2007/10/01 by Michael Knapp, Ragini Mudaliar, David Havell +2 · 3 citations
Medicine · Environmental Science · #Injury Epidemiology and Prevention #Marine animal studies overview
paper · pdf · doi:10.1080/10635150701636412
openalex publication_date 2007/10/01 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/29
Eighty million years ago (Ma) the landmass that was to become New Zealand broke away from the Gondwanan supercontinent. During the subsequent Oligocene period (26 to 38 Ma), there was a significant reduction in the landmass of New Zealand (Cooper and Cooper, 1995, and references therein). However, whether or not New Zealand was completely submerged is a matter of controversy and recent debate (Pole, 1994; Macphail, 1997; Winkworth et al., 1999; Stöckler et al., 2002; Campbell and Landis, 2003; Waters and Craw, 2006; Heads, 2006). If it was completely submerged during the Oligocene, then all extant biota must have arrived by transoceanic distance dispersal since that time. This is a hypothesis that has been considered implausible by some researchers (Nelson, 1975; Craw et al., 1999, McCarthy, 2003; Heads, 2006) and possible or even very likely by others (Pole, 1994; Waters and Craw, 2006). Molecular systematic studies on the New Zealand flora and fauna provide one means of testing the Oligocene drowning hypothesis. Recent studies have included invertebrates (Donald et al., 2005), fish (Waters and Craw, 2006, and references therein), birds (Cooper and Cooper, 1995; Cooper et al., 2001; Ericson et al., 2002), and plants (Stöckler et al., 2002; Zhang and Renner, 2003; Knapp et al., 2005). Findings in many of these studies have indicated post-Oligocene arrival of extant New Zealand lineages. However, three studies have been interpreted otherwise. An unbroken Gondwanan link with respect to some New Zealand bird fauna has been inferred from reconstructed gene trees for mitochondrial (Cooper et al., 2001) and nuclear genes (Ericson et al., 2002). Stöckler et al. (2002) have also advanced similar arguments to explain trans-Tasman distributions of Agathis in the plant family Araucariaceae. Their phylogenetic reconstruction of chloroplast rbcL genes identified the New Zealand Agathis australis (Kauri) as the sole representative of an early diverged lineage within the genus that is genetically very distinct from its Australian relatives. This finding, together with the discovery of Early Cretaceous fossils of Agathis species in the Clarence Valley, South Island, New Zealand (Daniel, 1989; Parrish et al., 1998; Ag.seymouricum), was interpreted by Stöckler et al. (2002) as support for Agathis having existed in situ in New Zealand since the breakup of Gondwana. Waters and Craw (2006) have recently argued that the evidence of Stöckler et al. (2002) is weak in terms of the contribution it can make to the debate of whether or not New Zealand was completely drowned in the Oligocene, because molecular clock analyses were not reported in this study. They have also questioned the relevance of 90- to 100-million-year-old Agathis fossils from the Clarence Valley of New Zealand (Daniel, 1989) that Stöckler et al. (2002) have used to support their hypothesis. These fossils predate the breakup of Gondwana and might therefore belong to a widely distributed lineage with close relatives in Australia. Here we report divergence time estimates made using relaxed molecular clock methods for nine different species of Araucariaceae based on three chloroplast regions: two coding regions (matK and rbcL) and one intergenic region (trnD-trnT). These concatenated regions total 3011 bp of unambiguously aligned chloroplast genome sequence. We report that phylogenetic analyses of these data cannot reject the hypothesis that Agathis australis survived the Oligocene drowning of the New Zealand landmass. Rather, they suggest that parts of the New Zealand archipelago remained above sea level throughout the Tertiary period. Chloroplast DNA sequences (3011 bp comprising the matK and rbcL genes as well as the trnD-trnT intergene region) were determined for each of nine accessions of Agathis, Araucaria, and Wollemia (Araucariaceae) sampled from the botanical gardens in Valdivia, Chile (Araucaria angustifolia), the Royal Botanical Gardens, Sydney, Australia (Agathis robusta, Ag. microstachya, Ag. atropurpurea, Araucaria cunninghamii, Ar. bidwillii, Wollemia nobilis), the University of the South Pacific, Suva, Fiji (Agathis macrophylla), the Esplanade Gardens, Palmerston North, New Zealand (Agathis australis), Massey University, Palmerston North, New Zealand (Prumnopitys ferruginea), and the Ruahine Range, New Zealand (Dacrydium cupressinum). All rbcL sequences as well as the matK gene for the New Zealand endemic podocarps Dacrydium cupressinum (Rimu) and Prumnopitys ferruginea (Miro), and Pinus koraiensis (Korea pine), a native of Korea, China, and Japan, were obtained from GenBank (accession numbers in Table 1). Also sequence data for the trnD-trnT intergene region were obtained from GenBank for the Korea pine. All other sequence data were obtained in our laboratory. The sequence data were aligned using progressive multiple-sequence alignment: ClustalX version 1.81 (Thompson et al., 1997). The alignments are available in Supplementary Material at http://www.systematicbiology.org. Accession numbers for sequences used in phylogenetic reconstructions Accession numbers for sequences used in phylogenetic reconstructions Phylogenetic reconstructions were made with PAUP* version 4.0b10 (Swofford, 2003) using heuristic maximum likelihood (ML) criteria. Trees were initially built for the matK and rbcL genes and also the trnD-trnT intergene region. These data were subsequently concatenated, producing an unambiguous alignment of 3011 bp. Analyses were then conducted on this combined data matrix. A model sensitivity test was performed, investigating a range of 60 symmetrical models of DNA substitution corresponding to the 56 implemented in ModelTest version 3.06 (Posada and Crandall, 1998) (http://darwin.uvigo.es/software/modeltest.html) plus F84, F84 + I, F84 + Γ8, and F84 + I + Γ8. ML parameters of these models were estimated by PAUP* following the approach used in ModelTest. These parameters were then used to conduct 60 individual ML heuristic searches in PAUP* with tree bisection-reconnection branch swapping and a neighbor-joining starting tree. ML bootstrap proportions were obtained after 100 replications, using the same search strategy and ML parameters as for the analysis of the original data set. All molecular dating analyses, except for the model sensitivity analysis, were conducted separately for each gene and for the concatenated data set. The model sensitivity analysis was only conducted for the concatenated data set. Divergence dates for all data sets were estimated using both the penalized likelihood (PL) method (Sanderson, 2002, 2003) and a Bayesian relaxed molecular clock approach (BRMC) (Thorne et al., 1998; Kishino et al., 2001; Thorne and Kishino, 2002). Divergence dates were obtained using the PL method of Sanderson (2002) as implemented in the program r8s, version 1.60 (Sanderson, 2003) (http://ginger.ucdavis.edu/r8s/) with the TN algorithm. The outgroup was excluded using the “prune” command. The degree of autocorrelation within lineages was estimated using cross-validation as suggested by Sanderson (2002), and the correcting smoothing parameter defined accordingly. Divergence dates were also estimated on the 60 ML phylograms recovered in the phylogenetic model sensitivity analysis. Ages for each node across the 60 ML trees were summarized using the “profile” command. Confidence limits on dating estimates were computed by using nonparametric bootstrapping of the original data set as suggested by Sanderson and Doyle (2001). PAUP* version 4.0b10 was used to generate 100 bootstrap resampled data sets of sites in length. ML branch lengths of the optimal topology were then estimated under the optimal model for each of the bootstrap resampled data sets using PAUP*. Divergence estimates were then calculated for each of the 100 bootstrap replicates using r8s to obtain standard deviations on each node by the “profile” command and the settings described above. The BRMC approach was applied using the program Multidivtime as implemented in the Thornian Time Traveller (T3) package (Yang, 2003). First, the program Baseml of the Paml package version 3.13 (Yang, 1997) (http://abacus.gene.ucl.ac.uk/software/paml.html) was used to estimate the ML parameters of the F84+8 substitution model, using the ML topology previously identified. Second, the program Estbnew (ftp://abacus.gene.ucl.ac.uk/pub/T3/) was used to estimate branch lengths of the ML topology and the corresponding variance-covariance matrix. Finally, the program Multidivtime was used to run a Monte Carlo Markov chain for estimating mean posterior divergence times on nodes with associated standard deviations from the variance-covariance matrix produced by Estbnew. The Markov chain was sampled 10,000 times every 100 cycles after a burn-in stage of 100,000 cycles. As for the PL method, the outgroup was not included in this analysis. All fossil calibration dates by definition represent minimum ages. However, in order to run the program r8s, at least one maximum age must be defined. Hill et al. (2000) consider it possible that Araucariaceae evolved after the major extinction phase at the end of the Permian period (299 to 251 Ma), which was caused by widespread arid conditions throughout Gondwana. Thus a conservative estimate of 225 Ma was assumed for the maximum age of the root. This date is weakly supported by the fossil record, in which pollen grains assigned to Araucariaceae have been recorded from the early Triassic period (251 to 245 Ma; De Jersey, 1968). In the Araucariaceae phylogeny the Wollemia/Agathis clade is sister to the Araucaria clade (see Fig. 1a). Although the oldest unequivocal Araucaria macrofossils are known from the Jurassic (Stockey, 1982), the oldest reliable records of Wollemia-like pollen (Dilwynites) only date from the Turonian period (91 to 89 Ma) (Macphail et al., 1995). The minimum age of the root was therefore set at 89 Ma. ML trees indicating evolutionary relationships for Araucariaceae based on the matK and rbcL coding regions and trnD-trnT intergene region of the chloroplast genome (3011 bp). Divergence dates (in Ma) were obtained with a (K81uf+Γ) substitution model using the PL approach of Sanderson (2002). For the dates indicated, the age of the root node was constrained to 89 to 225 Ma; 1, 3, and 4 were also constrained in accordance with fossil data at a minimum of 38, 136, and 100 Ma, respectively; (a) topology estimated using concatenated data set and no constraint; (b) topology constrained to topology of Setoguchi et al. (1998). AU, Australia; NG, New Guinea; SA, South America; NZ, New Zealand; FJ, Tropical Australasia; Mi, Miocene; Ol, Oligocene; Eo, Eocene; Pa, Palaeocene. The Araucaria sections Eutacta (in this analysis represented by Araucaria cunninghamii) and Bunya (Araucaria bidwillii) are known from the Jurassic, whereas the oldest reliable records of the section Araucaria (represented by Araucaria angustifolia) are of Early Cretaceous age. (Hill and Brodribb, 1999). Therefore, it was assumed that the most recent common ancestor (MRCA) of Ar. bidwillii and Ar. angustifolia is a minimum of 100 million years old, whereas Ar. cunninghamii and Ar. bidwillii are separated by at least 136 Myr of evolution. Fossils assigned to Agathis have been recorded from the Jurassic and Cretaceous. However, they lack sufficient diagnostic characters to allow confident identification (Hill and Brodribb, 1999). The oldest unequivocal Agathis record is known from the Late Eocene Vegetable Creek sediments in Northern New South Wales, Australia (Hill, 1995). The MRCA of Agathis and Wollemia was therefore assumed to be not younger than 38 Myr. The Bayesian approach to molecular dating allows for additional calibrations points. In addition to the calibration dates described above, a prior for the age of the root and a prior for the expected rate of substitution at the root node were set. Based on fossil evidence, Miller (1977) suggested that Agathis and Araucaria split in the Late Jurassic/Early Cretaceous. According to Hill (1995), more fossil evidence and phylogenetic analyses are needed to support this conclusion. Nevertheless, the date suggested by Miller (1977) provides a prior expectation for the age of the root. Thus we have assumed the basal node of the tree to be 136 Myr. To take into consideration the of this date a prior for the standard of that date was set to Myr. The prior for the substitution rate from root to was calculated separately for each data set branch from root to by estimated of time from root to In every a standard of was The for of the rate at root node and the the rate the degree of rate autocorrelation the of the were from the branch length. The estimate for the of time from root to was set at Myr. We assumed that this a estimate for the age of the and the optimal ML reconstruction made using concatenated matK and rbcL genes and the trnD-trnT intergene region for nine species from the Araucariaceae Agathis, and For this Pinus Dacrydium and Prumnopitys ferruginea were used as The tree estimated for the and trnD-trnT data sets were the trees in their level of These were with respect to the trees reported in studies et al., 1998; Stöckler et al., 2002). The only the of Wollemia Setoguchi et al. a phylogeny of a range of species from the Araucariaceae. Their were inferred from the rbcL gene using a In their analyses Wollemia was sister to the In Araucaria is sister to Wollemia and for this are outgroup in of the data and the To test which of these might be most we reconstructed Araucariaceae relationships using the rbcL data set of Setoguchi et al. and a maximum likelihood tree We were to their that our of ML was not the of the in root we the of outgroup under the maximum likelihood criteria. We that the most support for Araucaria sister to the Agathis and Wollemia lineage we used our set of Dacrydium and Prumnopitys However, the of Pinus the most diverged reconstruction of Setoguchi et We also of phylogenetic relationships for our rbcL data set of nine species using only podocarps as outgroup As with Setoguchi et data Wollemia was to be sister to the The only to Setoguchi et topology was the of Araucaria bidwillii that was with Ar. cunninghamii than with Ar. This in the of Araucaria bidwillii be more were to the data set. This that for the rbcL data set outgroup has significant on the phylogenetic of Wollemia in reconstructed trees for Araucariaceae. We also the above analyses on the concatenated data sets (matK and rbcL genes and the trnD-trnT intergene region) for the nine species for which we recently determined Finally, we conducted a sensitivity analysis of 60 substitution models et al., on the concatenated data set. these and in these analyses, the tree in was recovered with very in branch lengths of the substitution model all substitution models + was identified as the one for the concatenated data based on This substitution model and the F84 + model were used for The model was included because the Bayesian relaxed molecular clock (BRMC) as implemented in the program Multidivtime (see and only allows the of the and the F84 This analysis with the F84 + model a of date estimates to be made using different relaxed molecular clock All nodes of the optimal ML tree recovered in the sensitivity analysis nonparametric bootstrap support than with the only the of Ag. with Ag. atropurpurea, which and the of Ag. to these two which To be to the of the two different on molecular clock we also built a maximum likelihood tree with the concatenated data set the Setoguchi this produced an for Agathis, and Araucaria with respect to the Divergence times for the nodes on the optimal tree were estimated for each gene and intergene region separately and for the concatenated data set using the penalized likelihood (PL) method (Sanderson, and BRMC method (Thorne et al., 1998; Kishino et al., 2001; Thorne and Kishino, 2002). to the of sequence divergence in the individual the standard deviations for the individual data sets not be calculated using the nonparametric bootstrapping method for r8s by Sanderson and Doyle (2001). deviations for the individual data sets were therefore only calculated using The inferred of the nodes of the obtained under the optimal model or the F84 + model of substitution are in Table age and standard estimates based on and rbcL genes and trnD-trnT intergene age and standard estimates based on and rbcL genes and trnD-trnT intergene The standard deviations for the age of nodes for individual regions were for the completely node The date estimates for individual nodes was the different individual data The of divergence time estimates was for node For the concatenated data set the standard deviations for the date estimates were than for the individual data sets and the were by the of substitution model The total age and total standard for each node across all data sets and dating methods are in Table of node age and standard estimates based the concatenated data and constrained of node age and standard estimates based the concatenated data and constrained age of nodes across all genes and methods in Ma and standard of these estimates and sensitivity of estimated node age to 60 models of DNA age of nodes across all genes and methods in Ma and standard of these estimates and sensitivity of estimated node age to 60 models of DNA The of this was to estimate the divergence time of Agathis australis from New Zealand and its relatives from Australia and of the species of Agathis (Hill and Brodribb, were as representative for this based on their and the phylogeny reconstructed by Stöckler et al. (2002) for of three diverged Araucaria lineages as well as Wollemia were as the fossil record for the lineages they represent for calibration of a molecular The topology of the Araucariaceae phylogeny produced by all data sets and under 60 different models of DNA substitution and ML tree is with rbcL gene trees in studies et al., 1998; Stöckler et al., 2002). This that the of used in our was sufficient for reliable phylogenetic The only to of Wollemia Setoguchi et al. a phylogeny of a range of species from the Araucariaceae. Their were inferred from the rbcL gene using a In their analyses, Wollemia was sister to the analyses suggest an root and that this can be to of the Setoguchi et al. data set. it is most to that the phylogenetic Agathis, and Araucaria is close to an In our analyses, sequences from three different data sets produced a root that was recovered of the outgroup However, it is to be confident that this more the because sequences can support to in data and model et al., For this we estimated divergence times with the concatenated data set on two As with analyses of the rbcL Agathis australis (Kauri) was identified in our analyses as the sole representative of an early diverged lineage within the genetically very distinct from the Australian Agathis Agathis which is native to Tropical from the to was sister to the Australian species diverged more recently from this than Ag. The Araucaria species used in our analyses represented three diverged lineages within the and this was supported by our phylogenetic the in date estimates from the different genes were all except one estimate the Oligocene drowning The r8s estimate for the matK gene was only Myr and therefore Early However, all estimates have to be as minimum divergence times the conservative of the calibrations a tree to Setoguchi et topology was used the estimated time for the split Agathis the Australian and Agathis species 3, Fig. 1). that separated from its relatives some time in the This is with the of New Zealand and Australian and a hypothesis of in situ of Agathis in New Zealand throughout the that the calibrations used in our are a for the trans-Tasman of Agathis is However, an is also Agathis lineages have for in Australia. of Agathis australis might have the and become in Australia. This hypothesis might it is supported by studies indicating the of dispersal for the of the fauna and flora of New Zealand Winkworth et al., 2002; et al., Waters and Craw, 2006; et al., 2006) and other and as and 2006) and and 2006). the fossil record provides evidence for a of Araucariaceae species that have in Australia during the million years Hill and Brodribb, 1999). the of these species to extant Agathis species A of analyses based on molecular data is that dispersal is not Divergence time estimates can suggest or reject of they cannot reject estimates most gene trees are and support the of it is to dispersal as an In the of the there was fossil evidence that unambiguously the same throughout the Oligocene, and in New this make dispersal because of the same to be In fossil evidence in the described above is However, nuclear studies by et al. suggest the in many New Zealand Oligocene, and Agathis australis the hypothesis of an Oligocene of New Zealand is by recent and molecular et al. (2006) described an from the of New consideration for the of New the dispersal of the and its from the fossil record of et al. (2006) have suggested that the fossil a lineage that has survived the Oligocene in New Cooper et al. and Ericson et al. (2002) have conducted phylogenetic analyses on birds and In these New Zealand of these different are on sister to a considered to be a of and 2006). However, molecular clock arguments were used in both studies to that trans-Tasman sea relationships are as a of the breakup of the of reliable fossil calibrations for these has also been (Waters and Craw, 2006). In the divergence times used in our of Araucariaceae lineages are based on fossils from different which are The conservative of our calibrations explain our dates estimated for the divergence of New Zealand and Australian lineages are younger than the dates estimated in bird studies by Cooper et al. and Ericson et al. (2002). In our phylogenetic estimates are with the from these studies and that of et al. which suggest of the New Zealand landmass throughout the They not reject the hypothesis that trans-Tasman distributions of Araucariaceae are by To there have been very identified the of the of in the New Zealand throughout the However, there was a very significant reduction in to a of as suggested by then the likelihood of many lineages to the be is not that New Zealand is in et al., molecular and studies on are needed to test the hypothesis of an Oligocene drowning of New Zealand analysis of the of the fossil Araucariaceae from Clarence Valley, is of major If it can be to be known of Agathis australis as has then this together with of the of similar fossils from the Australian fossil record, be a for the of Agathis australis in New Zealand following the of Australia and New Zealand Ma. is evidence from fossil data and molecular to the of Late Tertiary transoceanic for of the New Zealand biota 1994; Macphail, 1997; Winkworth et al., 2002; Knapp et al., 2005). However, Agathis an evolutionary to be and on and New Zealand Araucariaceae fossils not support the hypothesis that New Zealand was completely submerged during the Oligocene (Waters and Craw, 2006). evidence also cannot reject the hypothesis of in situ of at least some plant and lineages in New Zealand following the breakup of Gondwana. We for with the Royal Botanical in Sydney, the Botanical in Valdivia, and the Esplanade in Palmerston North, New for and the of New for the for Agathis We for with DNA and as well as and for and on the which were We also and for to data and for with the We to the New Zealand and the for our study. was supported by of Massey University and the