TAXON 58 (4) • November 2009: 1153–1167 Rohwer & al. • Is Persea monophyletic? Is Persea (Lauraceae) monophyletic? Evidence from nuclear ribosomal ITS sequences Jens G. Rohwer1, Jie Li2, Barbara Rudolph1, Sabrina A. Schmidt1, Henk van der Werff 3 & Hsi-wen Li2 1 Biozentrum Klein Flottbek und Botanischer Garten, Universität Hamburg, Ohnhorststr. 18, 22609 Hamburg, Germany. rohwer@botanik.uni-hamburg.de (author for correspondence) 2 Laboratory of Plant Phylogenetics and Conservation Biology, Xishuangbanna Tropical Botanical Garden, The Chinese Academy of Sciences, 88 Xuefu Rd., Kunming, Yunnan 650223, P.R. China 3 Missouri Botanical Garden, P.O.Box 299, St. Louis, Missouri 63166-0299, U.S.A. The delimitation of genera within the Persea group (Lauraceae) has always been controversial. In an attempt to resolve the phylogenetic lines within this group, we analyzed ITS sequences of 61 species of the Persea group (Lauraceae) and 30 other species of Lauraceae. Several of the traditional genera or subgenera form well-supported groups, viz., Persea subg. Eriodaphne, Machilus, Persea subg. Persea, and Alseodaphne including Dehaasia. The included species of Phoebe form two clades that are unresolved with respect to Alseodaphne. However, Persea subg. Eriodaphne (together with the Macaronesian Apollonias barbujana) forms one of the clades of an unresolved basal trichotomy within the Persea group, whereas Persea subg. Persea is well supported as member of an otherwise Asian clade including Alseodaphne and Phoebe. This indicates that Persea, as currently circumscribed, is not monophyletic. The affinities of the Macaronesian Persea indica are clearly American rather than Asian. KEYWORDS: ITS, Lauraceae, neotropics, paleotropics, Persea group, phylogeny INTRODUCTION The Persea group, with about 400 to 450 species, is a subset of the Lauraceae, consisting of the currently often recognized genera Alseodaphne Nees, Apollonias Nees, Dehaasia Blume, Machilus Nees, Nothaphoebe Blume, Persea Mill. and Phoebe Nees. Most of them are mainly distributed in the tropics to subtropics of Asia, while Persea has about 90 species in subtropical to tropical America. Apollonias and Persea each have one species in the Macaronesian Islands (Canary Islands and Madeira, Persea indica also on the Azores). The genus Mutisiopersea, described by Kostermans (1993) to accommodate a number of American species previously placed in Persea subg. Eriodaphne Nees, so far did not gain widespread acceptance (see van der Werff, 2002). The Persea group was first recognized almost in its present circumscription by Rohwer (1993, as Persea subgroup) based on morphological evidence, except for the explicitly provisional inclusion of the aberrant genus Caryodaphnopsis Airy-Shaw. The group largely corresponds to Kostermans’ (1957) subtribe Perseineae of tribe Perseeae, except that Kostermans referred the genera with disporangiate anthers, Apollonias and Dehaasia, to the subtribe Beilschmiediineae, which also included several genera now known to belong to other evolutionary lineages. All non-cupulate genera of van der Werff & Richter’s (1996) Perseeae are members of the Persea group. The species in this group have paniculate-cymose inflorescences, in which the lateral flowers of a cyme are strictly opposite. In most species the staminodes of the fourth androecial whorl are distinct, often with a glandular head. The fruit is not seated in a cupule, but sometimes on a swollen pedicel. Recent molecular studies confirmed that the Persea group (excluding Caryodaphnopsis) is monophyletic (Rohwer, 2000; Chanderbali & al., 2001; Rohwer & Rudolph, 2005), but the generic delimitation in the group has always been controversial. Bentham (1880) treated Alseodaphne, Nothaphoebe and Phoebe as sections within Persea (keeping Machilus separate), whereas Kostermans (1957) placed Alseodaphne, Machilus and Nothaphoebe (plus Caryodaphnopsis, see above) in the synonymy of Persea (keeping Phoebe separate). Later, however, he reinstated Alseodaphne and Nothaphoebe (Kostermans, 1973a), as well as Caryodaphnopsis (Kostermans, 1974), but he retained Machilus within Persea. Other authors (e.g., Li & al., 1984) continued to use Machilus as an accepted genus. Rohwer (1993) followed Kostermans in treating Machilus as a subgenus of Persea, but he remarked that the Macaronesian species, Persea indica, was “possibly misplaced in this genus”, and that the two species of Apollonias (one from Macaronesia, the other from India) were “probably independently derived from Phoebe.” Van der Werff (2001) also used the generic name Persea in a sense including Machilus. He 1153 Rohwer & al. • Is Persea monophyletic? placed Nothaphoebe in the synonymy of Alseodaphne because none of the differences between them appeared to be discontinuous and concepts of the two genera had always been rather vague. In addition, he expressed doubts about the delimitation between Persea and Phoebe, and between Alseodaphne and Dehaasia. On the basis of their disporangiate anthers, Apollonias and Dehaasia used to be kept separate from the other genera (with predominantly tetrasporangiate anthers), often at different infrafamilial levels varying from subtribe (Kostermans, 1957) to subfamily (Pax, 1889). On the other hand, Kostermans (1973b) noted that “Dehaasia and Alseodaphne are very closely related; the only difference is the number of anther cells, 2 in the former, 4 in the latter.” Van der Werff (2001) came to the same conclusion. This statement also receives some support from an analysis based on trnK intron sequences (Rohwer & Rudolph, 2005), in which Alseodaphne perakensis (Gamble) Kosterm. and Dehaasia cuneata (Blume) Blume form a strongly supported clade. Kopp (1966) already accepted species with disporangiate and with tetrasporangiate anthers within Persea subg. Eriodaphne, sometimes as closest relatives to one another. She also documented different numbers of pollen sacs in the different androecial whorls within the same flower, and even a variable number of pollen sacs in the stamens of the third whorl of P. urbaniana Mez. In recent years it has become increasingly clear that number of pollen sacs is not sufficient to distinguish genera, and may sometimes vary within species (Rohwer & al., 1991; Rohwer, 1993; Li & Christophel, 2000; van der Werff, 2001; Chanderbali, 2004; Li & al., 2004). Other important characters for generic delimitation in the Persea group are (1) the structure of the perianth (equal vs. unequal tepals) and (2) its fate in fruit (deciduous vs. persistent; indurate or not; reflexed, spreading or clasping the base of the fruit), as well as (3) the structure of the pedicel in fruit (see Kostermans, 1957; Rohwer, 1993; van der Werff, 2001). The tepals (1) are supposed to be equal or subequal in Apollonias, Machilus, Phoebe and Persea subg. Persea, while they are described as more or less unequal in Alseodaphne, Dehaasia, Nothapoebe and Persea subg. Eriodaphne. However, this is by no means a sharp discontinuity. The illustrations for Alseodaphne in Kostermans (1973a) show scarcely to moderately unequal tepals (outer ones reaching about 2/3 of the length of the inner ones), whereas those for Dehaasia in Kostermans (1973b) show scarcely to strongly unequal tepals (outer ones reaching about 1/3 of the length of the inner ones). Personal observations on Apollonias (by JGR, based mainly on herbarium material, plus a few old inflorescences of A. barbujana on Tenerife) revealed a similar degree of variation even within this species, from scarcely different to outer 1154 TAXON 58 (4) • November 2009: 1153–1167 tepals reaching only slightly more than half the length of the inner tepals. The same is found in cultivated Persea americana (P. subg. Persea ; Fig. 1A, B), which is often keyed out using statements like “tepals (sub)equal.” Therefore, this character should be used only with caution for the delimitation of genera. The fate of the perianth in fruit (2) appears to be a more reliable character, but it isn’t entirely discontinuous, either. The tepals are usually described as deciduous in fruit in Alseodaphne, Dehaasia, and Persea subg. Persea, whereas they are persistent in Apollonias, Machilus, Phoebe, and Persea subg. Eriodaphne (Fig. 1C–E). In Nothaphoebe, they are described as either deciduous or persistent, but small and inconspicuous. Indurate (i.e., more or less lignified) tepals in fruit are characteristic of Apollonias, Phoebe and Persea subg. Eriodaphne. In Apollonias (Fig. 1D) and Phoebe they are appressed to the base of the fruit, whereas in Persea subg. Eriodaphne they are often (but not always, see Fig. 1C) spreading. The tepals of Machilus are sometimes slightly enlarged but not significantly thickened in fruit (Fig. 1E), and in herbarium specimens they are often shriveled and reflexed. These differences, however, are not as clear as they may seem. Quite often, the tepals do not fall off as a whole, but their bases persist while their tips break off. Then it depends on the relative size of the deciduous and the persistent parts, as well as on the size of the fruit, whether they are regarded as persistent (as in Apollonias barbujana ; Fig. 1D) or as deciduous (as in Persea americana ; Fig. 1F, where the tepal bases persist as well). In addition, a few species in each of the larger genera do not conform to the most frequent pattern of their group. In Persea, e.g., most species have unequal tepals persistent in fruit (= P. subg. Eriodaphne), while a few have (supposedly) equal to (sub) equal tepals deciduous in fruit (= P. subg. Persea) or equal tepals persistent in fruit (e.g., P. albida Kostermans and P. rigens C.K. Allen); one species (Persea julianae van der Werff) has strongly unequal tepals deciduous in fruit (van der Werff, 1989, 2001, 2002; Rohwer & al., 1991). The pedicel (3) changes relatively little during fruit development in most genera, except that it usually gets somewhat stronger, depending on the size of the fruit. Thickened, fleshy, and vividly colored fruit pedicels are characteristic of Alseodaphne and Dehaasia, but already Kostermans (1973a, b) noted that in some species they are scarcely thickened. This also seems to be the most frequent condition in Nothaphoebe, which had been included in Alseodaphne by van der Werff (2001). Vividly colored pedicels, however, do also occur in Apollonias, Machilus, and Persea, and occasionally they are slightly thickened as well. In summary, the traditional morphological characters used to define genera in the Persea group are not fully satisfactory. Therefore, it seems most promising to further TAXON 58 (4) • November 2009: 1153–1167 Rohwer & al. • Is Persea monophyletic? Fig. 1. A + B, flowers of Persea americana. A, flower with subequal tepals, female phase, anther valves still closed; Bot. Gard. Hamburg. B, flower with distinctly unequal tepals, male phase, anther valves open; Parque Etnografico Piramides de Güimar, Tenerife. C–F, fruits with persistent perianth. C, undetermined species of Persea subg. Eriodaphne (courtesy of Maximilian Weigend, MW 8624, Peru). D, Apollonias barbujana; near Taganana, Tenerife. E, Machilus grijsii; Bot. Gard. Hamburg. F, Persea americana; merchandise. All photos except C by JGR. expand the range of characters by including molecular evidence. Although sequences from cpDNA matK showed higher levels of sequence evolution and better resolution of species- and generic-level phylogeny in a range of plant families (Johnson & Soltis, 1994, 1995; Steele & Vilgalys, 1994; Hilu & Liang, 1997), Rohwer’s (2000) work revealed that the sequence divergence among Lauraceae taxa was surprisingly low (only 9.7% informative characters in the family as a whole, and less then 1% in the terminal and Persea groups). Therefore, the greatest challenge seems to be to find an appropriate marker for this purpose. 1155 Rohwer & al. • Is Persea monophyletic? Recent molecular work on the entire family Lauraceae by Chanderbali & al. (2001) has shown that sequence variation in chloroplast markers (trnL-trnF, psbA-trnH, trnT-trnL, rpl16) and partial 26S from the nuclear genome was almost limited to basal branches in Lauraceae, with very little and often no variation among members of the tribes Perseeae and Laureae. Only the nuclear internal transcribed spacer (ITS) showed substantial variation within these groups. Therefore, ITS was expected to yield phylogenetic resolution in the Persea group as well. In addition, we tried a number of other markers (see below) in order to evaluate their potential for resolving relationships within this group. Previous molecular studies in Lauraceae (Rohwer, 2000; Chanderbali & al., 2001; Rohwer & Rudolph, 2005) were aimed at resolving the basic relationships within the family. Therefore, they included only a relatively small sample (6–10 spp.) of species from the Persea group. Although our taxon sample (70 sequences of 51 determined and 10 undetermined species; Appendix) is much larger, the present paper should still be considered an exploratory study in view of the large size of the group. MATERIALS AND METHODS Taxon sampling. — As indicated above, 61 (bona fide) species of the Persea group were included in the present study (Appendix). An attempt was made to include a broad representation of all genera and sections in bigger genera such as Machilus and Phoebe according to the treatment in Flora Reipublicae Popularis Sinicae vol. 31 (Li & al., 1984) as well as in Persea in the sense of Kopp (1966). However, taxon sampling was also guided by availability of plant material and success in amplification and sequencing. Determinations of the plant material were either made or checked for their plausibility by H.W. Li for the Chinese species, and by either J.G. Rohwer or H. van der Werff for the American, Macaronesian, and the remaining Asian species. In material without reproductive structures, however, some determinations remained tentative. Two previously published sequences of species of the Persea group, Apollonias barbujana (EMBL accession number AF272257) and Dehaasia incrassata (AF272268) were excluded from our matrix because of serious doubts about their correctness (see results). In addition to the species of the Persea group, we included 30 species of the closely related Laureae-Cinnamomeae clade in the sense of Rohwer & Rudolph (2005), in order to see if any of the species was misplaced in the Persea group. This was done mainly because the American species now placed in Cinnamomum Schaeff. as well as several species now placed in Ocotea Aubl. (Cinnamomeae), among them O. helicterifolia, were included in 1156 TAXON 58 (4) • November 2009: 1153–1167 Phoebe (Persea group) by Mez (1889). The former were moved to their current position by Kostermans (1961), whereas the position of the latter had been discussed in detail by Rohwer (1986, 1991). Eleven species of the Persea group as well as eight species of the Laureae-Cinnamomeae clade were represented by two collections each, in order to give us an estimate of the intraspecific variability of the ITS region. If the sequences proved identical, however, only one of them was included in the final analysis. Based on the recent molecular studies (Rohwer, 2000; Chanderbali & al., 2001; Rohwer & Rudolph, 2005), five species from the Mezilaurus group (Anaueria brasiliensis, Chlorocardium venenosum, Chlorocardium rodiei, Mezilaurus triunca, Sextonia pubescens) were selected as the outgroup. DNA extraction. — Total genomic DNA was extracted from silica-gel dried material or herbarium specimens following Doyle & Doyle (1987) as modified by Li & al. (2004), or using the Invisorb® Spin Plant Mini Kit (Invitek) with modifications described by Rohwer & Rudolph (2005). PCR amplification and sequencing. — The ITS and 5.8S regions were amplified following the general methodology (White & al., 1990) and PCR primers of Chanderbali & al. (2001) with minor modifications as reported in Li & al. (2004). If the quality of the material allowed, we also often used the primers ITS18-F (5′-GTC CAC TGA ACC TTA TCA TTT AGA GG-3′; Käss & Wink, 1997, modified by Beyra-Matos & Lavin, 1999) and ITS26-R (5′-GCC GTT ACT AAG GGA ATC CTT GTT AG-3′; Käss & Wink, 1997), which amplify a larger part of the ITS region. In order to reduce problems caused by the secondary structure of the rather GC-rich ITS, 10% DMSO was included in all amplifications (Buckler & Holtsford, 1996; Buckler & al., 1997). In addition to ITS, we tried several other markers (see Table 1). These, however, yielded so little information that we did not consider them useful for our phylogenetic analysis. All PCR amplifications included negative controls to detect contamination. PCR products were subsequently purified using the QIAquick® PCR Purification Kit (Qiagen) or Montage™ PCR Centrifugal Filter Devices (Millipore). Purified amplification products were sequenced forward and reverse using ABI PRISM® BigDye™ terminator cycle sequencing. The sequencing reaction products were separated using an ABI 3100 (Applied Biosystems) automated sequencer at Kunming or an ABI Prism 3770 Capillary Sequencer (Applied Biosystems) at the University Hospital Eppendorf (UKE, Hamburg), Institute of Cell Biochemistry and Clinical Neurobiology. Sequence analysis. — Sequence chromatogram output files for each species were initially aligned and edited base by base with Sequencher™ (Gene codes Corporation, TAXON 58 (4) • November 2009: 1153–1167 Rohwer & al. • Is Persea monophyletic? Inc. 1994). Sequences from both strands (forward and reverse) were compared to guarantee accurate base calling. The completed sequences of the different species were then aligned using Sequencher™, BioEdit (Hall, 1999) and/or ClustalX, version 1.8 (Thompson & al., 1997), and manual adjustments were made as judged necessary. All ITS sequences analyzed in this study were deposited in GenBank under the accession numbers provided in the Appendix, and the aligned data matrix was submitted to TreeBASE with accession number M4361. When two taxa turned out to have identical ITS sequences, only one of them was included in the phylogenetic analysis. Therefore, there are fewer terminal taxa in Fig. 2 than are listed in the Appendix. Phylogenetic analysis. — Data analyses employed either maximum parsimony (MP) using PAUP* 4.0b10 (Swofford, 2001) or Bayesian inference (BI) using MrBayes v3.1.2 (Huelsenbeck & Ronquist, 2001; Ronquist & Huelsenbeck, 2003; see also http://morphbank .ebc.uu.se/mrbayes/manual.php). Maximum parsimony. – Equally weighted MP heuristic searches with 100 addition sequence replicates and tree bisection-reconnection (TBR) were conducted. All MP analyses were performed with MULTREES on, ACCTRAN optimization and gaps treated as missing data. Positions in which the alignment might have been ambiguous were excluded, but if the excluded positions were potentially informative, they were coded in an indel matrix (TreeBASE M4360). In the data presented here, all indel characters were treated as unordered. Previous experiments with asymmetric step matrices for the transitions between different character states (based on estimates of minimally required evolutionary steps) had greatly inflated the calculation time without significantly improving the results (not shown). In order to avoid excessive swapping on suboptimal trees, the maximum number of trees two or more steps longer than the shortest trees that were found in preliminary trials was set to 10,000. Clade support was estimated using fast bootstrap analyses with 10,000 bootstrap replicates (each with 100 addition sequence replicates), but without branch swapping (Mort & al., 2000), after tests on earlier versions of the matrix had shown that they yielded identical topologies and very similar bootstrap values as full bootstrap analyses with branch TBR swapping. Bayesian inference. – For the Bayesian inference of phylogeny the data were separated into four unlinked partitions, for the ITS1 region, the ITS2 region, all sequence parts coding for ribosomal RNA (5.8S rDNA and a small piece of the 26S rDNA), and the indel matrix. As suggested in the MrBayes manual, the default priors were used for the analysis. As MrBayes 3.1.2. allows using only a single species as outgroup, we specified each of the five species from the Mezilaurus group (Anaueria brasiliensis, Chlorocardium rodiei, Chlorocardium venenosum, Mezilaurus triunca, Sextonia pubescens) as outgroup for a separate run of the final analysis. In each of these analyses, two simultaneous runs of four Markov chain Monte Carlo chains each were run for 1,000,000 generations, with the current tree saved every 100 generations. Thus we generated ten probability files (.p) and ten tree files (.t), two for each outgroup taxon. Visual inspection Table 1. Additional markers explored for the Persea group. Locus Sequencing Species compared primer to Persea indica Aligned No. of bp (ca.) differences Difference in same length ITS Remarks accD accD-1f Laurus azorica 360 4 (1.11%) 33 ndhJ ndhJ-1f Laurus azorica 380 0 33 psbC-trnS psbC Laurus azorica 630 1 (0.16%) 44 psbC-trnS trnS Laurus azorica 600 5 (0.83%) 43 rpoB rpoB-2f Laurus azorica 470 2 (0.42%) 37 rpoC1 rpoC1-1f 5′rpoC1-3′rpoC1 3′rpoC1 Laurus azorica 560 1 (0.18%) 41 Apollonias barbujana 550 0 34 One 7-bp inversion Plus a few uncertainties rpsF-rpsR2 rpsF Apollonias barbujana 550 0 34 rpsF-rpsR2 rpsF Persea americana 550 1 (0.18%) 45 Poly-A 3 pb longer One 18-bp indel, 2 single base indels trnD-trnT2 trnD Laurus azorica 240 3 (1.25%) 29 trnD-trnT2 trnT2 Laurus azorica 350 3 (0.86%) 33 trnH-trnK trnH Laurus azorica 140 5 (3.57%) 17 trnH-trnK trnK Laurus azorica 670 5 (0.75%) 47 YCF5 YCF5-2f Machilus grijsii 380 4 (1.05%) 31 No longer readable after poly-A 1157 TAXON 58 (4) • November 2009: 1153–1167 Rohwer & al. • Is Persea monophyletic? * 81 92 76 * 88 67 65 99 87 93 72 * 69 80 97 * * 72 75 * 96 52 68 60 91 74 90 97 * 84 96 82 95 63 98 75 89 63 63 52 96 84 to outgroup 86 74 * 65 99 63 99 * S Am C Am Persea subg. Persea Asia S Am Asia South America * Alseodaphne incl. Dehaasia 83 Phoebe 56 99 56 99 63 96 86 Alseodaphne nigrescens #1 Nothaphoebe umbelliflora Alseodaphne nigrescens #2 Alseodaphne sp. A38 Dehaasia sp. A34 Dehaasia sp. A37 Alseodaphne sp. W17084 Dehaasia cuneata Alseodaphne semecarpifolia Phoebe formosana Phoebe sheareri #1 Phoebe sheareri #2 Nothaphoebe cavaleriei #1 Nothaphoebe cavaleriei #2 Phoebe neurantha Phoebe lanceolata Phoebe excelsa Phoebe megacalyx Phoebe macrocarpa Phoebe forrestii Phoebe puwenensis Persea nudigemma Persea americana #1 Persea americana #2 Persea steyermarkii Persea schiedeana Alseodaphne sp. W14264 Persea sphaerocarpa Persea sp. W21874 Alseodaphne andersonii Alseodaphne petiolaris Persea caerulea #1 Persea caerulea #2 Persea venosa Persea willdenovii Persea aff. subcordata Persea pajonalis Persea lingue #1 Persea lingue #2 Persea mutisii Persea sp. W14857 Persea sp. W19517 Persea sp. V25232 Persea meridensis Persea weberbaueri Persea alba Persea borbonia Persea palustris #1 Persea palustris #2 Persea veraguasensis Persea indica Persea areolatocostae #1 Persea areolatocostae #2 Apollonias barbujana Machilus sp. W14071 Machilus thunbergii #1 Machilus thunbergii #2 Machilus sp. W14068 Machilus zuihoensis Machilus obovatifolia Machilus japonica Machilus rimosa Machilus chienkweiensis Machilus decursinervis Machilus grijsii Machilus shweliensis Machilus yunnanensis Machilus minkweiensis N Am Persea subg. Eriodaphne 81 C Am Mac S Am Mac Asia 73 71 85 Machilus A First this was done for all ten tree files separately, but as they did not show any significant differences (neither conflicting topologies nor differences of more than ten percentage points from the average in the support values of individual clades), the results of all ten files were ultimately combined. Asia of the likelihood values in Excel showed that all analyses converged to stable likelihood values after about 300 to 500 generations. Therefore, the first 501 of 10,001 saved trees were discarded as burn-in. The posterior probabilities for the individual clades were computed by creating a majority rule consensus of the remaining trees in PAUP*. * Fig. 2. Cladogram retrieved from the Bayesian analysis. A (this page), Persea group; B (opposite page), outgroup, Cinnamomeae and Laureae. Numbers above branches are fast bootstrap values, bold numbers below branches are posterior probabilities. ○ = 100% posterior probability in Bayesian analysis only, ● = both posterior probability and bootstrap value 100%. Dotted lines represent clades present in some runs but collapsed in the consensus. Asterisks mark clades and species conflicting with either the maximum parsimony or the fast bootstrap consensus (see text). C Am = Central America, Mac = Macaronesia, Med = Mediterranean, N Am = North America, S Am = South America. 1158 TAXON 58 (4) • November 2009: 1153–1167 Rohwer & al. • Is Persea monophyletic? * * 73 83 * 53 57 99 99 90 53 92 60 98 50 77 * * 50 74 87 98 87 97 80 84 89 Asia Med Mac Med Laureae 72 Neolitsea sericea #1 Neolitsea sericea #2 Neolitsea levinei Actinodaphne maingayi Actinodaphne sesquipedalis Laurus nobilis #2 Laurus azorica Laurus nobilis #1 * Litsea accedens Litsea glutinosa Lindera benzoin #1 Lindera benzoin #2 Parasassafras confertiflora #1 * Parasassafras confertiflora #2 Endlicheria gracilis Rhodostemonodaphne crenaticupula Ocotea pulchella Nectandra amazonum #1 Nectandra amazonum #2 Umbellularia californica #1 Umbellularia californica #2 Ocotea foetens Licaria canella Ocotea helicterifolia Cinnamomum duartianum Cinnamomum quadrangulum Cinnamomum amoenum Cinnamomum burmanii Cinnamomum daphnoides Sassafras albidum Sassafras tzumu Chlorocardium rodiei Chlorocardium venenosum Anaueria brasiliensis Mezilaurus triunca Sextonia pubescens Asia N Am Asia N Am Mac S Am Cinnamomeae 95 98 Asia N Am Asia Mezilaurus group to Persea group South America B South America The total length of the aligned ITS region (including the 5.8S unit) in the taxa investigated was 821 base pairs (bp), of which 168 were excluded from the analyses for reasons indicated above (see Phylogenetic analyses). The length of the unaligned ITS sequences (excluding the often unequally represented 26S part) varied from 554 bp (Cinnamomum quadrangulum) to 648 bp (Anaueria brasiliensis). The GC contents varied from 68.7% (Lindera benzoin) to 75.9% (Alseodaphne nigrescens). Among the non-excluded 653 bp that constitute the ITS dataset, 360 (55%) were constant, 78 (12%) were variable but parsimony-uninformative, and 215 (33%) were parsimony-informative. Of the informative positions, 118 were found in the ITS1, 8 in the 5.8S rDNA, 81 in the ITS2, and 8 in the 26S rDNA. The gap matrix consisted of 21 binary and 34 unordered multistate characters, which were of course all informative. Nevertheless, two of them were excluded from the final analysis because they were defined as alternative codings for characters already represented in the gap matrix. Pairwise comparisons between the taxa showed that the number of total character differences was usually roughly similar in the ITS1 and the ITS2, with two conspicuous exceptions: the sequences of Apollonias barbujana and Dehaasia incrassata retrieved from EMBL (accession numbers AF272257 and AF272268, respectively) turned out to be identical with Machilus thunbergii (AF272327) in the ITS2, but in the ITS1 they differed from that species by 18 and 13 bp, respectively. This led to very different positions in the cladograms retrieved from separate parsimony analyses of the ITS1 and ITS2 (not shown), whereas the groups formed by other taxa South America RESULTS 1159 Rohwer & al. • Is Persea monophyletic? changed relatively little (if at all). This prompted us to sequence another collection of Machilus thunbergii and two further collections of Apollonias barbujana. Unfortunately, our attempts to sequence another collection of Dehaasia incrassata failed because of poor DNA quality from the herbarium material, but we were able to get good sequences from other, undetermined collections of Dehaasia. Our sequence of Machilus thunbergii turned out to be very similar to the sequence from EMBL, whereas the two sequences of Apollonias barbujana (one from Tenerife, the other from La Palma) were very different from the EMBL sequence in the ITS2 but identical among themselves (therefore only one of them is included in the final analysis). The sequences of our Dehaasia collections were also very different from the EMBL sequence in the ITS2, but very similar in the ITS1. Therefore, we assume that the EMBL sequences of Apollonias barbujana and Dehaasia incrassata are a result of contamination (with Machilus thunbergii or a similar species of Machilus) and should not be used in further analyses. Other markers that we tried in addition to ITS did not yield promising amounts of variation. Pairwise comparisons between Persea indica and a second species, either from within the Persea group (Apollonias barbujana or Machilus grijsii) or from the Laureae (Laurus azorica) showed very few (if any) differences (Table 1). Maximum parsimony analysis. — Parsimony analysis of the ITS data resulted in 51,664 most parsimonious trees in a single island, 1,262 steps long, with a retention index of 0.751 and a consistency index of 0.456. In addition, 38 of the 100 searches terminated on islands requiring a higher number of steps (6× length 1,263 in six different islands, 30× length 1,264, 2× length 1,265; the latter two not swapped to completion). The majority rule consensus of the 51,664 trees (not shown) differed in the composition of a few clades from the results of the Bayesian analysis (Fig. 2, marked with an asterisk), but the maximum fast bootstrap support (FBS) for any of these conflicting clades never exceeded 51%. Therefore, we show only the result of the Bayesian analysis (Fig. 2) and describe the differences below. Bayesian analysis. — As indicated above, the topology retrieved from the Bayesian analysis was largely congruent or at least compatible with the majority rule consensus from the parsimony analysis. The outgroup is consistently separated from the core Lauraceae (100% posterior probability, PP), and within the core Lauraceae the Persea group forms a well-defined clade (100% PP). Its sister group, however, the Laureae-Cinnamomeae clade, receives only very weak support (60% PP), while the Laureae and the Cinnamomeae individually both get 100% PP. As it was not the aim of this study to examine the relationships within these groups, we do not describe the topology within them any further. 1160 TAXON 58 (4) • November 2009: 1153–1167 Within the Persea group, there are three well-supported principal clades (except for one species, Apollonias barbujana, see below). Based on their main constituents, the three clades will in the following be called (1) the Alseodaphne-Phoebe clade, (2), the Eriodaphne clade, and (3) the Machilus clade. (1) The Alseodaphne-Phoebe clade consists not only of all representatives of these two genera included in the present study, but also of species of Dehaasia, Nothaphoebe (but see Discussion), and a small number of Persea species. Among the three principal clades within this group, only the species pair Alseodaphne andersonii/A. petiolaris receives strong support (100% PP), whereas support for the other two clades is (relatively) weak. The larger of these two clades consists mainly of Asian species (of Alseodaphne, Dehaasia, Phoebe, and possibly Nothaphoebe), which form a relatively well-supported group (92% PP). Persea nudigemma from Ecuador is weakly supported (67% PP) as sister species to this group. Among the Asian species, there are again three well supported clades (100% PP), one consisting of species of Alseodaphne and Dehaasia, the other two consisting of Phoebe species. The second main clade within the Alseodaphne-Phoebe group consists mainly of American species of Persea, with the exception of an undetermined Alseodaphne species from Vietnam. Both nodes below this species, however, receive only relatively weak support (75% and 72% PP, respectively). The only really well supported group within this clade is Persea subg. Persea (100% PP). (2) The Eriodaphne clade consists mainly of American species of Persea subg. Eriodaphne, plus Persea indica from the Canary Islands. All these Persea species form a well-supported group (100% PP). Apollonias barbujana from the Canary Islands is shown as sister species to this group, albeit with rather insignificant support (63% PP). Several subclades among the Persea species are well supported, but the nodes below Persea indica receive only very weak support (52 and 63% PP, respectively). (3) The Machilus clade consists exclusively of species of the Asian genus Machilus. It shows very little internal resolution, due to the fact that there are relatively few pairwise nucleotide differences among these species (see below). Differences between maximum parsimony and Bayesian analysis. — Although the results of both analyses are largely congruent or at least compatible, there are a few differences between the majority rule consensus of the maximum parsimony analysis (one of the 51,664 trees shown in Fig. 3) and the Bayesian analysis (marked with an asterisk in Fig. 2). Unless otherwise indicated, none of these alternative groupings reached 50% fast bootstrap support. Among the Laureae, Lindera benzoin and Parasassafras confertiflora appear as sister groups in the parsimony analysis, and both together as sister to the TAXON 58 (4) • November 2009: 1153–1167 Rohwer & al. • Is Persea monophyletic? Phoebe formosana Phoebe sheareri #1 Phoebe sheareri #2 Phoebe lanceolata Phoebe neurantha Nothaphoebe cavaleriei #1 Nothaphoebe cavaleriei #2 Phoebe macrocarpa Phoebe megacalyx Phoebe excelsa Phoebe forrestii Phoebe puwenensis Persea nudigemma Alseodaphne nigrescens #1 Nothaphoebe umbelliflora Alseodaphne nigrescens #2 Alseodaphne sp. A38 Dehaasia sp. A34 Dehaasia sp. A37 Alseodaphne sp. W17084 Dehaasia cuneata Alseodaphne semecarpifolia Persea sphaerocarpa Persea sp. W21874 Alseodaphne andersonii Alseodaphne petiolaris Alseodaphne sp. W14264 Persea americana #1 Persea americana #2 Persea steyermarkii Persea schiedeana Machilus sp. W14071 Machilus thunbergii #1 Machilus thunbergii #2 Machilus sp. W14068 Machilus zuihoensis Machilus obovatifolia Machilus decursinervis Machilus yunnanensis Machilus shweliensis Machilus grijsii Machilus rimosa Machilus japonica Machilus minkweiensis Machilus chienkweiensis Persea caerulea #1 Persea caerulea #2 Persea venosa Persea willdenovii Persea lingue #1 Persea lingue #2 Persea aff. subcordata Persea pajonalis Persea mutisii Persea sp. W14857 Persea sp. W19517 Persea sp. V25232 Persea meridensis Persea weberbaueri Persea alba Persea borbonia Persea palustris #1 Persea palustris #2 Persea veraguasensis Persea indica Apollonias barbujana #2 Persea areolatocostae #1 Persea areolatocostae #2 1 step Fig. 3. Phylogram of one of the 51,664 most parsimonious trees retrieved from the maximum parsimony analysis, Perseagroup only, illustrating the alternative topology and the widely differing branch lengths. Actinodaphne-Neolitsea clade. In addition, the two collections of Laurus nobilis receive rather weak support (51% FBS) as monophyletic group in the bootstrap consensus. Among the Cinnamomeae, Umbellularia californica appears as sister species to the clade consisting of Endlicheria gracilis, Rhodostemonodaphne crenaticupula, Ocotea pulchella and Nectandra amazonum, leaving Ocotea foetens and Licaria canella unresolved. Within the Persea group, Machilus becomes sister to the Alseodaphne-Phoebe clade, leaving the Eriodaphne clade as the basal group. Within the Eriodaphne clade, Persea indica appears as sister species to P. areolatocostae, and P. lingue swaps its position with the clade consisting of P. pajonalis and P. aff. subcordata, so that it becomes sister species to the clade consisting of P. caerulea, P. venosa and P. willdenovii (59% FBS). Within the Machilus clade, M. chienkweiensis swaps its position with M. minkweiensis. Within the Alseodaphne-Phoebe clade, there are several rearrangements. Persea subg. Persea becomes the basal clade, followed by a clade consisting of Alseodaphne 1161 Rohwer & al. • Is Persea monophyletic? andersonii, A. petiolaris and the undetermined Alseodaphne species W14264. Persea sphaerocarpa and the undetermined Persea species W21874 form the next clade, followed by the Alseodaphne-Dehaasia clade. Persea nudigemma becomes sister species to an apparently monophyletic Phoebe. Among the Phoebe species, Phoebe lanceolata appears as sister species to Phoebe neurantha, and both together as sister to the Phoebe formosana/ Ph. sheareri clade. DISCUSSION The present molecular analysis for the first time reveals significantly supported and phylogenetically meaningful clades within the Persea group. Although one of the main questions that prompted us to start this study (viz., the circumscription of the genus Persea) could not be solved completely, we can already provide significant information about the phylogenetic structure of the Persea group, which will serve as an orientation for further studies. Several of the traditional genera or subgenera are retrieved as monophyletic groups in our analysis. Machilus. — This genus is by far the most homogeneous group in our molecular data. Pairwise total character differences between species of Machilus range from 0 (between M. oreophila and M. zuihoensis) to 13 (between M. chienkweiensis and M. thunbergii). In other genera represented by more than two species, the differences are on the average much larger (Alseodaphne 6–54; Cinnamomum 5–42; Ocotea 26–31; Persea subg. Eriodaphne 2–39; Phoebe 1–32). This might suggest that the species differentiation within Machilus was rather recent, compared to other Lauraceae such as Cinnamomum (Groth, 2003), Endlicheria Nees and Ocotea (Chanderbali, 2004), which show a much higher degree of divergence. However, the phylogram of one of the most parsimonious trees (Fig. 3) rather suggests that Machilus may have a decreased substitution rate, whereas the substitution rate of the Alseodaphne clade appears to be increased compared to the other genera. Our preliminary data as well as a previous taxonomic treatment (Li & al., 1984) suggest that some of the currently accepted sections or subsections of Machilus may be questionable. The only group within Machilus receiving a high posterior probability in the Bayesian analysis, i.e., the clade including Machilus thunbergii, M. obovatifolia, M. oreophila (omitted in Fig. 2 because sequence identical with the following), M. zuihoensis and the two undetermined species W14068 and W14071, thus includes (at least) two species from M. sect. Glabriflorae S. Lee and two from M. sect. Pubiflorae S. Lee, in both cases from different subsections. This suggests that traditionally used morphological features, such as the presence or absence of hairs on the outside of the tepals in flower, 1162 TAXON 58 (4) • November 2009: 1153–1167 and the shape and size of the fruit may not be sufficient as characters for defining infrageneric taxa within Machilus. However, a much denser sampling of the species and an additional molecular marker with higher resolution will be necessary to resolve the relationships within the genus. Persea subg. Eriodaphne, and the affinities of Persea indica. — Most American species of the Persea group so far investigated (except Persea americana, P. nudigemma, P. schiedeana, P. sphaerocarpa, P. steyermarkii and the undetermined Persea species W21874, see below) form a moderately (75% FBS) to strongly (100% PP) supported clade. This clade comprises all species (so far investigated) that had been placed in Persea subg. Eriodaphne by Kopp (1966), plus a few that had been unknown at her time. In addition, it includes Persea indica from Macaronesia. Thus the idea by one of us (Rohwer, 1993) that this species was “possibly misplaced in this genus” can be rejected—as long as P. subg. Eriodaphne is retained in Persea. In our molecular data as well as morphologically, P. subg. Eriodaphne is at least as different from P. subg. Persea as are the other genera, Machilus, Alseodaphne, and Phoebe. From this point of view it would make sense to treat it as a separate genus, which then would have to be called Mutisiopersea (Kostermans, 1993). This, however, would require the transfer of dozens of names (Kostermans made only a few new combinations). Because the Lauraceae are always good for taxonomic surprises, we feel that a far larger number of species of Persea subg. Eriodaphne as well as several of the aberrant Persea species need to be investigated before we make such far-reaching formal changes. In biogeographic terms, the most plausible explanation for the American–Macaronesian disjunction within Persea subg. Eriodaphne seems to be the relict hypothesis for the Macaronesian laurel forests. The Lauraceae have a well-documented fossil record in the Northern Hemisphere, from the early mid-Cretaceous onwards, and they persisted in Europe with considerable diversity at least until the Miocene, in southern Europe even until the Pliocene (Mai, 1971; Taylor, 1988; Drinnan & al., 1990; Crane & al., 1994; Herendeen & al., 1994; Eklund & Kvaček, 1998; Takahashi & al., 1999; Eklund, 2000 and references cited therein). Therefore, Persea indica as well as the other three Macaronesian species of Lauraceae (Apollonias barbujana, Laurus azorica, Ocotea foetens) may well be interpreted as relicts from the European-Mediterranean Tertiary laurel forests. Alternative explanations appear less likely. With a maximum age of 20 million years (Schmincke, 1976), the Macaronesian Islands are far too young for a Cretaceous disjunction. Long-distance dispersal from America to Macaronesia is still a possibility, but not very likely either. Lauraceae fruits are generally dispersed by birds (Snow, 1981; Moore & Willson, 1982; Wheelwright & TAXON 58 (4) • November 2009: 1153–1167 al., 1984), but most of the fruits, especially in the Persea group, are relatively large and heavy. Birds generally do not fly long distances with such fruits in their beak or their intestines. It seems very unlikely that they crossed the Atlantic eastwards, against the prevailing trade winds. Long-distance dispersal in the opposite direction, on the other hand, currently appears unlikely because Persea indica appears nested among the American species in our analysis. However, the nodes below Persea indica within the Eriodaphne clade neither received bootstrap support nor a significantly high posterior probability, so that a slight possibility remains that Persea indica may be the sister group of Persea subg. Eriodaphne. In the consensus from the parsimony analysis, it is sister to P. areolatocostae, the only other species in the Eriodaphne clade that has equal tepals like P. indica, and the two together are the sister group to the North and Central American species in the Eriodaphne clade. In our opinion, this uncertainty in the topology of the phylogenetic tree as well as the relatively large distance to a possible (reliable?) calibration point (Sassafras J. Presl; Poole & al., 2000) would make any attempt to estimate the age of Persea indica rather speculative. In any case, Gondwanan vicariance can be discounted as a possible reason for the trans-Atlantic disjunction in the Eriodaphne clade. When Chanderbali & al. (2001) tried to calibrate their molecular clock with such a disjunction in Ocotea, they arrived at ages of more than 600 million years for the basal lineages in the Lauraceae, far older than life on land. Since Persea, like Ocotea, is a member of the closely knit core Lauraceae (Rohwer & Rudolph, 2005), the result would undoubtedly be similar. Apollonias. — After exclusion of the sequence from EMBL (AF272257) because of probable contamination (see above), the genus is represented in our analysis by only a single sequence. This sequence, however, was retrieved twice from different specimens of Apollonias barbujana, one from Tenerife, the other from La Palma. In the present analysis it appears as sister to the Eriodaphne clade, albeit with only minimal support (63% PP). The fact that it is not a member of any well-supported clade within the Persea group may be interpreted as evidence of a long isolation of this phylogenetic lineage. Unfortunately, our attempts to isolate DNA from herbarium specimens of the second species currently ascribed to this genus, Apollonias arnottii Nees from India, were not successful. However, we expect that this species will group with one of the Asian clades (Machilus or Phoebe) rather than with A. barbujana. The Alseodaphne-Phoebe clade. — This is one of the clades that are strongly supported in the Bayesian analysis (100% PP), but do not reach 50% fast bootstrap support in the maximum parsimony analysis. Its composition of both Asian and American species is somewhat surprising, especially as they appear mixed in the present Rohwer & al. • Is Persea monophyletic? analysis. It should be noted, however, that only geographically homogeneous clades reach fast bootstrap values > 50% or posterior probabilities > 80%. Thus it would still be possible that the American and the Asian species within the Alseodaphne-Phoebe clade form separate subclades. In addition, exclusion of any of the apparently misplaced taxa usually leads to changes in the topology of the Alseodaphne-Phoebe clade beyond the immediately neighboring taxa. Those subclades that are stable and well supported in the present analysis are discussed below. The Alseodaphne andersonii /A. petiolaris clade. — These two species invariably form a clade, but its position relative to the other subclades of the AlseodaphnePhoebe clade is not stable. Exclusion of Persea nudigemma or of Alseodaphne species W14264 lets it jump to a sister group position to one of the two well-supported Persea clades in this group. In morphological terms, Alseodaphne andersonii and A. petiolaris differ from the other species of Alseodaphne included in this study by smaller terminal buds, larger leaves and large ovoid fruits, with remains of tepals persisting for some time. Thus they are quite similar to some Phoebe species, e.g., to Phoebe yunnanensis. The other species have larger terminal buds, smaller leaves, and small globose fruits seated on a naked pedicel, without remains of tepals. Persea subg. Persea. — As expected, the species attributed to Persea subg. Persea by Kopp (1966), viz., Persea americana, P. steyermarkii and P. schiedeana, form a strongly supported group. Persea schiedeana had even been treated as a variety of P. americana by Meissner (1864, as Persea gratissima schiedeana). What we also expected to find was a sister-group relationship of Persea subg. Persea with either Persea subg. Eriodaphne, which shares its American distribution, or with Machilus. The latter had been treated as a subgenus of Persea by several authors, and morphologically it appears rather close to Persea subg. Persea, perhaps even closer than Persea subg. Eriodaphne. This expectation is not confirmed in the present study. Persea subg. Persea and a few additional Persea species that were unknown to Kopp (1966), P. nudigemma, P. sphaerocarpa and the undetermined species Persea W21874, are well supported as members of the otherwise Asian Alseodaphne-Phoebe clade. These additional species show characters that are uncommon in Persea. As described by van der Werff (1994), Persea nudigemma appears to belong to a small group of neotropical Persea species with equal tepals that are pubescent on both sides and persistent in the fruiting stage (e.g., P. bernardii Kopp, P. rigens Allen, P. silvatica van der Werff). In so far these species are similar not only to P. indica (see above), but also to Phoebe. Unfortunately, we did not yet manage to get good sequences of any of these species, so that we could only speculate about their position in the phylogeny and their influence on the 1163 Rohwer & al. • Is Persea monophyletic? analysis. The position of Persea sphaerocarpa within the Alseodaphne-Phoebe clade, or even more among the Alseodaphne species as in the parsimony analysis, is not surprising. Van der Werff (2002) already noted that this species, together with P. albiramea van der Werff and P. laevifolia van der Werff, “would very likely be included in Alseodaphne had they been collected in tropical Asia.” Although the undetermined Persea species W21874 is not particularly similar to P. sphaerocarpa, its position is not really surprising, either. With its stubby twigs with large inflorescence scars and conspicuous bract scars, it is reminiscent of P. americana, whereas glabrous inflorescences are quite common in Alseodaphne (see Kostermans, 1973a). As indicated above, the position of Alseodaphne species W14264 among the Persea species is neither well supported nor stable. In parsimony analyses it usually groups with A. andersonii and A. petiolaris (see above). The Alseodaphne-Dehaasia clade. — Except for the three Alseodaphne species discussed above (A. andersonii, A. petiolaris, and species W14264), all remaining species of Alseodaphne and all species of Dehaasia examined form a moderately to strongly supported clade (73% FBS, 100% PP). The present arrangement would suggest that Dehaasia originated at least twice within Alseodaphne, but the taxon sample is still too small for such a conclusion. The sterile specimen received as Nothaphoebe umbelliflora from Singapore has to be a misidentification; it differs by just one and two base pairs, respectively, from the two specimens of Alseodaphne nigrescens from the same source. The two Phoebe clades. — Species of the genus Phoebe are found in two relatively well-supported clades (100% PP; 81% and 88% FBS, respectively). Inspection of the individual trees retrieved in our analysis reveals that both clades form a monophyletic group in some of them, whereas in others one of these clades (the one with Ph. formosana) is sister to the Alseodaphne-Dehaasia clade, or the three clades are unresolved as in the consensus. This leaves a good chance that ultimately Phoebe will turn out as monophyletic. A recent paper by Soltis & al. (2008) has drawn our attention to another possibility: the two clades could be evidence of incomplete homogenization of different ITS copies. The two clades differ not only by simple base changes, but also by minor length mutations. Unlike the situation in other genera, we have a few samples of Phoebe species from which we were unable to obtain clean sequences, even from silica gel–dried material. In at least one of these samples the sequence becomes “messy” at the position of the first of these length mutations, and from the chromatogram it seems that both patterns could be present. Resolving this question, however, would require sequencing of numerous clones per species, which was beyond our financial resources. 1164 TAXON 58 (4) • November 2009: 1153–1167 Nothaphoebe cavaleriei, which is nested in one of the Phoebe clades here, has a quite complicated taxonomic history. The species originally had been described as Lindera cavaleriei by H. Léveillé (in Repert. Spec. Nov. Regni Veg. 10: 371. 1912), and it was transferred to Nothaphoebe by Yang (in J. West China Border Res. Soc., Ser B 15: 75. 1945). Later Kostermans (1957) included Alseodaphne, Machilus and Nothaphoebe in Persea, and in 1962 he made the formal new combination for this species, Persea cavaleriei (H. Lév.) Kosterm., based on Lindera cavaleriei H. Lév. Again some years later (1973a) he changed his mind and reinstated Alseodaphne and Nothaphoebe. In this paper, he made another new combination Alseodaphne cavaleriei (H. Lév.) Kosterm., based on Machilus cavaleriei H. Lév. (in Bull. Géogr. Bot. 24: 142. 1914). He incorrectly cited Nothaphoebe cavaleriei and Persea cavaleriei in synonymy, and in the discussion he wrote: “Some mistakes crept in, in my Bibliography [i.e., Bibliographia Lauracearum, 1964], where erroneously Lindera cavaleriei is cited as the basionym. Lindera cavaleriei, based on the specimen Cavalerie 1222 represents Phoebe nanmu, the type is conserved in Edinburgh and consists of a sterile branchlet, showing the collar of budscale scars, found in Phoebe nanmu.” Apparently he now wanted to have his Persea cavaleriei to be based on Machilus cavaleriei rather than on Lindera cavaleriei. In our opinion, such a change cannot be made a posteriori. As to the generic identity of Lindera cavaleriei, one of us (H. v.d.W.) came to a similar conclusion as Kostermans (1973a). After checking the specimens (including the type) of Nothaphoebe cavaleriei (Lév.) Yang, as the species is most frequently called, he found that it did not seem to differ from Phoebe. This is confirmed by the present study. It should be noted, however, that this species—correctly or erroneously—at some time had been linked to each of the five genera in the Persea group which have tetrasporangiate anthers. This again illustrates the weakness of the generic concepts in the Persea group. Sassafras. — Although the aim of this study was to resolve relationships within the Persea group, of which the genus Sassafras is not a member, we would like to draw attention to the fact that in our analysis Sassafras belongs to the Cinnamomeae, as sister group to the remaining genera of that clade, as in the unconstrained ITS analysis of Chanderbali & al. (2001) and in the matK analysis of Rohwer & Rudolph (2005). Sassafras had been placed in the Laureae by Rohwer (1993), although he noticed correctly that this genus had neither umbellate nor involucrate inflorescences. Reasons to keep Sassafras in the Laureae were its deciduous foliage, otherwise known only from Lindera Thunb. and Litsea Lam., and its introrse anthers of the third staminal whorl. However, deciduous foliage is of course an adaptation to a seasonal habitat, and more or less introrse anthers are often found in unisexual flowers, TAXON 58 (4) • November 2009: 1153–1167 also in Ocotea and Endlicheria within the Cinnamomeae. Therefore, it is not surprising to find Sassafras as sister to the rest of the Cinnamomeae also in this study. CONCLUSION Our analysis shows that Machilus should be treated as a separate genus, not as a subgenus of Persea. Similarly, it shows that Persea subg. Eriodaphne ultimately will have to be treated as a separate genus Mutisiopersea, which probably will also include the Macaronesian Persea indica. Phoebe is distinct from both Machilus and Persea (unless this latter name would be used for the entire Alseodaphne-Phoebe clade). The inclusion of Dehaasia in Alseodaphne is undoubtedly justified; as in many other cases, the number of pollen sacs alone is not sufficient as a generic character. A careful taxonomic revision will be necessary to decide whether or not it is diagnostic on the species level. Alseodaphne, Dehaasia, Nothaphoebe and those species currently placed in Persea that do not conform to the “typical” Eriodaphne pattern (unequal tepals, the outer ones glabrous on the adaxial surface, persistent in fruit) need a much denser sampling. In Phoebe, cloning will be necessary to check for different ITS copies. The most important obstacle on our way towards an understanding of the Persea group now appears to be the poor state of basic taxonomic and floristic knowledge. Besides the (somewhat outdated) monograph of American Persea by Kopp (1966), there is no monograph with keys to the species for any of the genera of the Persea group, and for most countries of subtropical to tropical America and Asia there are no floristic treatments either. This makes it extremely difficult to determine species of this group or to verify the determinations of cultivated material or herbarium specimens. Therefore, a combined effort of classical morphological and molecular work will be necessary to resolve the problems within this group. ACKNOWLEDGMENTS We thank the curators of various herbaria given in the Appendix for loans and allowing DNA extraction from certain specimens. Andrea Jounais, Anna Maria Vogt, Lang Li and Junqiu Chen are gratefully acknowledged for their skilled technical assistance in the lab. Kristina Ruthe (née Groth) allowed us to use unpublished Cinnamomum sequences from her work, and Matthias Wolf (Univ. Würzburg) helped with the delimitation of the ITS-1 and ITS-2 regions. Financial support was provided by DFG (446 CHV 112/1/04) and NSFC (30310103218). Jie Li and Hsi-wen Li were partly funded by the Chinese Academy of Sciences (KSCX2-YW-Z-001) and National Natural Science Foundation of China (30870170). Rohwer & al. • Is Persea monophyletic? LITERATURE CITED Bentham, G. 1880. Laurineae. Pp. 146–168 in: Bentham, G. & Hooker, J.D. (eds.), Genera Plantarum, vol. 3. Lovell Reeve & Co., Williams & Norgate, London. Beyra-Matos, A. & Lavin, M. 1999. A monograph of Pictetia (Leguminosae-Papilionoideae) and review of the Aeschynomeneae. Syst. Bot. Monogr. 56: 1–93. Buckler, E.S., 4th & Holtsford, T.P. 1996. Zea systematics: ribosomal ITS evidence. Molec. Biol. Evol. 13: 612–622. Buckler, E.S., Ippolito, A. & Holtsford, T.P. 1997. The evolution of ribosomal DNA: divergent paralogues and phylogenetic implications. Genetics 145: 821–832. Chanderbali, A.S. 2004. Endlicheria (Lauraceae). Flora Neotropica Monographs 91. New York Botanical Garden Press, Bronx. Chanderbali, A.S., van der Werff, H. & Renner, S.S. 2001. Phylogeny and historical biogeography of Lauraceae: evidence from the chloroplast and nuclear genomes. Ann. Missouri Bot. Gard. 88: 104–134. Crane, P.R., Friis, E.M. & Pedersen, K.R. 1994. Palaeobotanical evidence on the early radiation of magnoliid angiosperms. Pl. Syst. Evol., Suppl. 8: 51–72. Doyle, J.J. & Doyle, J.L. 1987. A rapid DNA isolation procedure for small quantities of fresh leaf tissue. Phytochem. Bull. 19: 11–15. Drinnan, A.N., Crane, P.R., Friis, E.M. & Pedersen, K.R. 1990. Lauraceous flowers from the Potomac Group (mid-Cretaceous) of eastern North America. Bot. Gaz. 151: 370–384. Eklund, H. 2000. Lauraceous flowers from the Late Cretaceous of North Carolina, U.S.A. Bot. J. Linn. Soc. 132: 397–428. Eklund, H. & Kvaček, J. 1998. Lauraceous inflorescences and flowers from the Cenomanian of Bohemia (Czech Republic, Central Europe). Int. J. Pl. Sci. 159: 668–686. Groth, K. 2003. Molecular Systematic Studies on the Genus Cinnamomum Schaeffer (Lauraceae). Diploma thesis, University of Hamburg, Hamburg. Hall, T.A. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/ NT. Nucleic Acids Symp. Ser. 41: 95–98. Herendeen, P.S., Crepet, W.L. & Nixon, K.C. 1994. Fossil flowers and pollen of Lauraceae from the Upper Cretaceous of New Jersey. Pl. Syst. Evol. 189: 29–40. Hilu, K.W. & Liang, H. 1997. The matK gene: sequence variation and application in plant systematics. Amer. J. Bot. 84: 830–839. Huelsenbeck, J.P. & Ronquist, F. 2001. MRBAYES: Bayesian inference of phylogenetic trees. Bioinformatics 17: 754–755. Johnson, L.A. & Soltis, D.E. 1994. MatK DNA sequences and phylogenetic reconstruction in Saxifragaceae s. str. Syst. Bot. 19: 143–156. Johnson, L.A. & Soltis, D.E. 1995. Phylogenetic inference in Saxifragaceae sensu stricto and Gilia (Polemoniaceae) using matK sequences. Ann. Missouri Bot. Gard. 82: 149–175. Käss, E. & Wink, M. 1997. Phylogenetic relationships in the Papilionoideae (family Leguminosae) based on nucleotide sequences of cpDNA (rbcL) and ncDNA (ITS 1 and 2). Molec. Phylog. Evol. 8: 65–88. Kopp, L.E. 1966. A taxonomic revision of the genus Persea in the Western Hemisphere (Perseae-Lauraceae). Mem. New York Bot. Gard. 14: 1–120. 1165 Rohwer & al. • Is Persea monophyletic? Kostermans, A.J.G.H. 1957. Lauraceae. Pengum. Balai Besar Penjel. Kehut. Indonesia 57: 1–64. Kostermans, A.J.G.H. 1961. The new world species of Cinnamomum Trew (Lauraceae). Reinwardtia 6: 17–24. Kostermans, A.J.G.H. 1962. The Asiatic species of Persea Mill. (Lauraceae). Reinwardtia 6: 189–194. Kostermans, A.J.G.H. 1964. Bibliographia Lauracearum. Archipel, Bogor. Kostermans, A.J.G.H. 1973a. A synopsis of Alseodaphne Nees (Lauraceae). Candollea 28: 93–136. Kostermans, A.J.G.H. 1973b. A synopsis of the genus Dehaasia Bl. (Lauraceae). Bot. Jahrb. Syst. 93: 424–480. Kostermans, A.J.G.H. 1974. A monograph of Caryodaphnopsis A. Shaw. Reinwardtia 9: 123–137. Kostermans, A.J.G.H. 1993. Mutisiopersea Kostermans, a new genus in Lauraceae. Rheedea 3: 132–135. Li, H.W., Pai, P.Y., Lee, S.K., Wei, F.N., Wei, Y.T., Yang, Y.C., Huang, P.H., Tsui, H.P., Shia, Z.D. & Li, J.L. 1984. Lauraceae. Pp. 1–463 in: Li, H.W. (ed.), Flora Reipublicae Popularis Sinicae, vol. 31. Science Press, Beijing. Li, J. & Christophel, D.C. 2000. Systematic relationships within the Litsea complex (Lauraceae): a cladistic analysis on the basis of morphological and leaf cuticle data. Austral. Syst. Bot. 13: 1–13. Li, J., Christophel, D.C., Conran, J.G. & Li, H.W. 2004. Phylogenetic relationships within the ‘core’ Laureae (Litsea complex, Lauraceae) inferred from sequences of the chloroplast gene matK and nuclear ribosomal DNA ITS regions. Pl. Syst. Evol. 246: 19–34. Mai, D.H. 1971. Über fossile Lauraceae und Theaceae in Mitteleuropa. Feddes Repert. 82: 313–341. Meissner, C.F. 1864. Lauraceae. Pp. 1–260 in: Candolle, A.P. de (ed.), Prodromus Systematis Universalis Regni Vegetabilis, vol. 15 (1). Fortin, Masson et soc., Paris. Mez, C. 1889. Lauraceae Americanae. Gebrüder Borntraeger, Berlin. Moore, L.A. & Willson, M.F. 1982. The effect of microhabitat, spatial distribution, and display size on dispersal of Lindera benzoin by avian frugivores. Canad. J. Bot. 60: 557–660. Mort, M.E., Soltis, P.S., Soltis, D.E. & Mabry, M.L. 2000. Comparison of three methods for estimating internal support on phylogenetic trees. Syst. Biol. 49: 160–171. Pax, F. 1889. Lauraceae. Pp. 106–126 in: Engler, A. & Prantl, K. (eds.), Die natürlichen Pflanzenfamilien, vol. III, 2. Engelmann, Leipzig. Poole, I., Richter, H.G. & Francis, J.E. 2000. Evidence for Gondwanan origins for Sassafras (Lauraceae)? Late Cretaceous fossil wood of Antarctica. I.A.W.A. Journal 21: 463–475. Rohwer, J.G. 1986. Prodromus einer Monographie der Gattung Ocotea Aubl. (Lauraceae), sensu lato. Mitt. Inst. Allg. Bot. Hamburg 20: 3–278. Rohwer, J.G. 1991. Borderline cases between Ocotea, Nectandra, and Phoebe (Lauraceae): the “marginal” species of the Ocotea helicterifolia-group, including the O. heydeanagroup. Bot. Jahrb. Syst. 112: 365–397. Rohwer, J.G. 1993. Lauraceae. Pp. 366–391 in: Kubitzki K., Rohwer, J.G. & Bittrich, V. (eds.), The Families and Genera of Vascular Plants, vol. 2. Springer, Berlin. Rohwer, J.G. 2000. Toward a phylogenetic classification of 1166 TAXON 58 (4) • November 2009: 1153–1167 the Lauraceae: evidence from matK sequences. Syst. Bot. 25: 60–71. Rohwer, J.G., Richter, H.G. & van der Werff, H. 1991. Two new genera of neotropical Lauraceae and critical remarks on the generic delimitation. Ann. Missouri Bot. Gard. 78: 388–400. Rohwer, J.G. & Rudolph, B. 2005. Jumping genera: the phylogenetic positions of Cassytha, Hypodaphnis, and Neocinnamomum (Lauraceae) based on different analyses of trnK intron sequences. Ann. Missouri Bot. Gard. 92: 153–178. Ronquist, F. & Huelsenbeck, J.P. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572–1574. Schmincke, H.-U. 1976. The geology of the Canary Islands. Pp. 67–184 in: Kunkel, G. (ed.), Biogeography and Ecology in the Canary Islands. Monographiae Biologicae 30. W. Junk, Den Haag. Snow, D.W. 1981. Tropical frugivorous birds and their food plants: a world survey. Biotropica 13: 1–14. Soltis, D.E., Mavrodiev, E.V., Doyle, J.J., Rauscher, J. & Soltis, P.S. 2008. ITS and ETS sequence data and phylogeny reconstruction in allopolyploids and hybrids. Syst. Bot. 33: 7–20. Steele, K.P. & Vilgalys, R. 1994. Phylogenetic analyses of Polemoniaceae using nucleotide sequences of the plastid gene matK. Syst. Bot. 19: 126–142. Sworfford, D.L. 2001. PAUP*. Phylogenetic Analysis Using Parsimony (*and Other Methods), version 4.0b10. Sinauer, Sunderland. Takahashi, M., Crane, P. R. & Ando, H. 1999. Fossil flowers and associated plant fossils from the Kamikitaba locality (Ashizawa formation, Futaba group, Lower Coniacian, Upper Cretaceous) of Northeast Japan. J. Pl. Res. 112: 187–206. Taylor, D.W. 1988. Eocene floral evidence of Lauraceae: corroboration of the North American megafossil record. Amer. J. Bot. 75: 948–957. Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F. & Higgins, D.G. 1997. The Clustal X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 25: 4876–4882. Van der Werff, H. 1989. A new species of Persea (Lauraceae) from Surinam, with a discussion of its position within the genus. Ann. Missouri Bot. Gard. 76: 939–941. Van der Werff, H. 1994. Novelties in Neotropical Lauraceae. Novon 4: 58–76. Van der Werff, H. 2001. An annotated key to the genera of Lauraceae in the Flora Malesiana region. Blumea 46: 125–140. Van der Werff, H. 2002. A synopsis of Persea (Lauraceae) in Central America. Novon 12: 575–586. Van der Werff, H. & Richter, H.G. 1996. Toward an improved classification of Lauraceae. Ann. Missouri Bot. Gard. 83: 409–418. Wheelwright, N.T., Haber, W.A., Murray, K.G. & Guidon, C. 1984. Tropical fruit-eating birds and their food plants: a survey of a Costa Rican lower montane forest. Biotropica 16: 173–192. White, T.J., Bruns, T., Lee, S. & Taylor, J.W. 1990. Amplification and direct sequencing of fungal ribosomal RNA genes for phylogenetics. Pp. 315–322 in: Innis, M.A., Gelfand, D.H., Sninsky, J.J. & White, T.J. (eds.), PCR Protocols: A Guide to Methods and Applications. Academic Press, San Diego. TAXON 58 (4) • November 2009: 1153–1167 Rohwer & al. • Is Persea monophyletic? Appendix. Species examined. Taxon, origin, voucher information and GenBank/EMBL accession numbers. Accession numbers beginning with AB are from Fijridiyanto & Murakami (unpub.), with AF2 from Chanderbali & al. (2001), with AF3 from Chanderbali (2004), with AY from Li & al. (2004) or Li & al. (unpub.); numbers beginning with FM are new sequences. Actinodaphne maingayi Hook. f., Malaysia, Sarawak, Lambir National Park, Fijridiyanto IZ 2068 (KYO), AB260851; A. sesquipedalis Meisn., Malaysia, Kuala Lumpur, Saw Leng Guan s.n. (KEP), AF272247; Alseodaphne andersonii (King ex Hook. f.) Kosterm., China, Yunnan, JL & LL [Jie Lie & Lang Li] 20070074 (HITBC), FM957793; A. nigrescens (Gamble) Kosterm., #1: Singapore Bot. Gard., SING 2000-27, FM957794; #2: Singapore Bot. Gard., SING 2000-29, FM957795; A. petiolaris (Meissn.) Hook. f., China, Yunnan, Xishuangbanna Tropical Bot. Gard. (XTBG), CJQ [Jun-qiu Chen] 07003 (HITBC), FM957796; A. semecarpifolia Nees, Sri Lanka, Central Prov., Kandy, Malcomber 2753 (MO), AF272252; A. sp. A38, Indonesia, Java, Arifiani & van der Werff 38 (MO), FM957799; A. sp. W14264, Vietnam, Tuyen Quang, van der Werff & Nguyen 14264 (MO), FM957797; A. sp. W17084, Vietnam, Lang Son, van der Werff & al. 17084 (MO), FM957798; Anaueria brasiliensis Kosterm., Peru, Loreto, Maynas, Vásquez 25228 (MO), FM957800; Apollonias barbujana (Cav.) Bornm., #1: Spain, Canary Islands, Tenerife, Bramwell 628 (MO), AF272257 [probably errroneous, excluded from final analysis]; #2: Spain, Canary Islands, Tenerife, Rohwer s.n. (HBG), AY934889; Chlorocardium rodiei (R.H. Schomb.) Rohwer, H.G. Richt. & van der Werff, Guyana, Demerara, Mabura Hill, Chanderbali 246 (MO), AF272258; C. venenosum (Kosterm. & Pinkley) Rohwer, H.G. Richt. & van der Werff, Peru, Loreto, Iquitos, Vásquez 25236 (MO), AF272259; Cinnamomum amoenum (Nees & Mart.) Kostermans, Brazil, Paraná, Curitiba, Ribas & Poliquesi 1072 (HBG), FM957801; C. burmanii (Nees & T. Nees) Blume, Germany, Berlin Bot. Gard., Leuenberger s.n. (HBG), FM957802; C. daphnoides Siebold & Zucc., Germany, Berlin Bot. Gard., Leuenberger s.n. (HBG), FM957803; C. duartianum Vattimo-Gil, Brazil, Kubitzki & Figuereido 83-35 (HBG), FM957804; C. quadrangulum Kosterm., Brazil, Minas Gerais, Lorea-Hernandez 5585 (MO), AF272265; Dehaasia cuneata (Blume) Blume, Indonesia, North Sumatra, Gunung Leuser Nature Reserve, W.J.J.O. de Wilde & B.E.E. de Wilde-Duyfjes 12742 (HBG), AY934890; D. incrassata (Jack) Kosterm., Philippines, Palawan, Taytay, Soejarto 7693 (MO), AF272268 [probably errroneous, excluded from final analysis]; D. sp. A34, Indonesia, Java, Arifiani & van der Werff 34 (MO), FM957805; D. sp. A37, Indonesia, Java, Arifiani & van der Werff 37 (MO), FM957806; Endlicheria gracilis Kosterm., Guyana, Iwokrama Reserve, Chanderbali 250 (MO), AF363374; Laurus azorica (Seub.) Franco, Spain, Canary Islands, Tenerife, Anaga Mts., Rohwer s.n. (HBG), FM957807; Laurus nobilis L., #1: U.S.A., St. Louis, Missouri Botanical Garden, Chanderbali 327 (MO), AF272278; #2: Germany, Hamburg Bot. Gard., Rohwer s.n. (HBG) FM957808; Licaria canella (Meisn.) Kosterm., Guyana, Demerara, Mabura Hill, Chanderbali 234 (MO), AF272280. Lindera benzoin (L.) Blume, #1: U.S.A., St. Louis, Missouri Bot. Gard., Chanderbali 324 (MO), AF272283; #2: Germany, Hamburg Bot. Gard., Rohwer s.n. (HBG), FM957809; Litsea accedens (Blume) Boerl., Malaysia, Sarawak, Lambir National Park, Fijridiyanto IZ 2066 (KYO), AB260860; L. glutinosa (Lour.) C.B. Rob., China, Yunnan, Mengla, Li H.-W. 21 (HITBC), AY265403; Machilus chienkweiensis S. Lee, China, Guangxi, Shangsi, Li J. 2002153 (HITBC), AY934895; M. decursinervis Chun, China, Guangxi Prov. Jinxiu, Li J. 2002195 (HITBC), AY934893; M. grijsii Hance, Germany, Hamburg Bot. Gard., Rohwer 193 (HBG), FM957810; M. japonica Sieb. & Zucc., Korea, Cheju, Hanra Bot. Gard., Kim Chang Suk s.n., 14 Oct 1998 (HBG), AY934891; M. minkweiensis S. Lee, China, Guangxi, Jinxiu, Li J. 2002194 (HITBC), AY934897; M. obovatifolia (Hay.) Kanehira et Sasaki, China, Taiwan, Pingtung, Liu S.M. 198 (KUN), FM957811; M. oreophila Hance, China, Guangxi, Nanning Bot. Gard., Li J. 2002176 (HITBC), AY934898; Machilus rimosa (Blume) Blume, Indonesia: Cibodas Botanical Garden, Fijridiyanto VII.C. 58D (KYO), AB260888; M. shweliensis W.W. Sm., China, Yunnan, Jingdong, J. Li 2002065 (HITBC), AY934899; M. thunbergii Sieb. & Zucc. [= Persea thunbergii (Sieb. & Zucc.) Kosterm.], #1: U.S.A., St. Louis, Missouri Bot. Gard., Chanderbali 328 (MO), AF272327; #2: Japan, Honshu, Yasuda 1352 (MO), FM957814; M. yunnanensis Lec., China, Yunnan, Kunming Bot. Gard., Li J. 2004001 (HITBC), AY934892; M. zuihoensis Hayata, Germany, Hamburg Bot. Gard., Rohwer s.n. (HBG), FM957815; M. sp. W14068, Vietnam, Vinh Phu, van der Werff & al. 14068 (MO), FM957812; M. sp. W14071, Vietnam, Vinh Phu, van der Werff & al. 14071 (MO), FM957813; Mezilaurus triunca van der Werff, Peru, Amazonas, Iquitos, Vásquez 25227 (MO), AF272287; Nectandra amazonum Nees, #1: Guyana, Essequibo, Iwokrama Reserve, Chanderbali 217 (MO), AF272289; #2: Germany, Hamburg Bot. Gard., Rohwer s.n. (HBG), FM957816; Neolitsea levinei Merr., China, Yunnan, Mengla, Li H.-W. 29 (HITBC), AY265401; N. sericea (Blume) Koidz., #1: Japan, Kyoto, Yasuda 1355 (MO), AF272296; #2: Germany, Hamburg Bot. Gard., Rohwer s.n. (HBG), FM957817; Nothaphoebe cavaleriei (Levl.) Yang, #1: China, Yunnan, Shuijiang, Sun B.X. 0190 (KUN), AY934900; #2: China, Sichuan, LL [Lang Li] 20070260 (HITBC), FM957818; N. umbelliflora (Blume) Blume, Singapore Bot. Gard., SING 2000-30, FM957819;Ocotea foetens (Aiton) Baill., Portugal, Madeira, Porto Moniz, Maas 8642 (MO), AF272300; O. helicterifolia (Meisn.) Hemsl., Mexico, Oaxaca, Miabuatlan, Torres 11911 (MO), AF272303; O. pulchella Mart., Brazil, Minas Gerais. Fazenda Caiera, LoreaHernandez 5575 (MO), AF272312 [not AF262312, as erroneously cited in Chanderbali & al. 2001]; Parasassafras confertiflora (Meisn.) D.G.Long, #1: China, Yunnan, Lishui Co., Li Heng 10030 (MO), AF 272321; #2: China, Yunnan, Qian Y.Y. 682 (KUN), AY265395; Persea alba Nees, Brazil, Paraná, Guartelá, Hatschbach 59099 (HBG), FM957820; Persea americana Mill., #1: U.S.A., St. Louis, Missouri Bot. Gard., Chanderbali 323 (MO), AF272322; #2: Germany, Hamburg Bot. Gard., Rohwer s.n. (HBG), FM957821; P. areolatocostae (C.K. Allen) van der Werff, #1: Peru, Amazonas, Prov. Bagua, Yamayakat, Jaramillo 1227 (HBG), FM957822; #2: Peru, San Martín, Rioja, van der Werff 16459 (HBG), FM957823; P. borbonia Spreng., Germany, Bochum Bot. Gard., B. Kirchner s.n. (HBG), AY934901; P. caerulea (Ruiz & Pav.) Mez., #1: Peru, Amazonas, van der Werff 14744 (MO), AF272323; #2: Peru, Pasco, Oxapampa, van der Werff 17983 (MO), FM957824; P. indica (L.) Spreng., Germany, Hamburg Bot. Gard., Rohwer s.n. (HBG), AY934902; P. lingue (Ruiz & Pav.) Nees ex Kopp, #1: Chile, Greissl 640-99 (MJG), AF272324; #2: Chile, Prov. Quillota, Quebrada Alvarado, Zöllner 20858 (HBG), FM957825; P. meridensis Kopp, Venezuela, Trujillo, Bocono, Cuello 943 (MO), AF272325; P. mutisii H.B.K., Colombia, Antioquia, Medellín, Zarucchi 5306 (HBG), FM957826; P. nudigemma van der Werff, Ecuador, Zamora-Chinchipe, van der Werff & al. 19209 (MO), FM957828; P. pajonalis van der Werff, #1: Peru, Pasco, Oxapampa, Vasquez 27793 (HBG), FM957829; #2: Peru, Pasco, Oxapampa, Monteagudo & al. 4581 (HBG), not submitted to EMBL because identical with #1; P. palustris (Raf.) Sarg., #1: U.S.A., Florida, Okaloosa, J.S. Miller & al. 8380 (MO), FM957830; #2: U.S.A., Florida, Leon, J.S. Miller & al. 9018 (MO), FM957831; P. schiedeana Nees, Costa Rica, Puntarenas, Monteverde, Bello 3806 (HBG), FM957832; P. sphaerocarpa (H. Winkl.) Kosterm., Peru, Pasco, Oxapampa, van der Werff & al. 17889 (MO), FM957837; P. steyermarkii C.K. Allen, El Salvador, Dept. Santa Ana, Bosque Nebuloso Montecristo, Los Planes, M.L. Reyna 1398 (HBG) FM957838; P. subcordata (Ruiz & Pav.) Nees, Peru, Cusco, La Convención, Vilcabamba, Ututo, Succli 1456 (HBG), FM957839; P. aff. subcordata (Ruiz & Pav.) Nees, Bolivia, Dept. La Paz, Prov. Nor Yungas, Coroico, Beck 21834 (HBG), FM957827; P. venosa Nees, Brazil, Paraná, Bocaiuva do Sul, Rio Capivari, Silva 1400 (HBG), FM957840; P. veraguasensis Seem., Costa Rica, Santa Maria de Dota, Burger 12085 (HBG), FM957841; P. weberbaueri Mez, Ecuador, van der Werff 21626 (MO), FM957842; P. willdenovii Kosterm., Brazil, Paraná, Quatro Barras, Roderjan 1117 (HBG), FM957843; P. sp. W14857, Peru, Amazonas, van der Werff & al. 14857 (MO), FM957833; P. sp. W19517, Ecuador, Zamora-Chinchipe, van der Werff & al. 19517 (MO), FM957834; P. sp. W21874, Ecuador, van der Werff 21874 (MO), FM957835; P. sp. V25232, Peru, Loreto, Maynas, Vásquez & Ortiz-Gentry 25232 (MO), FM957836; Phoebe excelsa (Blume) Nees, Indonesia, Cibodas Botanical Garden, Fijridiyanto VII.C. 60 (KYO), AB260889; Ph. formosana (Hayata) Hayata, Germany, Bonn Bot. Gard., Rohwer 156 (MJG), AF272328; Ph. forrestii W.W. Smith, China, Yunnan, Jingdong, Li J. 2002083 (HITBC), AY934904; Ph. lanceolata (Wall. ex Nees) Nees, China, Yunnan, XTBG, Li J. 2004003 (HITBC), FM957844; Ph. macrocarpa C.Y. Wu, China, Guangxi, Guangxi Medical Botanical Garden, Li 2002207 (HITBC), FM957845; Ph. megacalyx H.W. Li, Vietnam, van der Werff 17399 (MO), FM957846; Ph. neurantha (Hemsl.) Gamble, China, Yunnan, XTBG, CJQ & ZJS [Jun-qiu Chen & Jin-shun Zhong] 2005024 (HITBC), FM957847; Ph. puwenensis Chang, China, Yunnan, Jingdong, Li J. 2002081 (HITBC), AY934905; Ph. sheareri (Hemsl.) Gamble, #1: Germany, Hamburg Bot. Gard. 504-86 plant #1, originally from China, Hangchow, Rohwer s.n. (HBG), FM957848; #2: Germany, Hamburg Bot. Gard. 50486 plant #2, Rohwer s.n. (HBG), FM957849; Rhodostemonodaphne crenaticupula Madriñán, Guyana, Essequibo, Iwokrama Reserve, Chanderbali 265 (MO), AF272331; Sassafras albidum (Nutt.) Nees, #1: U.S.A., St. Louis, Missouri Bot. Gard., Chanderbali 325 (MO), AF272335; #2: Germany, Hamburg Bot. Gard., Rohwer s.n. (HBG), FM957850; S. tzumu (Hemsl.) Hemsl., #1: China, Hunan, Xining Co., Luo Lin-bo 1242 (MO), AF272336; #2: China, Yunnan, Li H.-W. 15 (HITBC), AY265391 [identical with preceding in all overlapping positions, therefore consensus sequence used in analysis]; Sextonia pubescens van der Werff, Peru, Loreto, Iquitos, Vásquez 25229 (MO), AF268808; Umbellularia californica (Hook. & Arn.) Nutt., #1: U.S.A., St. Louis, Missouri Bot. Gard., Chanderbali 326 (MO), AF272337; #2: U.S.A., without locality, van der Werff s.n. (MO), AY265393. 1167