carpenter and Pole, 1995, p. 1108; 49 Ma, rounded from 48.6, Gradstein et al., 2004) and by removing the third constraint altogether. Second, we performed analyses with eight other combinations of rtrate and brownmean values (and the original three age constraints). These were the combinations of the rtrate values of 0.0039, 0.039 (the original value), and 0.39 and the brownmean values of 0.1, 1 (the original value), and 10, with the standard deviation of each prior set equal to the mean. This set of sensitivity tests is modeled after those performed by Wiegmann et al. (2003) and Bell et al. (2005). Third, we incrementally increased the root age prior (rttm), the minimum and maximum age of the root, and the bigtime value by 10 Ma, and made appropriate changes to the mean rate at root node prior (rtrate) and the standard deviation of the rate at the root node prior (rtratesd) using the calculation given earlier. We did not use the Gevuininae constraint for this third sensitivity analysis; all other assumptions were identical to our preferred set of assumptions. This third sensitivity analysis is designed to determine the effect that new fossil discoveries pushing back the age of the eudicots would have on our conclusions. The incremental changes were made over a range of rttm values from 115 Ma (our preferred prior) to 225 Ma (100 Ma before the first appearance of eudicots in the fossil record).
RESULTS
Defining substitution characters
We generated >250 kilobases of aligned data for this study. We describe each DNA region in Table 1. We used 8248 aligned nucleotide positions in total: 7702 in the non-ITS data sets and 546 in the ITS data set. We were unable to locate a colony with a waxy locus 1 insert for Orites megacarpus, Macadamia grandis C.L.Gross & B.Hyland, and M. ternifolia F.Muell., and we were unable to amplify the 5' end of ndhF from O. megacarpus. We were also unable to amplify any region other than ITS from the DNA of M. hildebrandii Steenis. We found the nDNA regions to have higher percentages of informative characters and greater maximum distances within the tribe than the cpDNA regions (Table 1).
|
|
|
95% posterior probability) for each cpDNA and nDNA region, the three cpDNA regions combined, and all six non-ITS regions combined. Figure 4 displays the 95% majority rule consensus of trees sampled after the burn-in for the ITS data alone, the morphological data alone, and the combined DNA and morphological data. Figure 4 also displays the strict consensus of the most parsimonious trees found for the morphological data.
|
|
95% posterior probability (PP) had parsimony bootstrap frequencies 70% (a commonly used cut-off). Eight branches had parsimony bootstrap frequencies between 50 and 70% (denoted with "<"), and five had parsimony bootstrap frequencies <50% (denoted with "<<"). Conversely, four branches in the tribe had parsimony bootstrap frequencies 70%, but did not have PP 95%. A sister relationship between Heliciopsis and Athertonia is supported by bootstrap frequencies 70% (but PP 
95%) in three analyses (PHYA alone, six non-ITS regions combined, and all data combined, in which the relationship had a PP of 90%, 89%, and 92%, respectively). A sister relationship between Macadamia hildebrandii and M. claudiensis C.L.Gross & B.Hyland is supported by a bootstrap frequency 70%, but a PP of only 85% in the analysis of all data in combination.
The tree inferred using the six non-ITS DNA regions in combination (Fig. 3) is well resolved, with just two points at which it is not bifurcating: a trichotomy in subtribe Virotiinae and a trichotomy in subtribe Gevuininae. Within the tribe, there is just one point of conflict between the branches resolved with 95% posterior probability in this tree and those resolved with 95% posterior probability with any DNA region individually (in the position of M. whelanii Bailey in the matK results; Fig. 3). A second point of conflict between the tree resolved with all six non-ITS DNA regions in combination and those generated with the data sets separately occurs among the extratribal taxa (in the position of Lambertia in the PHYA results). The tree inferred using the six non-ITS DNA regions in combination resolves a paraphyletic tribe Macadamieae and subtribe Macadamiinae with respect to O. megacarpus. Each of the other three subtribes is resolved as monophyletic. Floydia, Lambertia, and Roupala—the three genera that Johnson and Briggs (1975) considered to be included in tribe Macadamieae, but which Weston and Barker (2006) excluded from it—are more closely related to other extratribal taxa than to the taxa considered by Weston and Barker (2006) to be in the tribe. Cardwellia—the genus that Johnson and Briggs (1975) excluded from tribe Macadamieae, but which Weston and Barker (2006) included in it—is most closely related to other genera that Weston and Barker (2006) included in subtribe Gevuininae. The tree inferred using the six non-ITS DNA regions in combination also resolves a paraphyletic Macadamia with respect to O. megacarpus, Brabejum, and Panopsis. That result is seen in the separate analysis of half of the six DNA regions (all of the cpDNA regions); the results from the separate analysis of the nuclear DNA regions neither conflict with, nor support, the paraphyly of Macadamia with respect to those other taxa.
The tree inferred using the ITS data (Fig. 4) also resolves a paraphyletic tribe Macadamieae and subtribe Macadamiinae with respect to O. megacarpus. The two samples of that species are resolved as sisters in the ITS tree. Each of the other three subtribes are again resolved to be monophyletic. Four of the five genera from which multiple species were sampled (Panopsis, Virotia, Kermadecia, and Euplassa) are also resolved as monophyletic in the tree. The fifth genus, Macadamia, is resolved as two clades in a polytomy with Brabejum and a clade of Panopsis and O. megacarpus.
The tree inferred from the morphological data using the Bayesian approach contains just two branches with PP 95% (Fig. 4): one subtending a polytomy composed of Macadamia tetraphylla, M. ternifolia, M. integrifolia, and M. jansenii C.L.Gross & P.H.Weston and one subtending a polytomy composed of Kermadecia sinuata Brongn. & Gris, K. elliptica Brongn. & Gris, and K. rotundifolia Brongn. & Gris. The tree inferred from the morphological data using a parsimony approach has three branches with 70% bootstrap frequencies (Fig. 4): the two branches with PP 95% and that resolving a sister relationship between Euplassa inaequalis Engl. and E. duquei Killip & Cuatrec. The genera Heliciopsis and Virotia are each resolved as monophyletic in the strict consensus of the most parsimonious trees, but with only 64% and <50% bootstrap support, respectively. Floydia— a genus that Johnson and Briggs (1975) considered to be included in tribe Macadamieae, but which Weston and Barker (2006) excluded from it—is resolved to be closely related to some members of the tribe in parsimony analysis, but with <50% bootstrap support.
The tree inferred using all seven of the DNA regions in combination with the morphological data (Fig. 4) is congruent with the trees resolved for the six non-ITS DNA regions in combination (Fig. 3) and the ITS data alone (Fig. 4). The two polytomies within the tribe in the combined non-ITS DNA result (noted above) remain unresolved when all data are combined. Three branches with PP 95% in the combined data tree have parsimony bootstrap frequencies <50%. Two of these, the sister relationship between Bleasdalea and Hicksbeachia and the sister relationship between Banksia L.f. and many of the other outgroup taxa, had bootstrap frequencies 70% in the analysis of the six non-ITS DNA data sets in combination. The third, the sister relationship between Brabejum and O. megacarpus/Panopsis, had a bootstrap frequency between 50 and 70% in that analysis.
Inference of the origin of disjunctions
For the purpose of inferring the origins of geographic disjunctions and state shifts in morphological characters, we resolved two of the three trichotomies in the tree inferred using the combined molecular and morphological data (Fig. 4) in just one way. We resolved Heliciopsis and Athertonia as sister to one another, as supported by a bootstrap frequency of 81% and PP of 92% in analyses of the combined molecular and morphological data, and M. hildebrandii and M. claudiensis as sister, as supported by a bootstrap frequency of 70% and a PP of 85% in those same analyses. The final trichotomy within the tribe (that in subtribe Gevuininae) is not resolved with a PP >50% or bootstrap support 70% in analyses of the combined molecular and morphological data. Its resolution is relevant to the reconstruction of ancestral geographic distributions but not relevant to the ancestral states that we discuss in the next section.
Fitch parsimony and DIVA resolve geographic disjunctions arising at identical nodes, and because our dating analysis infers all but one of these nodes to be younger than the last continental connections between the landmasses involved, we discuss the Fitch parsimony results (a dispersalist approach) at greater length here. Fitch parsimony reconstruction of ancestral geographic distributions infers a distribution in Australia for the MRCA of tribe Macadamieae and the MRCA of each subtribe, with subsequent dispersals from Australia (Fig. 5A). In subtribe Macadamiinae, dispersal events from Australia to Sulawesi, South America, and Africa led to the origins of M. hildebrandii (and presumably M. erecta J.A.McDonald & Ismail R.), Panopsis, and Brabejum, respectively. In subtribe Malagasiinae, a dispersal event from Australia to Madagascar led to Malagasia. In subtribe Virotiinae, dispersal events from Australia to New Caledonia and Southeast Asia led to Virotia and Heliciopsis, respectively. The trichotomy in subtribe Gevuininae complicates the interpretation of inferred ancestral distributions and dispersal events. The inference is unambiguous in resolving dispersal events from Australia to New Guinea in Bleasdalea and from New Caledonia to Fiji and Vanuatu in the ancestors of Turrillia. However, explanation of the remaining disjunctions in the subtribe depends on resolution of the trichotomy (Fig. 5B). If Gevuina is later resolved as sister to Euplassa and relatives, then dispersal from Australia to South America with a subsequent dispersal to New Caledonia is most parsimonious. However, if the trichotomy is resolved in one of the other two possible ways, then two scenarios are equally parsimonious. In the one, there is an early dispersal to South America (at node 9; Fig. 6), with a dispersal back to Australia leading to Bleasdalea and Hicksbeachia; in the other, there are two later dispersals to South America leading to Gevuina and Euplassa and relatives (Fig. 5B). With these exceptions in subtribe Gevuininae, Fitch parsimony infers that all of the expansions of the tribe’s range occurred as a result of dispersal out of Australia. Nodes at which disjunctions are inferred to occur (or the earliest node at which they could occur, in the case of the polytomies) are labeled 1–9 in Fig. 6.
|
|
Inference of ancestral states and test of correlations between morphological change and the origin of disjunctions
Of the diaspore characters examined, three are significantly correlated with dispersals in the tribe at a threshold of P = 0.05, as determined using Midford and Maddison’s (2006) implementation of Pagel’s (1994) test of correlations between discrete characters. These three characters are fruit dehiscence at maturity vs. germination (character 42; P = 0.002), outer mesocarp texture dry vs. succulent (43; P = 0.022), and inner mesocarp texture leathery vs. strongly lignified and bony (45; P = 0.027; Fig. 5A). In the case of the first character, Fitch parsimony infers 4–5 state changes to later dehiscence (at germination) immediately preceding, or potentially co-occurring with, dispersal events to result in later dehiscence in every extra-Australian genus. Correl (Midford and Maddison 2006) calculated an instantaneous transition rate of 24.199 for dispersal from Australia when dehiscence occurs at germination compared to a rate of 0.282 for dispersal from Australia when dehiscence occurs at fruit maturity. In the case of the second character, Fitch parsimony infers a state change to a succulent outer mesocarp immediately preceding, or potentially co-occurring with, the dispersal events occurring in subtribes Malagasiinae, Virotiinae, and Gevuininae, but not Macadamiinae (Fig. 5A). The program Correl calculated an instantaneous transition rate of 10.848 for dispersal from Australia when the outer mesocarp is succulent compared to a rate of 2.369 for dispersal from Australia when the outer mesocarp is dry. In the case of the third character, alternative reconstructions are equally parsimonious. A bony mesocarp arose either on the branch leading to the MRCA of the tribe and the outgroup clade that includes Floydia, with one loss on the branch leading to the MRCA of subtribe Macadamiinae and two losses within that outgroup clade, or it arose independently on the branches leading to the MRCA of subtribes Malagasiinae and Virotiininae, to the MRCA of subtribe Gevuininae, to Lambertia, and to Banksia. The latter scenario is similar, though not identical, to the ancestral reconstruction for the outer mesocarp succulence within the ingroup (Fig. 5A). Correl calculated an instantaneous transition rate of 7.343 for dispersal out of Australia when the inner mesocarp is strongly lignified and bony compared to a rate of 2.639 for dispersal out of Australia when the inner mesocarp is leathery. Cauliflory (character 16) is inferred to have arisen three separate times—once in each of Virotia, Kermadecia, and Hicksbeachia (Fig. 1G). The former two genera were recognized as producing typical bat fruits by Weston and Crisp (1996); cauliflory is often associated with bat pollination and/or dispersal (van der Pijl, 1972). Exocarp color (41) is quite labile and is discussed later in the context of synapomorphies. The optimization of testa thickness (50) is ambiguous in Macadamiinae and is also discussed later in the context of synapomorphies. Winged seeds (51) are inferred to have arisen twice in the tribe, in O. megacarpus and Cardwellia (Fig. 6).
Inference of ancestral states for characters 42, 43, and 45 on the chronogram using a likelihood approach is congruent with Fitch optimization of character state changes on the tree topology inferred using the combined molecular and morphological data. The two-parameter model of gain and loss does not result in a significant improvement in the likelihood of the data as compared to its likelihood with a one-parameter model for any of these characters. The pattern of fruit dehiscence (character 42) in extant taxa has a higher proportional likelihood when the MRCA of tribe Macadamieae is constrained to have fruit that dehisce at maturity (proportional likelihood = 0.785; proportional likelihoods given unless otherwise noted) vs. dehiscing at germination (0.215), though this difference is not statistically significant with the threshold chosen. Dehiscence at the time of maturity at the MRCA of subtribe Macadamiinae and dehiscence at the time of germination at the MRCA’s of subtribes Malagasiinae and Virotiinae result in higher proportional likelihoods of the data when those nodes are constrained to have those states in turn (0.915, 0.932, and 0.958, respectively), and these differences are each statistically significant. The data have a statistically significant higher proportional likelihood when the MRCA of all of subtribe Gevuininae except Cardwellia is constrained to have dehiscence at the time of germination (0.999). The pattern of outer mesocarp succulence (character 43) in extant taxa has a statistically significant higher proportional likelihood when the MRCA of tribe Macadamieae is constrained to have a dry outer mesocarp (0.971). A dry outer mesocarp at the MRCA of subtribe Macadamiinae and a succulent mesocarp at the MRCAs of subtribes Malagasiinae, Virotiinae, and Gevuininae (minus Cardwellia) result in statistically significant higher proportional likelihoods of the data when those nodes are constrained to have those states, in turn (0.999, 0.981, 0.981, and 0.998, respectively). The pattern of inner mesocarp texture (character 45) in extant taxa has a higher proportional likelihood when the MRCA of tribe Macadamieae is constrained to have a strongly lignified and bony inner mesocarp (0.566), but this higher proportional likelihood is not statistically significant. A leathery inner mesocarp at the MRCA of subtribe Macadamiinae and a strongly lignified and bony inner mesocarp at the MRCAs of subtribes Malagasiinae, Virotiinae, and Gevuininae (minus Cardwellia) result in statistically significant higher proportional likelihoods of the data when those nodes are constrained to have those states, in turn (0.980, 0.980, 0.981, and 0.999, respectively).
Inference of ages
The mean and 95% credible interval of ages inferred for the 12 focal nodes discussed here varied when we systematically varied the age constraints (Table 4) and the assumed prior probability distributions for rtrate and brownmean (Table 5). However, only in those dispersal scenarios involving node 9—scenarios sensitive to the resolution of a trichotomy, as discussed—did the 95% credible interval for the ages of geographic disjunctions include or predate the last continental connection between the landmasses involved (Tables 4, 5; Fig. 6). Those landmasses, Australia and South America, are thought to have had the most recent continental connections among the fragments of Gondwana (via Antarctica; McLoughlin, 2001). Node 2 is unambiguously resolved to also involve an Australian/South American disjunction (Fig. 5), and the older boundary of the 95% credible interval for the age of that node is 20.8 Ma with our preferred set of prior assumptions (minimum age of MRCA of Gevuininae = 34 Ma, rtrate = 0.039, brownmean = 1.0; Table 4). That older boundary is 26.6 Ma when the minimum age of Gevuininae is increased to 49 Ma (Table 4), and it ranges as high 24.4 Ma when alternative priors for rtrate and brownmean are used with our preferred minimum age of Gevuininae. One of the three possible resolutions of node 8 (Resolution 1) results in the unequivocal inference of an Australian/South American disjunction first arising among the members of that node. However, optimizations on the other two resolutions of the node (2 and 3) resolve that scenario as equally parsimonious with one in which the Australian/South American disjunction arises first between Cardwellia and the stem leading to node 8 (node 9). As was found with node 2, none of the alternative assumptions shifted the older boundaries of the 95% credible intervals for the age of node 8 to include 29 Ma (Tables 4, 5). With one exception (when the minimum age constraint for subtribe Gevuininae was not used), the 95% credible intervals for the age of node 9 were consistently older than the period during which the latest continental connections between Australia and South America occurred (29–33 Ma BP; Tables 4, 5). The geographic disjunctions involving the earliest fragmentations of Gondwana (those involving Africa and Madagascar) occur at nodes 3 and 4. The older boundary of the 95% credible interval for the age of node 3 (a disjunction between Africa and Australia) is 29.0 Ma with our preferred set of assumptions (Table 4), less than a third of the age of the last continental connection between those two fragments. The older boundary of the 95% credible interval for the age of node 4 is 21.4 Ma with our preferred set of assumptions (Table 4), about a sixth of the age of the last continental connection between those two fragments.
|
|
The mean inferred age of the geographic disjunctions (nodes 1–9) across alternative priors varied by as much as 4.7 Ma (for node 4) and as little as 1.8 Ma (for node 7; Table 5). The mean inferred age of the MRCA of Macadamia and tribe Macadamieae varied by 5.5 Ma and 10.6 Ma, respectively, across the alternative priors. The lowest mean ages for the nodes were inferred using the highest rtrate prior (0.39), the highest mean ages for the nodes was inferred using the lowest rtrate prior (0.0039; Table 5). This result is consistent with the sensitivity analyses of Wiegmann et al. (2003) and Bell et al. (2005).
Incrementally increasing the root age prior (rttm), the minimum and maximum age of the root, and the bigtime value up to values 110 Ma greater than our preferred values resulted in temporal shifts sufficient to include the age of last continental connections in the 95% credible intervals for nodes 2 and 8 (two nodes involved in disjunctions between Australia and South America, or potentially so). The mean inferred age of the other geographic disjunctions did not change enough across these alternative assumptions to change our conclusion regarding their origins. The 95% credible interval for the age of node 2 includes 29 Ma (the last continental connection between Antarctica and South America) when the root age prior (the age of the MRCA of Proteaceae and Platanaceae) is 185 Ma; the 95% credible interval for node 8 includes 29 Ma when the root prior is 215 Ma. The 95% credible interval for the age of node 9 no longer includes 33 Ma (the last continental connection between Antarctica and Australia) when the root age prior 175 Ma.
None of the mean or 95% credible intervals for nodes 1–12 differed by >1 Ma between any of the replicate pairs of MCMC analyses done using Multidivtime. All of the 95% credible intervals for the age of the MRCA of Malagasiinae and Virotiinae were older than the upper boundary of the Oligocene (23 Ma BP; Tables 4, 5), and we consider this to be congruent with the Oligocene age of the fossils that share derived sculpturing of the inner mesocarp with extant members of the Virotiinae (Rozefelds, 1992).
DISCUSSION
The monophyly of genera and subtribes
The phylogeny that we reconstruct here for the 16 genera of tribe Macadamieae (Fig. 6) is a major advance in our understanding the group’s evolutionary relationships. The most recent prior phylogenetic hypothesis for the tribe (Weston and Barker, 2006) did not resolve relationships among the subtribes or among most genera within the subtribes. The most striking features of these newly resolved relationships are the paraphyly of Macadamia and the sister relationship of Orites megacarpus with Panopsis, rather than with O. diversifolius (Fig. 6). In this section, we will focus mainly on molecular and morphological support for these results, insofar as it is necessary to assess the congruence between these two sources of evidence for phylogenetic relationships and to decide which action to take to maintain monophyly at the generic rank. However, we also note that the phylogenetic relationships support the recent assertion (e.g., by Weston and Barker, 2006) that the five New Caledonian species with combinations in Macadamia are more closely related to the New Caledonian Virotia leptophylla (Figs. 1J, 4) and that Floydia, Lambertia, and Roupala should be excluded from tribe Macadamieae (Weston and Barker, 2006; cf., Johnson and Briggs, 1975; Fig. 6).
As currently circumscribed, Macadamia is composed of two clades—one subtropical Australian and one tropical Australian and Sulewesian—that are not each others’ closest relatives (Fig. 6). The subtropical clade is composed of four species, including the two widely cultivated for their "nuts" (actually their embryos, M. integrifolia and M. tetrapylla) and the type species (M. ternifolia). These species have narrow distributions that are all found in a short (ca. 500 km) stretch of the eastern Australian coast (Gross, 1995) within, and immediately north of, the McPherson-Macleay area of endemism (Burbidge, 1960; Weston and Crisp, 1994; Crisp et al. 1995). The "McPherson-Macleay overlap" was originally identified by Burbidge (1960) as a region where temperate and tropical Australian "migration tracks" overlap most conspicuously. Many otherwise tropical rainforest taxa have their southern limits of distribution here (e.g., Macadamia, Hicksbeachia, Triunia L.A.S.Johnson & B.G.Briggs, Alloxylon P.H.Weston & Crisp, and subtribe Floydiinae in the Proteaceae), and it is the northern limit of distribution of many predominantly temperate, often sclerophyllous taxa (e.g., Telopea R.Br., Strangea Meisn., and Isopogon R.Br. in the Proteaceae). More recently, though, the McPherson-Macleay overlap itself has been viewed as an area of endemism (e.g., Weston and Crisp, 1994; Crisp et al., 1995) that might have developed as such beginning in the early Miocene as mesic biomes contracted into wetter refugia along the eastern coast of Australia in response to the dramatic drying of Australia’s climate. The tropical clade is composed of five, narrowly distributed species found in northeast Australia and Sulawesi—a range within ca. 1000 km of the subtropical clade. This tropical clade is not sister to the subtropical clade but to a widespread clade composed of the southern African genus Brabejum (1 sp.), the South and Central American genus Panopsis (25 spp.), and an Australian species not previously considered a part of tribe Macadamieae, O. megacarpus.
One of the striking features of the chronogram (Fig. 6) is the relative age of the MRCA of these subtropical and tropical clades of Macadamia (mean inferred age = 29.4 Ma; unless otherwise noted, we cite the mean inferred age from the analysis using our preferred set of prior assumptions; node 10, Table 4). With one exception, the MRCA is older than the divergence between any other taxon recognized at generic rank in tribe Macadamieae and its sister. For example, the divergence between Panopsis and O. megacarpus is about half the age of the MRCA of Macadamia (15.3 Ma; node 2, Table 4). The exception is the divergence between Cardwellia and the clade composed of the rest of subtribe Gevuininae (36.2 Ma; node 9, Table 4)—a divergence that approaches the period of time when the subtribes were diverging one from another (Fig. 6). Given the age of their MRCA, one might wonder how similar the two clades of Macadamia indeed are and how different they are from other members of the subtribe. We address that question, as well as discuss inferred synapomorphies, next.
The subtropical clade of Macadamia differs from most or all members of its sister clade (the tropical clade of Macadamia and its widespread sister clade) in six morphological characters (characters 14, 15, 17, 18, 29, and 41; Table 3). Fitch optimization unambiguously resolves a state of one of these, a "telopetal" direction of anthesis in the conflorescence (developing from the middle toward the base and tip; character 18), as providing a synapomorphy for the subtropical clade of Macadamia. Another state, the adnation of the staminal filament to the subtending tepal in its basal half only (character 29), provides a synapomorphy for the clade composed of the tropical macadamias and their widespread relatives, but this character reverses to more extensive adnation in M. hildebrandii and M. claudiensis. Character 41, the color of the mature exocarp, is sufficiently variable to make Fitch optimization of the states ambiguous at the base of the subtribe. However, the brown color seen in the tropical macadamias and their widespread relatives (polymorphic brown and green in M. claudiensis) is not seen elsewhere in the tribe and arguably provides a synapomorphy for the clade composed of the tropical macadamias and their widespread relatives, despite the ambiguity implied by the algorithm. For the remaining three characters (14, 15, 17), Brabejum (and M. whelanii, for character 14) shares the state of the subtropical clade of Macadamia, introducing ambiguity into the Fitch optimization.
Most or all members of the tropical clade of Macadamia differ from most or all members of its widespread sister clade in four morphological characters (13, 33, 42, 50; Table 3). Coincidentally, for each of these four characters, most or all members of the subtropical clade of Macadamia are scored with the same state as most or all members of the tropical clade of Macadamia. This character state distribution means that none of the four characters are inferred to provide an unambiguous synapomorphy for the tropical clade of Macadamia. While this might seem suggestive of a close relationship between the two clades of Macadamia to the exclusion of the widespread clade composed of Brabejum, O. megacarpus, and Panopsis, we point out that only one of these (a bony testa, character 50) is unambiguously resolved as a synapomorphy for Macadamia when the two clades are made sister to one another and that this is secondarily lost in the MRCA of M. claudiensis and M. hildebrandii. This single unambiguous synapomorphy can be compared to the two synapomorphies discussed earlier for a sister relationship between the tropical macadamias and their widespread relatives. None of these characters provide an unambiguous synapomorphy for the widespread clade either.
The absence of synapomorphies for the tropical clade of Macadamia and for its widespread sister clade might arise because the stems subtending their crown groups are inferred to be relatively short (Fig. 6), leaving little time for the origin and fixation of new phenotypes. The brevity of the stem lineages of both the tropical clade of Macadamia and its widespread sister clade is especially striking when compared to the stem subtending the crown group of the subtropical clade of Macadamia (Fig. 6). While short stem lineages for the former two clades might have led to an absence of morphological synapomorphies among extant members, it produced a larger window of opportunity for the generation of morphological diversity within these correspondingly older crown groups. This longer time period appears to have translated into greater morphological diversity in these groups. Members of the subtropical clade, with a mean inferred crown age of 7.1 Ma, vary in just two of the characters sampled in our study, but the tropical clade of Macadamia and its widespread sister clade, with mean inferred crown ages more than triple that (24.1 and 23.1 Ma, respectively), vary in eight and 13 characters, respectively.
The position of O. megacarpus within the clade that is otherwise composed of taxa from subtribe Macadamiinae was unexpected. This species differs spectacularly from subtribe Macadamiinae and other Proteaceae, including Orites, in its unusual fruit and seed morphology. Its obovoid, unripe fruits are almost globose, and thus superficially similar to those of most other Macadamiinae. However, the thick pericarp dehisces widely (Fig. 1E), quite unlike that of any other member of the Macdamiinae, releasing two, flat, narrowly winged seeds that are quite unlike the solitary, ±globose, wingless seed found in most Macadamiinae. That said, comparison of the flowers of O. megacarpus with that of some species of Panopsis (such as P. cinnamomea Pittier from Venezuela) reveals them to be remarkably similar. The species name was recently published by George and Hyland (1995) for a species found from 100–1200 m a.s.l. on three mountains in a small area of northern Queensland. They placed it in Orites without discussing alternative generic placements, and their note that "the linear bracts, villous flowers and large, thick-walled fruits are distinctive" (George and Hyland, 1995, p. 348) leaves one wondering why they chose to place it in Orites (compare Fig. 1E and F). In our morphological data set, O. megacarpus differs from O. diversifolius (the type for Orites) in 12 characters and from all sampled members of its sister group, Panopsis, in seven characters. Five of the differences between O. megacarpus and Panopsis are resolved as autapomorphies for O. megacarpus (characters 1, 4, 37, 39, 52; Tables 2, 3), and for each of these, the state seen in O. megacarpus is not seen elsewhere in subtribe Macadamiinae. We could not score O. diversifolius for the first two of these five characters (1 and 4), but seedlings of O. excelsus examined by one of us (PHW) have the same states as O. megacarpus for these characters. Like Panopsis, O. diversifolius differs in state from O. megacarpus in the last two characters (39 and 52). The ovule morphology of both O. megacarpus and O. diversifolius is hemitropous (state 0, character 37), unlike all but one other species in tribe Macadamieae, Cardwellia sublimis F.Muell. However, hemitropous ovules are not an autapomorphy for O. diversifolius and thus are not a putative synapomorphy for O. diversifolius and O. megacarpus to the exclusion of other sampled taxa. In fact, none of the sampled characters provide unambiguous autapomorphies for O. diversifolius. Like other members of the clade composed of the tropical macadamias and their widespread relatives, O. megacarpus has staminal filaments adnate to the subtending tepals in their basal halves only (29) and fruit with a brown exocarp at maturity (41)—the two synapomorphies for the clade.
Several options exist for establishing monophyly at the generic rank in subtribe Macadamiinae, but we favor the creation of new genera for O. megacarpus (Fig. 1E, I) and the tropical clade of Macadamia (Fig. 1H). This choice seems reasonable, given that O. megacarpus has five character states not otherwise seen in subtribe Macadamiinae and it maintains Panopsis as a morphologically and geographically cohesive group. However, it does create a genus out of a clade, the tropical macadamias, for which we could not find a morphological synapomorphy. Alternatives to this include circumscribing a large genus composed of a broader clade—composed of the tropical clade of Macadamia, O. megacarpus, Panopsis, and Brabejum—called Brabejum (the oldest name at generic rank). This alternative would circumscribe a genus with two synapomorphies. However, it would be quite disruptive to the nomenclature of the group because it would result in 30 nomenclatural changes for the flora of South America, Australia, and Sulawesi. We make the new combinations in the last section.
Origin and timing of geographic disjunctions
The biogeographic scenario emerging from the results of this study turns the Gondwanan narrative for explaining biotic distributions of extant members of the tribe on its head. Most, or all, lineages extant during Gondwanan fragmentation did not "co-speciate" with the supercontinent and survive until the present, as previously accepted by many authors (e.g., Venkata Rao, 1971; Johnson and Briggs, 1975; Weston and Crisp, 1996; Prance and Plana, 1998; Prance et al., 2007). Instead, it was only after Gondwana had fully fragmented (Fig. 6) that dispersal events out of Australia (to begin with) led to an extant distribution that today stretches across the southern hemisphere. Most dispersal events in the tribe are inferred to have occurred between 23.1 Ma and 8.1 Ma BP (Table 4), at a time after the last continental connections between Australia, South America, and Antarctica (ending 29–33 Ma BP). While some topological resolutions (2 and 3; Fig. 5B) of a trichotomy in subtribe Gevuininae lead to an Australian/South American disjunction prior to the last continental connection between these fragments via Antarctica, another scenario in which the disjunction occurs well after the fragmentation is equally parsimonious. Note that we do not mean to imply that the fragmentation of Gondwana did not have an evolutionary impact on its biota or that the tribe was not distributed outside of Australia earlier than the dispersal events inferred here. Rather, our point is that the extant diversity on the continents other than Australia is not descended from the biota that existed on them at the time of their isolation from other fragments of Gondwana. In this section, we will consider the congruence of our results with those of Barker et al. (2007), and, in the next section, we will discuss biotic correlates (the evolution of the tribe’s diaspores) and abiotic correlates (the advent of the Antarctic Circumpolar Current) to the dispersal events of the last 25 Ma.
The biogeographic scenario for subtribe Macadamiinae supported by our results fundamentally differs from that of Barker et al. (2007) in the mechanisms responsible for the disjunctions. We infer three dispersal events out of Australia in subtribe Macadamiinae: one to Sulawesi, one to South and Central America (mean inferred age = 15.2 Ma), and one to Africa (mean inferred age = 23.1 Ma; Table 4). We were unable to include the Sulawesian species M. hildebrandii and M. erecta in the six non-ITS data sets used for inferring the ages of disjunctions because of PCR failures (M. hildebrandii) and unavailability of tissue for DNA extractions (M. erecta). However, Macadamia hildebrandii and M. claudiensis are sister in the ITS results (PP 85%), and a sister relationship between M. hildebrandii and M. erecta is considered likely by McDonald and Ismail (1995) based on morphological evidence. This evidence leads us to believe that an ancestor of these species arrived by dispersal to Sulawesi less than 12.1 Ma (the mean age of the divergence between M. grandis and M. claudiensis; Table 4). A vicariance explanation for their position on Sulawesi then appears untenable. The tectonic plates that make up Sulawesi are thought to have fragmented from the Australia-New Guinea plate at much earlier times: in the Late Jurassic (ca. 160 Ma BP) and in the early Eocene (ca. 50 Ma Bp; Audley-Charles, 1987; van Welzen et al., 2005). Furthermore, it appears that these plates have only emerged above sea level in the last 25 Ma (Hall, 2001), precluding their service as rafts for species from Australian-New Guinea (van Welzen et al., 2005). Perhaps corroborating this recent dispersal scenario is the observation that M. hildebrandii is most frequently found on ultramafic substrate of young age at low elevations (McDonald and Ismail, 1995). Barker et al. (2007) infer a topology in which the African Brabejum and the South American Panopsis are sister and these together are sister to the Australian and Sulewesian Macadamia (using one operational taxonomic unit, OTU, per genus). They conclude that this topology, in combination with their divergence age estimates, is partly consistent with the widely accepted model of Gondwanan fragmentation (McLoughlin, 2001). Their favored scenario begins with the MRCA of the subtribe widespread across Australia, Antarctica, and South America. That distribution is then disrupted with the loss of connections to Antarctica, isolating Macadamia on Australia and the ancestor of Panopsis and Brabejum on South America. Finally, a long-distance dispersal from South America to Africa leads to Brabejum. Barker et al. (2007) prefer this scenario because it invokes the fewest number of long-distance dispersal events by explaining disjunctions that can be explained by cospeciation of fragments and their biota in that way.
The incongruities in these two biogeographic scenarios primarily arise from differences in the inferred age of relevant divergence events. Barker et al. (2007) consistently estimated older ages for their divergence events, and these older age estimates sometimes include the last continental connections between relevant Gondwanan fragments when ours do not. For example, they inferred a mean age of the divergence between Brabejum and Panopsis at 43.8 (± 10.0 SD) Ma and that between these two genera and Macadamia at 53.9 (± 10.0) Ma. We inferred a mean age of divergence between Brabejum and Panopsis at 23.1 (17.8–29.0 = 95% credible interval) Ma, between these two and the tropical clade of Macadamia at 25.6 (20.4–31.4) Ma, and between these two and the subtropical clade of Macadamia at 29.4 (24.0–35.3) Ma (Table 4; Fig. 6). Ours is not the only study that inferred ages that are younger than Barker et al. (2007). For example, Barker et al. (2007) inferred a mean age of the crown group Proteaceae at 118.5 (± 8.2) Ma, while Anderson et al. (2005) inferred an age of 85 or 96 Ma for the crown group using rbcL data analyzed with penalized likelihood and nonparametric rate smoothing approaches, respectively.
A possible source of this difference is Barker et al. (2007) use of fossil pollen from the Upper Cenomanian (Dettman and Jarzen, 1998; interpreted as 93 Ma BP by Barker et al., 2007) as a minimum age constraint for the crown-group Proteaceae, despite its possession of only plesiomorphic characters of the family (H. Sauquet and D. Cantrill, personal communication in Barker et al., 2007). Barker et al. (2007) justify this by citing the age of the stem-group Platanaceae as 15 Ma older than 93 Ma and the age of fossils that share synapomorphies with an internal clade of Proteaceae at 82 Ma BP. However, neither of these observations support use of 93 Ma as a minimal age constraint for the crown-group Proteaceae. The age of the stem-group Platanaceae only suggests a minimum age for the stem-group Proteaceae, and fossils sharing synapomorphies with internal clades of Proteaceae would only justify using their age as the minimum age of crown-group Proteaceae, not an age 11 Ma earlier. Their estimate of the age of the crown-group Proteaceae at 118.5 (± 8.2) leaves little time between the appearance of the eudicots in the fossil record (125 Ma BP; Magallón et al., 1999) and the origin of the crown group Proteaceae, and explaining the long branch subtending the crown (Fig. 6; estimated to represent 23–25 Ma by Anderson et al. [2005]) is difficult with that time frame.
That said, other minimum ages used appropriately in Barker et al. (2007) study, but not used in our study, might also explain the older ages in their study. The second and third oldest minimum age constraints used by Barker et al. (2007) are 70 Ma for the stem-group Embothriinae and 65 Ma for the stem-group Banksieae, based upon synapomorphies shared by the respective fossils and their crown groups. We did not use these in our study because of our limited sampling of subfamily Grevilleoideae outside of tribe Macadamieae. However, among those taxa that we sampled, subtribe Embothriinae is most closely related to Grevillea R.Br. and Banksia is part of tribe Banksieae. The 95% credible intervals for the inferred age of the MRCA of Grevillea and its sister (52.7–68.0 Ma) and the MRCA of Banksia and its sister (44.2–57.7) do not include or predate the reconstructed ages of these fossils, suggesting that including them could affect the ages inferred with this data set, bringing them closer to the ages inferred by Barker et al. (2007). Thorough exploration of this will occur elsewhere (A. Mast et al., unpublished manuscript), but we began to explore this here. When we constrained the minimum age of these nodes to 70 and 65 Ma (and otherwise using our preferred set of assumptions), the mean inferred ages of nodes 1–8 increased by only 1.8–5.8 Ma, and the 95% credible intervals still did not include the last continental connections between landmasses involved in their respective disjunctions. The mean inferred age of the disjunction between Brabejum and its sister group using this approximation of Barker et al. second and third oldest minimum age constraints was 28.9 Ma (22.3–36.2)—still 14.9 Ma younger than the mean age inferred in that study.
An additional factor that likely affected age estimates more locally, in subtribe Macadamiinae, is the agglomeration of DNA sequence data in Barker et al. (2007) for Macadamia, a genus that we show here to be paraphyletic. In their study, Barker et al. (2007) combine new data that they generate with data from Hoot and Douglas (1998), and for some genera, they use data from different species for their OTUs at generic rank. As it turns out, their OTU Macadamia uses an rbcL sequence from M. claudiensis (a tropical species) and atpB and atpB-rbcL spacer sequences from M. jansenii (a subtropical species). This agglomeration combines data from clades of differing ages since divergence with Panopsis and Brabejum and presumably results in an age intermediate between the two actual ages of divergence.
It is not entirely clear how inclusion of O. megacarpus would have affected the biogeographic scenario favored by Barker et al. (2007) because their construction depended on the inferred ages of disjunctions, and we do not know if they would have inferred the divergence of O. megacarpus and Panopsis as earlier or later than 29–33 Ma BP with their data. However, exclusion of O. megacarpus from the tree on which we performed Fitch optimization of distributions would have left us with a far more ambiguous result. We note both that discovery of the sister relationship of O. megacarpus and Panopsis was fortuitous and not something that we expected a priori and that biogeographic inference, in general, can be sensitive to taxon sampling. For example, paraphyly of the moderately large Panopsis (25 spp.) with respect to O. megacarpus would alter our interpretation of the dispersal events in this subtribe. However, we do not expect this to be the case because the four sampled species of Panopsis form a monophyletic group in the phylogeny inferred with ITS, the species are morphologically very similar (Prance and Plana, 1998; Prance et al., 2007), and the diaspore morphology of O. megacarpus is not consistent with the morphology of other diapsores thought to undergo long-distance dispersal in this group (with delayed dehiscence, as discussed later).
In subtribe Gevuininae, only the inferences that Bleasdalea dispersed from Australia to New Guinea and that Turrillia arose following a dispersal from New Caledonia to Fiji and Vanuatu are unambiguous. We sampled the Australian species Bleasdalea bleasdalei (F.Muell.) A.C.Sm. & J.E.Haas but not the New Guinean species Bleasdalea papuana (Diels) Domin. So, we did not infer an age for the disjunction between these two areas. However, we infer that it originated recently because the mean inferred age for the divergence between Bleasdalea and its sister, the Australian genus Hicksbeachia, is 11.5 Ma. There have been a number of terrestrial connections between Australia and New Guinea via sea level fluctuations in the shallow Torres Strait, and the latest of these occurred just 18000 yr ago (Veevers, 1991; Hope, 1994). The mean inferred age of the disjunction between Turrillia and its sister on New Caledonia, Kermadecia, is 8.1 Ma (Table 4). This age corresponds to a time after the origin of the volcanic ridges that make up the archipelagos of Fiji (Late Eocene) and Vanuatu (Late Oligocene) and the origin of the North Fiji Basin that separates the two island chains (11–12 Ma BP; Schellart et al., 2006). It is also well after the rifting of the ‘Eua Ridge from New Caledonia at 41 Ma BP and its subsequent accretion to Fiji at about 6 Ma BP—events postulated by Kroenke (1996). Weston and Crisp (1996) suggested that vicariance has likely played a larger role in the biota of Fiji and Vanuatu than previously recognized, based on the distributional limit of many "primitive" taxa at Fiji (e.g., Proteaceae, Winteraceae, Fagaceae, Balanopaceae, Magnoliales, and conifers). However, it now seems clear that, in the case of Turrillia, the timing of the relevant tectonic events substantially predated the biological event that they were supposed to explain. The trichotomy inferred in subtribe Gevuininae leads to an inability to discriminate between an early Australia/South America disjunction arising at node 9 or a later disjunction arising at node 8 (Figs. 5B, 6). The later Australia/South America disjunction at 13.6 Ma BP (node 8; Table 4) would bring that dispersal into a period of time during which all of the other disjunctions represented in today’s biota were originating.
Diaspore evolution and dispersal
Prior to asking if there are any correlates to this apparent increase in vagility starting ca. 25 Ma BP, it is important to note that extant distributions are a product of both dispersal and survival to the present. It is possible that the tribe dispersed as frequently from one landmass to another prior to 25 Ma BP as it did after that time but that these early immigrants are no longer represented in the biota of those continents due to extinction. However, a fossil record of earlier immigrants has not been discovered. The only extra-Australian fossil attributable to the group using synapomorphies is the fossil cuticle of subtribe Gevuininae from the early Miocene (ca. 23 Ma BP) of New Zealand mentioned earlier (Carpenter, 1994; Pole, 1998). In the absence of fossil evidence suggesting a more widespread lineage prior to 25 Ma BP, we will consider both the biotic and abiotic events occurring prior to, or at the same time as, this period of dispersal events that are represented in the extant biota.
The diaspores in the tribe vary from the winged seeds of Cardwellia to the wingless seeds protected by a hard testa in Macadamia, to the wingless fruits that have a (typically) succulent outer mesocarp and bony inner mesocarp in the majority of remaining taxa (Fig. 6). It is this third diaspore type that appears to be related to the greatest expansions in distributions (Fig. 5A). Dehiscence of the fruit at seed germination, rather than at fruit maturity, occurs in every extra-Australian genus in the tribe, as well as some Australian members. It is inferred to have arisen independently 4–5 times, and in each case this was prior to, or as early as the time of, each of the dispersal events (Fig. 5A), and the character state changes are significantly correlated with dispersal out of Australia, as determined using Pagel’s (1994) method. We conclude that the pericarp might then have served one or more important functions in the dispersal events. These might include initial attraction of the disperser (when it is an organism), protection of the seed during long periods under harsh conditions (e.g., in a gut or in seawater), or provision of buoyancy in water. Two other characters were also significantly correlated with dispersal out of Australia: the texture of the outer and inner mesocarp. A succulent outer mesocarp and bony inner mesocarp in subtribes Malagasiinae, Virotiinae, and Gevuininae undoubtedly makes them more attractive to vertebrate dispersers, such as birds, bats, and less vagile organisms (e.g., agoutis; Prance et al., 2007) and protects the seed from destruction by these dispersers, respectively. And these might be the common dispersers in those subtribes (Johnson and Briggs, 1975; Weston and Crisp, 1996).
However, we propose that a dispersal mechanism inconsistent with that suggested by the morphology of these fruits—dispersal by seawater as buoyant fruit or on natural rafts—might have played the greatest role in dispersing seeds of the tribe across the southern hemisphere. We suggest this for three reasons. First, the hard kernel (the inner mesocarp or the testa) of the fruits in the group is larger (>1 cm) than those known to be ingested by bats and vagile birds (reviewed by Whittaker and Jones, 1994; Shilton et al., 1999). This fact suggests that seeds do not pass through the guts of these organisms but are instead discarded short distances from the source trees. Whittaker and Jones (1994) considered colonization of Krakatau from adjacent islands—a distance much shorter than that between Australia and South America—via movement of seeds that are orally ejected to be highly unlikely. Furthermore, Higgins et al. (2003) found in a meta-analysis of available data sets that the relationship between morphological dispersal syndromes and long-distance dispersal is poor, in part because of the role that nonstandard dispersal mechanisms play in primary and secondary dispersal (e.g., movement by water after oral ejection of a kernel at a riparian perch). Second, some of the extra-Australian genera are known to be dispersed by water or live in riparian habitats and are presumed to be dispersed by water, suggesting ecological signatures that persist from the original dispersal event. For example, the South American genus Panopsis includes species dispersed by water (Prance et al., 2007), and the African genus Brabejum occurs mainly along permanent water courses (Rourke, 1998). And finally, 25 Ma BP is shortly after the onset of the Antarctic Circumpolar Current (ACC)—a powerful ocean current that rotates eastward around an isolated Antarctica. The ACC is assumed to have had a significant impact on intercontinental dispersal of the southern hemisphere biota and is used to explain asymmetry in the rates of dispersal between pairs of southern continents (e.g., Winkworth et al., 2002; Sanmartín et al., 2007). While the Australian genera are today in the northeast of that continent—outside of the latitudes in which the ACC is found—we point out that rainforest was much more widespread in Australia in earlier times and that lower latitude currents interact with the ACC. Subtribe Gevuininae is known to have occurred along the southern coast of Australia in the Eocene (Carpenter and Pole, 1995). And, while the drying of Australia began with the onset of the ACC, it was only ca. 15 Ma BP that the major northerly shift across the continent of anticyclonic high-pressure cells that block rain-bearing low-pressure cells occurred (Bowler, 1982). Thus, there seems to have been opportunity for the tribe to use the ACC for the dissemination of its diaspores across the greatest distances.
Establishing monophyly at the generic rank
Here, we establish two new names at generic rank and make new combinations for relevant species in those genera, as justified earlier. We also make new combinations for the New Caledonian species of Macadamia in Virotia because they are more closely related to the type of Virotia than Macadamia (Fig. 4).
New generic names
Nothorites P. H. Weston & A. R. Mast, gen. nov.
Description: Arbores. Hypocotylus evolutus. Trichomata simplicia. Folia alterna, simplicia, integra. Conflorescentia lateralis vel terminalis, ramosa. Par florum pedunculum communem deficiem, bractea lineari subtentum. Flores ebracteati, pedicellati, hermaphroditi. Perianthium actinomorphum. Glandes hypogynae connatae, annulatae, cupulatae. Ovarium breviter stipitatum; ovula 2, hemitropa, lateraliter inserta; stylus inter tepala ante anthesin non protrudens; pollenophorum plus minusve tumidum, plus minusve actinomorphum; stigma terminale. Fructus obovoideus, plene dehiscens, complanatus post dehiscentia; mesocarpium externum crassum, coriaceum, fibris dense radiantibus; mesocarpium internum tenue. Semina 2, plana, elliptica vel fere orbicularia, ala angusta marginali circumcincta; testa tenuis, chartacea.
Trees. Hypocotyl developed. Trichomes simple. Leaves alternate, simple, entire. Conflorescence terminal or lateral, branched. Flower pair lacking a common peduncle, subtended by a linear bract. Flowers ebracteate, pedicellate, hermaphrodite, actinomorphic. Hypogynous glands connate, annular, cuplike. Ovary shortly stipitate; ovules 2, hemitropous, laterally inserted; style not protruding between the tepals before anthesis; pollen presenter slightly swollen, more or less radially symmetrical; stigma terminal. Fruit obovoid, fully dehiscing, flattened after dehiscence; outer mesocarp thick, leathery, with dense radiating fibers; inner mesocarp thin. Seeds 2, flat, elliptic to almost orbicular; surrounded by a narrow, marginal wing; testa thin, chartaceous.
Type: Nothorites megacarpus (A. S. George & B. Hyland) P. H. Weston & A. R. Mast.
Derivation of name: from the Greek nothos (bastard, base-born) and Orites (the genus in which this species was originally placed), indicating the inappropriate original generic placement of this species.
Lasjia P. H. Weston & A. R. Mast, gen. nov.
Description: Arbores. Hypocotylus vestigialis. Trichomata simplicia. Folia adulta verticillata, simplicia, integra. Conflorescentia lateralis vel terminalis, plerumque ramosa, raro simplex. Par florum pedunculum communem deficiem, bracteatum. Flores ebracteati, pedicellati, hermaphroditi. Perianthium actinomorphum. Glandes hypogynae connatae, annulatae, cupulatae. Ovarium breviter stipitatum; ovula 2, orthotropa, pendula; medium styli inter tepala ante anthesin protrudens; pollenophorum non tumidum vel plus minusve basaliter tumidum, plus minusve actinomorphum; stigma terminale. Fructus globosus, tarde dehiscens; mesocarpium externum crassum, coriaceum, fibris dense radiantibus; mesocarpium internum tenue. Semen plerumque solitarium, globosum, non alatum; testa tenuis chartaceaque vel crassa osseaque.
Trees. Hypocotyl vestigial. Trichomes simple. Adult leaves whorled, simple, entire. Conflorescence terminal or lateral, usually branched, rarely simple. Flower pair lacking a common peduncle, bracteate. Flowers ebracteate, pedicellate, hermaphroditic. Perianth actinomorphic. Hypogynous glands connate, annular, cuplike. Ovary shortly stipitate; ovules 2, orthotropous, pendulous; middle of style protruding between the tepals before anthesis; pollen presenter not swollen or slightly swollen basally; stigma terminal. Fruit globose, tardily dehiscent; outer mesocarp thick, leathery, with dense radiating fibers; inner mesocarp thin. Seed usually solitary, globose, wingless; testa thin and chartaceous or thick and bony.
Type: Lasjia claudiensis (C. L. Gross & B. Hyland) P. H. Weston & A. R. Mast.
Derivation of name: formed from the initials of the late Dr. Lawrence Alexander Sidney Johnson (L.A.S.J.) to honor his outstanding contribution to our knowledge of the evolution and biogeography of the family Proteaceae.
New combinations
Nothorites megacarpus (A. S. George & B. Hyland) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Orites megacarpus A. S. George & B. Hyland (as "Orites megacarpa"), Fl. Australia, 16: 497. 1995.
Lasjia grandis (C. L. Gross & B. Hyland) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Macadamia grandis C. L. Gross & B. Hyland, Austral. Syst. Bot. 6: 347. 1993.
Lasjia claudiensis (C. L. Gross & B. Hyland) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Macadamia claudiensis C. L. Gross & B. Hyland, Austral. Syst. Bot. 6: 343. 1993.
Lasjia whelanii (F. M. Bailey) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Helicia whelanii F. M. Bailey, Report on New Plants, Preliminary to General Report on Botanical Results on Mestons Expedition to the Bellenden-Ker Range. p. 2. 1889.
Lasjia hildebrandii (Steenis) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Macadamia hildebrandii Steenis, Reinwardtia i: 475. 1952.
Lasjia erecta (J. A. McDonald & Ismail R.) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Macadamia erecta J. A. McDonald & Ismail R., Harvard Pap. Bot. 7: 7. 1995.
Virotia angustifolia (Virot) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Macadamia angustifolia Virot, Fl. N. Caled. & Depend. 2: 120. 1968.
Virotia francii (Guillaumin) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Roupala francii Guillaumin (as "Rhopala francii"), Bull. Mus. Hist. Nat. Paris. Ser. II. 5: 325. 1933.
Virotia neurophylla (Guillaumin) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Kermadecia neurophylla Guillaumin, Bull. Mus. Hist. Nat. Paris. Ser. II. 5: 325. 1933.
Virotia rousselii (Vieill.) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Roupala rousselii Vieill. (as "Rhopala rousselii"), Bull. Soc. Linn. Normand. 9: 394. 1865.
Virotia vieillardi (Brongn. & Gris) P. H. Weston & A. R. Mast, comb. nov.
Basionym: Roupala vieillardii (as "Rhopala vieillardi") Brongn. & Gris, Bull. Soc. Bot. France 10: 229. 1863.
Conclusions
The results presented here both fundamentally change our views on the evolutionary history of tribe Macadamieae and nicely illustrate the sensitivity of conclusions to assumptions made with limited knowledge. The most dramatic example of this was the discovery that O. megacarpus, a species not previously considered to be part of the tribe, is sister to Panopsis. This discovery shifts the origin of the disjunction between South America and Australia 10 Ma closer to the present (Fig. 6), making it much younger than the last continental connection between these Gondwanan fragments (via Antarctica). In another example, we show that Barker et al. (2007) assumption of monophyly in Macadamia permitted them to agglomerate data that turned out to be from two clades with different ages of divergence from their sister groups. This inappropriate agglomeration presumably produced an overestimate of the age of a disjunction between Australia and other fragments of Gondwana in their analysis. Our creation of a separate genus for the tropical macadamias, Lasjia, and our combinations in Virotia for the New Caledonian species previously discussed as macadmias (e.g., Virot, 1968) should lead to fewer mistakes of this sort in the future. For our part, we have assumed that moderately large genera (e.g., Panopsis and Euplassa) are monophyletic. Future results to the contrary have the potential to either strengthen or weaken the dispersal scenario that is supported in this study, as well as to resolve some of the ambiguity in our results (were Euplassa found to be paraphyletic with respect to genera resolved here as sister to it).
Our conclusion that most or all of the extra-Australian range of the tribe is the result of dispersal across substantial water barriers is generally robust across a wide range of alternative assumptions (Tables 4, 5). However, it does not remain valid for the Australian-South American disjunctions when the root age prior and the root’s minimum and maximum age constraints are increased by 70 Ma (for node 2) or 100 Ma (for node 8). These alternative assumptions move the origin of the Proteaceae back into the Jurassic and Triassic—a period during which Croizat (1962) thought they originated, but that we recognize today as much earlier than their first occurrence in the fossil record. Of course, if, for example, tricolpate pollen is later found that documents the eudicots at a time 70 Ma or more before their current earliest occurrence, then our assumptions would change sufficiently to conclude that the data do not reject a vicariance scenario for one or both of these nodes. But changes in the earliest occurrence of tricolpate pollen is unlikely because it is distinctive and pollen is abundant in sedimentary rocks (Crane, 1989).
We still have a lot to learn about this arguably under-exploited group. For example, chromosome number proved to be among the poorest known characters in the study. And future work is warranted. Macadamia is the principle orchard crop in Hawaii (Nagao and Hirae, 1992), and its embryos have the highest oil content (78%) of any "nut" on the market (Strohschen, 1986a). However, other promising genera—Brabejum, Gevuina, Heliciopsis, Hicksbeachia, Kermadecia, and Athertonia—are still only exploited locally, and little is known of their reproductive biology.
Appendix 1. Voucher specimens for DNA and GenBank accession numbers.
|
|
FOOTNOTES
1 The authors thank individuals and institutions who provided field and laboratory assistance, leaf material, unpublished data, images, and primer suggestions including N. Barker, G. Bourdy, B. Briggs, E. Brown, D. Cantrill, K. Edwards, D. Feller, A. Ford, C. Gross, G. Jordan, K. Kupsch, H. Kurzweil, I. Macconochie, B. Manara, S. Mathews, E. Milton, R. Ming, D. Paul, T. Pennington, V. Plana, G. Prance, J. Rourke, G. Sankowsky, H. Sauquet, G. Schatz, J.-M. Vieillon, K. Wilson, and F. Zich. This research was funded by a NSF grant (DEB-0516340) to A.R.M., an Undergraduate Fellowship in Mathematical and Computational Biology from the Howard Hughes Medical Institute to C.L.W., and support from the Royal Botanic Gardens and Domain Trust to P.H.W. 
4 Author for correspondence (e-mail: amast{at}bio.fsu.edu)
LITERATURE CITED
Anderson, C. L., K. Bremer, AND E. M. Friis. 2005. Dating phylogenetically basal eudicots using rbcL sequences and multiple fossil reference points. American Journal of Botany 92: 1737–1748.
Angiosperm Phylogeny Group. 2003. An update of the Angiosperm Phylogeny Group classification for the orders and families of flowering plants: APG II. Botanical Journal of the Linnaean Society 141: 399–436.[CrossRef][Web of Science]
Audley-Charles, M. G. 1987. Dispersal of Gondwanaland: relevance to evolution of the angiosperms. In T. C. Whitmore [ed.], Biogeographical evolution of the Malay Archipelago, 5–25. Clarendon Press, Oxford, UK.
Baldwin, B. G. 1992. Phylogenetic utility of the internal transcribed spacers of nuclear ribosomal DNA in plants: An example from the Compositae. Molecular Phylogenetics and Evolution 1: 3–16.[CrossRef][Medline]
Barker, N. P., P. H. Weston, J. P. Rourke, AND G. Reeves. 2002. The relationships of the southern African Proteaceae as elucidated by internal transcribed spacer (ITS) DNA sequence data. Kew Bulletin 57: 867–883.[CrossRef]
Barker, N. P., P. H. Weston, F. Rutschmann, AND H. Sauquet. 2007. Molecular dating of the ‘Gondwanan’plant family Proteaceae is only partially congruent with the timing of the break-up of Gondwana. Journal of Biogeography 34: 2012–2027.[CrossRef][Web of Science]
Bell, C. D., D. E. Soltis, AND P. S. Soltis. 2005. The age of the angiosperms: A molecular timescale without a clock. Evolution 59: 1245–1258.[CrossRef][Web of Science][Medline]
Bosser, J., AND R. Rabevohitra. 1991. Protéacées. In P. Morat [ed.], Flore de Madagascar et des Comores, Familles 45, 57, 93, 94, 107, 46–69. Muséum National d’Histoire Naturelle, Paris, France.
Bowler, J. M. 1982. Aridity in the late Tertiary and Quaternary of Australia. In W. R. Barker, and P. J. M. Greenslade [eds.], Evolution of the flora and fauna of arid Australia, 35–46. Peacock Publications, Frewville, South Australia, Australia.
Burbidge, N. T. 1960. The phytogeography of the Australian region. Australian Journal of Botany 8: 75–209.[CrossRef]
Carpenter, R. J. 1994. Cuticular morphology and aspects of the ecology and fossil history of North Queensland rainforest Proteaceae. Botanical Journal of the Linnean Society 116: 249–303.[CrossRef][Web of Science]
Carpenter, R. J., AND M. Pole. 1995. Eocene plant fossils from the Lefroy and Cowan paleodrainages, Western Australia. Australian Systematic Botany 8: 1107–1154.[CrossRef][Web of Science]
Coates, A. G., AND J. A. Obando. 1996. The geological evolution of the Central American Isthmus. In J. B. C. Jackson, A. F. Budd, and A. G. Coates [eds.], Evolution and environment in tropical America, 21–56. University of Chicago Press, Chicago, Illinois, USA.
Cooper, W., AND W. T. Cooper. 2004. Fruits of the Australian tropical rainforest. Nokomis Editions, Melbourne, Australia.
Crane, P. R. 1989. Paleobotanical evidence on the early radiation of nonmagnoliid dicotyledons. Plant Systematics and Evolution 162: 165–191.[CrossRef][Web of Science]
Crane, P. R., K. R. Pedersen, E. M. Friis, AND A. N. Drinnan. 1993. Early Cretaceous (Early to Middle Albian) Platanoid inflorescences associated with Sapindopsis leaves from the Potomac Group of eastern North America. Systematic Botany 18: 328–344.[CrossRef][Web of Science]
Crepet, W. L., K. C. Nixon, AND M. A. Gandolfo. 2004. Fossil evidence and phylogeny: The age of major angiosperm clades based on mesofossil and macrofossil evidence from Cretaceous deposits. American Journal of Botany 91: 1666–1682.
Crisp, M. D., H. P. Linder, AND P. H. Weston. 1995. Cladistic biogeography of plants in Australia and New Guinea: Congruent pattern reveals two endemic tropical tracks. Systematic Biology 44: 457–473.
Croizat, L. 1962. Space, time, form: The biological synthesis. Published by the author, Caracas, Venezuela.
Dettmann, M. E., AND D. M. Jarzen. 1998. The early history of the Proteaceae in Australia: The pollen record. Australian Systematic Botany 11: 401–438.[CrossRef][Web of Science]
De Vogel, E. F. 1980. Seedlings of dicotyledons. Centre for Agricultural Publishing and Documentation, Wageningen, Netherlands.
Douglas, A. W. 1996. Inflorescence and floral development of Carnarvonia (Proteaceae). Telopea 6: 749–774.
Douglas, A. W., AND S. C. Tucker. 1996a. The developmental basis of diverse carpel orientations in Grevilleoideae (Proteaceae). International Journal of Plant Sciences 157: 373–397.[CrossRef][Web of Science]
Douglas, A. W., AND S. C. Tucker. 1996b. Inflorescence ontogeny and floral organogenesis in Grevilleoideae (Proteaceae) with emphasis on the nature of the flower pairs. International Journal of Plant Sciences 157: 341–372.[CrossRef][Web of Science]
Edwards, A. W. F. 1972. Likelihood. Cambridge University Press, Cambridge, UK.
Felsenstein, J. 1984. Distance methods for inferring phylogenies: a justification. Evolution 38: 16–24.[Medline]
Filla, F. 1926. Der Perikarp der Proteaceae. Flora 120: 99–142.
Florindo, F., A. K. Cooper, AND P. E. O’Brien. 2003. Introduction to ‘Antarctic Cenozoic palaeoenvironments: Geologic record and models.’ Palaeogeography, Palaeoclimatology, Palaeoecology 198: 1–9.[CrossRef]
Gelman, A., AND D. B. Rubin. 1992. Inference from iterative simulation using multiple sequences. Statistical Science 7: 457–472.[CrossRef]
George, A. S. 1981. The genus Banksia. Nuytsia 3: 239–473.
George, A. S., AND B. P. M. Hyland. 1995. Orites. In P. McCarthy [ed.], Flora of Australia, vol. 16, 346–352. CSIRO Australia, Melbourne, Australia.
Gillespie, J. H. 1991. The causes of molecular evolution. Oxford University Press, Oxford, UK.
Gilmore, S., P. H. Weston, AND J. A. Thomson. 1993. A simple, rapid, inexpensive and widely applicable technique for purifying plant DNA. Australian Systematic Botany 6: 139–148.[CrossRef]
Goldman, N. 1993. Statistical tests of models of DNA substitution. Journal of Molecular Evolution 36: 182–198.[CrossRef][Web of Science][Medline]
Gradstein, F. M., J. G. Ogg, AND A. G. Smith. 2004. A geologic time scale. Cambridge University Press, Cambridge, UK.
Gross, C. L. 1995. Macadamia. In P. McCarthy [ed.], Flora of Australia, vol. 16, 419–425. CSIRO Australia, Melbourne, Australia.
Hall, R. 2001. Cenozoic reconstructions of SE Asia and the SW Pacific: Changing patterns of land and sea. In I. Metcalfe, J. M. B. Smith, M. Morwood, and I. D. Davidson [eds.], Faunal and floral migrations and evolution in SE Asia-Australasia, 35–56. A. A. Blkema (Swets and Zeitlinger Publishers), Leiden, Netherlands.
Hasegawa, M., H. Kishino, AND T. Yano. 1985. Dating of the human-ape splitting by a molecular clock of mitochondrial DNA. Journal of Molecular Evolution 22: 160–174.[CrossRef][Web of Science][Medline]
Higgins, S. I., R. Nathan, AND M. L. Cain. 2003. Are long-distance dispersal events in plants usually caused by nonstandard means of dispersal? Ecology 84: 1945–1956.[CrossRef][Web of Science]
Hoot, S. B., AND A. W. Douglas. 1998. Phylogeny of the Proteaceae based on atpB and atpB-rbcL intergenic spacer region sequences. Australian Systematic Botany 11: 301–320.[CrossRef][Web of Science]
Hope, G. S. 1994. Quaternary vegetation. In R. S. Hill [ed.], History of the Australian vegetation, Cretaceous to Recent, 368–389. Cambridge University Press, Cambridge, UK.
Hyland, B. P. M. 1995. Cardwellia. In P. McCarthy [ed.], Flora of Australia, vol. 16, 358–359. CSIRO Australia, Melbourne, Australia.
Johnson, L. A. S., AND B. G. Briggs. 1963. Evolution in the Proteaceae. Australian Journal of Botany 11: 21–61.[CrossRef]
Johnson, L. A. S., AND B. G. Briggs. 1975. On the Proteaceae—The evolution and classification of a southern family. Botanical Journal of the Linnean Society 70: 83–182.[CrossRef][Web of Science]
Kluge, A. G., AND J. S. Farris. 1969. Quantitative phyletics and the evolution of anurans. Systematic Zoology 18: 1–32.
Kroenke, L. W. 1996. Plate tectonics development of the western and southwestern Pacific: Mesozoic to the present. In A. Kest, and S. E. Miller [eds.], The origin and evolution of Pacific Island biotas, New Guinea to Eastern Polynesia: Patterns and processes, 19–34. SPB Academic Publishing, Amsterdam, Netherlands.
Ladiges, P. Y., AND D. Cantrill. 2007. New Caledonia—Australian connections: Biogeographic patterns and geology. Australian Systematic Botany 20: 383–389.[CrossRef][Web of Science]
Lawver, L. A., L. M. Gahagan, AND M. F. Coffin. 1992. The development of the paleoseaways around Antarctica. Antarctic Research Series 56: 7–30.
Lewis, P. O. 2001. A likelihood approach to estimating phylogeny from discrete morphological character data. Systematic Biology 50: 913–925.[CrossRef][Web of Science][Medline]
Maddison, W. P., AND D. R. Maddison. 2005. Ancestral state reconstruction packages for Mesquite, version 1.0. http://mesquiteproject.org.
Maddison, W. P., AND D. R. Maddison. 2007. Mesquite: A modular system for evolutionary analysis, version 2.0. http://mesquiteproject.org.
Magallón, S., P. R. Crane, AND P. S. Herendeen. 1999. Phylogenetic pattern, diversity, and diversification of eudicots. Annals of the Missouri Botanical Garden 86: 297–372.[CrossRef][Web of Science]
Makinson, R. O. 2000. Grevillea. In A. J. G. Wilson [ed.], Flora of Australia, vol. 17A, 1–460. ABRS/CSIRO Publishing, Melbourne, Australia.
Mast, A. R. 1998. Molecular systematics of subtribe Banksiinae (Banksia and Dryandra; Proteaceae) based on cpDNA and nrDNA sequence data: Implications for taxonomy and biogeography. Australian Systematic Botany 11: 321–342.[CrossRef][Web of Science]
Mast, A. R., AND T. J. Givnish. 2002. Historical biogeography and the origin of stomatal distributions in Banksia and Dryandra (Proteaceae) based on their cpDNA phylogeny. American Journal of Botany 89: 1311–1323.
Mast, A. R., E. H. Jones, AND S. P. Havery. 2005. An assessment of old and new DNA sequence evidence for the paraphyly of Banksia with respect to Dryandra (Proteaceae). Australian Systematic Botany 18: 75–88.[CrossRef][Web of Science]
McCarthy, P. [ed.]. 1995. Flora of Australia, vol. 16, Elaeagnaceae, Proteaceae 1. CSIRO Australia, Melbourne, Australia.
McDonald, J. A., AND R. Ismail. 1995. Macadamia erecta (Proteaceae), a new species from Sulawesi. Harvard Papers in Botany 7: 7–10.[Medline]
McLoughlin, S. 2001. The breakup history of Gondwana and its impact on pre-Cenozoic floristic provincialism. Australian Journal of Botany 49: 271–300.[CrossRef][Web of Science]
Midford, P., AND W. Maddison. 2006. Correl package for Mesquite, version 0.1. http://mesquiteproject.org.
Mullis, K., F. Faloona, S. Scharft, R. Saiki, G. Horn, AND H. Erlich. 1986. Specific enzymatic amplification of DNA in vitro—The polymerase chain-reaction. Cold Spring Harbor Symposia on Quantitative Biology 51: 263–273.
Nagao, M. A., AND H. H. Hirae. 1992. Macadamia: Cultivation and physiology. Critical Reviews in Plant Sciences 10: 441–470.[CrossRef][Web of Science]
Olmstead, R. G., AND J. A. Sweere. 1994. Combining data in phylogenetic systematics: An empirical approach using three molecular data sets in the Solanaceae. Systematic Biology 43: 467–481.
Pagel, M. 1994. Detecting correlated evolution on phylogenies: A general method for the comparative analysis of discrete characters. Proceedings of the Royal Society of London, B, Biological Sciences 255: 37–45.
Pole, M. 1998. The Proteaceae record in New Zealand. Australian Journal of Botany 11: 343–372.[CrossRef]
Prance, G. T., AND V. Plana. 1998. The American Proteaceae. Australian Systematic Botany 11: 287–299.[CrossRef][Web of Science]
Prance, G. T., V. Plana, K. S. Edwards, AND R. T. Pennington. 2007. Monograph of Neotropical Proteaceae. Flora Neotropica 100: 1–220.
Qiu, H., AND P. H. Weston. 2003. Proteaceae. In Z. Y. Wu, P. H. Raven, and D. Y. Hong [eds.], Flora of China, vol. 5, 192–199. Science Press, Beijing, China and Missouri Botanical Garden Press, St. Louis, Missouri, USA.
Ronquist, F. 1997. Dispersal–vicariance analysis: A new approach to the quantification of historical biogeography. Systematic Biology 46: 195–203.
Ronquist, F., AND J. P. Huelsenbeck. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics (Oxford, England) 19: 1572–1574.[CrossRef]
Rourke, J. P. 1998. A review of the systematics and phylogeny of the African Proteaceae. Australian Systematic Botany 11: 267–285.[CrossRef][Web of Science]
Rozefelds, A. C. 1992. The subtribe Hicksbeachiinae (Proteaceae) in the Australian Tertiary. Memoirs of the Queensland Museum 32: 195–202.
Sanderson, M. J. 1997. A nonparametric approach to estimating divergence times in the absence of rate constancy. Molecular Biology and Evolution 14: 1218–1231.[Web of Science]
Sanderson, M. J. 2002. Estimating absolute rates of molecular evolution and divergence times: A penalized likelihood approach. Molecular Biology and Evolution 19: 101–109.
Sanmartín, I., AND F. Ronquist. 2004. Southern hemisphere biogeography inferred by event-based models: Plant versus animal patterns. Systematic Biology 53: 216–243.[CrossRef][Web of Science][Medline]
Sanmartín, I., L. Wanntorp, AND R. C. Winkworth. 2007. West wind drift revisited: Testing for directional dispersal in the southern hemisphere using event-based tree fitting. Journal of Biogeography 34: 398–416.[CrossRef][Web of Science]
Schellart, W. P., G. S. Lister, AND V. G. Toy. 2006. A late Cretaceous and Cenozoic reconstruction of the southwest Pacific region: Tectonics controlled by subduction and slab rollback processes. Earth-Science Reviews 76: 191–233.
Shilton, L. A., J. D. Altringham, S. G. Compton, AND R. J. Whittaker. 1999. Old world fruit bats can be long-distance seed dispersers through extended retention of viable seeds in the gut. Proceedings of the Royal Society of London, B, Biological Sciences 266: 219–223.
Sleumer, H. 1955a. Proteaceae. In C. G. G. J. van Steenis [ed.], Flora Malesiana, ser. 1, vol. 5, 147–206. P. Noordhoff Ltd., Groningen, Netherlands.
Sleumer, H. 1955b. Studies in Old World Proteaceae. Blumea 8: 1–95.[Medline]
Smith, A. C., AND J. E. Haas. 1975. Studies of Pacific island plants. XXIX. Bleasdalea and related genera of Proteaceae. American Journal of Botany 62: 133–147.[CrossRef][Web of Science]
Steiger, D. L., P. H. Moore, F. Zee, Z. Liu, AND R. Ming. 2003. Genetic relationships of macadamia cultivars and species revealed by AFLP markers. Euphytica 132: 269–277.[CrossRef][Web of Science]
Strohschen, B. 1986a. Contributions to the biology of useful plants. 4. Anatomical studies of fruit development and fruit classification of the macadamia nut (Macadamia integrifolia Maiden and Betche). Angewandte Botanik 60: 239–247.[Web of Science]
Strohschen, B. 1986b. Contributions to the biology of useful plants. 5. Anatomical studies of fruit development and fruit classification of the monkey nut (Hicksbeachia pinnatifolia F. Muell.). Angewandte Botanik 60: 249–256.[Web of Science]
Swofford, D. L. 2003. PAUP*: Phylogenetic analysis using parsimony (*and other methods), version 4. Sinauer, Sunderland, Massachusetts, USA.
Tavaré, S. 1986. Some probabilistic and statistical problems on the analysis of DNA sequences. Lectures on Mathematics in the Life Sciences 17: 57–86.
Thorne, J. L., AND H. Kishino. 2002. Divergence time and evolutionary rate estimation with multilocus data. Systematic Biology 51: 689–702.[CrossRef][Web of Science][Medline]
Thorne, J. L., H. Kishino, AND I. S. Painter. 1998. Estimating the rate of evolution of the rate of molecular evolution. Molecular Biology and Evolution 15: 1647–1657.[Abstract]
van der Pijl, L. 1972. Principles of dispersal in higher plants, 2nd ed. Springer-Verlag, New York, New York, USA.
Van Steenis, C. G. G. J. 1952. Miscellaneous botanical notes—IV. Reinwardtia 1: 467–481.
Van Welzen, P. C., J. W. Ferry Slik, AND J. Alahuhta. 2005. Plant distribution patterns and plate tectonics in Malesia. Biologiske Skrifter 55: 199–217.
Veevers, J. J. 1991. Phanerozoic Australia in the changing configuration of proto-Pangea through Gondwanaland and Pangea to the present dispersed continents. Australian Systematic Botany 4: 1–11.
Veevers, J. J. 2000. Billion-year earth history of Australia and neighbors in Gondwanaland. Gemoc Press, Sydney, Australia.
Veevers, J. J. 2001. Atlas of billion-year history of Australia and neighbors in Gondwanaland. Gemoc Press, Sydney, Australia.
Venkata Rao, C. 1971. Proteaceae. Council of Scientific and Industrial Research, New Delhi, India.
Virot, R. 1968. Protéacées. In A. Aubreville [ed.] Flore de la Nouvelle Calédonie et Dépendances, vol. 2, 5–254. Muséum National d’Histoire Naturelle, Paris, France.
Von Mueller, F. 1879. Appendix D. Descriptive notes on the Tertiary flora of New South Wales. In Anonymous [ed.], Annual report, 169–172. Mines Department, New South Wales, Australia.
Von Mueller, F. 1883. Observations on new vegetable fruits of the auriferous drifts. Geological Survey of Victoria, John Ferres, Government Printer, Melbourne, Australia.
Weston, P. H. 1995a. Gevuina. In P. McCarthy [ed.], Flora of Australia, vol. 16, 408–410. CSIRO Australia, Melbourne, Australia.
Weston, P. H. 1995b. Hicksbeachia. In P. McCarthy [ed.], Flora of Australia, vol. 16, 410–413. CSIRO Australia, Melbourne, Australia.
Weston, P. H. 1995c. Athertonia. In P. McCarthy [ed.], Flora of Australia, vol. 16, 413–415. CSIRO Australia, Melbourne, Australia.
Weston, P. H. 1995d. Catalepidia. In P. McCarthy [ed.], Flora of Australia, vol. 16, 415–416. CSIRO Australia, Melbourne, Australia.
Weston, P. H. 2006. Proteaceae. In K. Kubitzki [ed.], The families and genera of vascular plants, vol. 9, 364–404. Springer, Berlin, Germany.
Weston, P. H., AND N. P. Barker. 2006. A new suprageneric classification of the Proteaceae, with an annotated checklist of genera. Telopea 11: 314–344.
Weston, P. H., AND M. D. Crisp. 1994. Cladistic biogeography of waratahs (Proteaceae; Embothrieae) and their allies across the Pacific. Australian Systematic Botany 7: 225–249.[CrossRef]
Weston, P. H., AND M. D. Crisp. 1996. Trans-Pacific biogeographic patterns in the Proteaceae. In A. Keast, and S. E. Miller [eds.], The origin and evolution of Pacific island biotas, New Guinea to Eastern Polynesia: Patterns and processes, 215–232. SPB Academic Publishing, Amsterdam, Netherlands.
White, T. J., T. Bruns, S. Lee, AND J. Taylor. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. In M. A. Innis, D. H. Gelfand, J. J. Sninsky, and T. J. White [eds.], PCR protocols: A guide to methods and applications, 315–322. Academic Press, San Diego, California, USA.
Whittaker, R. J., AND S. H. Jones. 1994. The role of frugivorous bats and birds in the rebuilding of a tropical forest ecosystem, Krakatau, Indonesia. Journal of Biogeography 21: 245–258.[CrossRef][Web of Science]
Wiegmann, B. M., D. K. Yeates, J. L. Thorne, AND H. Kishino. 2003. Time flies, a new molecular time-scale for Brachyceran fly evolution without a clock. Systematic Biology 52: 745–756.[CrossRef][Web of Science][Medline]
Winkworth, R. C., S. J. Wagstaff, D. Glenny, AND P. J. Lockhart. 2002. Plant dispersal N.E.W.S. from New Zealand. Trends in Ecology & Evolution 17: 514–520.[CrossRef][Web of Science]
Yang, Z. 1997. PAML: A program package for phylogenetic analysis by maximum likelihood. Computer Applications in the Biosciences 13: 555–556.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
