Journal of Biogeography, 30, 1311–1328 The phytogeographical affinities of the Pitcairn Islands – a model for south-eastern Polynesia? Naomi Kingston1*, Steve Waldren2 and Una Bradley1 1Department of Botany, Trinity College, Dublin 2, Ireland and 2Trinity College Botanic Gardens, Palmerston Park, Dartry, Dublin 6, Ireland Abstract Aim To identify how the Pitcairn group relates biogeographically to the south-eastern Polynesian region and if, as a subset of the regions flora, it can then be used as a model for biogeographical analyses. Location The Pitcairn group (25
4¢ S, 130
06¢ W) comprises four islands: Pitcairn, a relatively young, high volcanic Island; Henderson, an uplifted atoll, the uplift caused by the eruption of Pitcairn; and two atolls, Ducie and Oeno. The remote location, young age and range of island types found in the Pitcairn Island group makes the group ideal for the study of island biogeography and evolution. Methods A detailed literature survey was carried out and several data sets were compiled. Dispersal method, propagule number and range data were collected for each of the 114 species that occurs in the Pitcairn group, and environmental data was also gathered for islands in Polynesia. Analyses were carried out using non-metric multidimensional scaling and clustering techniques. Results The flora of the Pitcairn Islands is derived from the flora of other island groups in the south-eastern Polynesian region, notably those of the Austral, Society and Cook Islands. Species with a Pacific-wide distribution dominate the overall Pitcairn group flora. However, each of the islands show different patterns; Pitcairn is dominated by species with Pacific, Polynesian and endemic distributions, with anemochory as the dominant dispersal method (39.5%); Henderson is also dominated by species with Pacific, Polynesian and endemic distributions, but zoochory is the dominant dispersal method (59.4); Oeno and Ducie are dominated by Pantropic species with hydrochory as the most common dispersal method (52.9% and 100%, respectively). Main conclusions • Habitat availability is the most significant factor determining the composition and size of the flora. • South-east Polynesia is a valid biogeographical unit, and should include the Cook, Austral, Society, Marquesas, Gambier, Tuamotu and Pitcairn Islands with Rapa, but should exclude Easter Island, Tonga and Samoa. • Regionalization schemes should take island type into consideration. • The Pitcairn Island group can serve as a useful model for Pacific biogeographical analyses. Keywords Colonization, dispersal, habitat diversity, island biogeography, insular biotas, NMDS, Pitcairn, Sorensen distance, south-eastern Polynesia, UPGMA. *Correspondence: Naomi Kingston, National Parks and Wildlife, Department of the Environment, Heritage and Local Government, 7 Ely Place, Dublin 2, Ireland. E-mail: nkingston@duchas.ie  2003 Blackwell Publishing Ltd 1312 N. Kingston et al. INTRODUCTION Since Schouw (1823) first recognized the Polynesian region, numerous biogeographical schemes, using virtually all groups of organisms, have distinguished Polynesia as a subset of the Indomalayan (Oriental) Realm (Kay, 1979). Many schemes have also recognized south-eastern Polynesia as a further subdivision of the Polynesian region, although the position of the boundaries varies (van Balgooy, 1960, 1971; Gressit, 1963; van Balgooy et al., 1996). The problems associated with dividing south-eastern Polynesia are best summarized by van Balgooy et al. (1996) who stated that as a result of the considerable overlap in provinces, the floristic provinces could not be classified into a hierarchical system. Stoddart (1992) considers that island type should be included in biogeographical schemes, as once the relatively homogeneous biotas of atolls and makatea islands are removed the problem resolves itself into the consideration of high islands. Brown (1935) was the first to assess the flora of southeastern Polynesian, dividing it into six regions: Marquesas, Society, Tuamotu, Austral, Rapa, Pitcairn and Gambier Islands. However, the information from which these divisions were derived was very limited, with only 251 species being known from the region at that time, mostly from the Society and Marquesas Islands (there are currently over 600 species known from the Society Islands alone; Florence, 1987). As many of the collections were very recent, and hence poorly understood, 87% of these species were thought to be confined to the Polynesian region, and only two species were assigned a pantropical distribution. Brown also suggested that 82% of the species were of American origin. In terms of general patterns, there is a gradual reduction in the total number of species from west to east across the Pacific; islands in the eastern half of the Pacific have virtually nothing in common with the flora of the Americas; disjunct, discontinuous and patchy distributions are frequent phenomena; and distribution patterns are associated with island type (Kay, 1979). The biota of atolls are composed of few species, most of which are widely distributed on tropical strandlines, while even a slight elevation is accompanied by a substantial increase in the floristic diversity (Sachet, 1967; Fosberg, 1984; van Balgooy et al., 1996). Differences in the flora based on island type relate to the fact that atolls are ephemeral structures, subject to constant change and having resulting ÔweedyÕ native floras exhibiting little endemism, while high islands have significant species diversity where dynamic processes of speciation and extinction occur (Brownlie, 1965; Kay, 1979). We studied the remote Pitcairn group (25
4¢ S, 130
06¢ W), which lies just south of the Tropic of Capricorn, about halfway between New Zealand and South America. The group comprises four islands: Pitcairn, a relatively young, high volcanic Island; Henderson, an uplifted atoll, the uplift caused by the eruption of Pitcairn; and two atolls, Ducie and Oeno. The remote location, relatively young age and range of island types found in the Pitcairn Island group makes the group ideal for the study of island biogeography and evolution. Henderson, Oeno and Ducie Islands are c. 16, 13 and 8 Myr old, respectively, although all have undergone cycles of submergence and re-emergence during that period (Spencer, 1989). Pitcairn Island itself is < 1 Myr old (Spencer, 1989). The atolls of Oeno and Ducie have a limited range of available habitats for plant colonization, and as a result have a total flora of only 18 and three species, respectively (Florence et al., 1995). Waldren et al. (1995) identified four distinct vegetation communities on Oeno; open littoral vegetation, Argusia argentea L.f. scrub, Closed forest and Cocos nucifera L. grove. Henderson, the largest island in the group has 65 species (Florence et al., 1995), having not only strand habitats as on the atolls, but also cliff and inland forest communities on the uplifted makatea. Waldren et al. (1995) identified six main vegetation communities; beachfront communities, embayment forests, open limestone scrub, cliff and ledge communities, exposed cliff top communities and plateau forests. The nature of Pitcairn Island, as a high island, means that a more diverse array of habitats is available than on other islands in the group. Kingston & Waldren (2003) identified eight main vegetation communities; Metrosideros collina (J.R. Forst. & G. Forst.) A. Gray woodland, Homalium taypau H.St. John woodland, Syzygium jambos (L.) Alston woodland, Mixed woodland, weedy scrub, fernlands, coastal rock communities and coastal scrub communities. Unfortunately, as Pitcairn is the only inhabited island in the group, human induced habitat change has resulted in a reduction in the amount of native forest and significant reduction in the populations of the 81 native species. In addition over 250 species have been introduced to the island, several of which have become nuisance invasive species and dominate large tracts of land (Kingston & Waldren, 2003). Aims The aim of this study was to make a biogeographical assessment for the flora of the Pitcairn group, and to determine: • Where the flora came from. • How the species colonized the island. • The factors which explain any distribution patterns found. Using this information it could then be determined how the Pitcairn group relates biogeographically to the south-eastern Polynesian region and if, as a subset of the regions flora, it can then be used as a model for biogeographical analyses. METHODS Several separate but related data sets were compiled from many sources, listed at the end of this section. Only native plants were used as the natural processes of plant migration and dispersal were under study, rather than anthropological progression in the area, and only vascular plants were chosen for use in this study, as these are the groups for which  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Phytogeographical model for south-eastern Polynesia 1313 information is most readily available. Finally, only the 114 indigenous species occurring on the Pitcairn group of islands were used (see Appendix), as the main aim of this study was to consider the origin and formation of the flora of the Pitcairn group, rather than a biogeographical analysis of plant distribution in Polynesia or the Pacific region. observations made on the Islands of Rimatara, Rurutu, Tubuai and Raivavae in the Austral Islands, Mangareva in the Gambier Islands, and Tahiti, Moorea, Huahine and Raiatea in the Society Islands, during 1997 and 2000 were also incorporated. A record was kept of all of the data sources, listed below: Dispersal data Species were also assigned one of five dispersal classes based on the dispersal mechanism employed as follows: 1 ¼ Endozoochory (actively by an animal agent), 2 ¼ Ectozoochory (passively by an animal agent), 3 ¼ Hydrochory (water dispersal), 4 ¼ Anemochory (wind dispersal), 5 ¼ More than one of the dispersal mechanisms 1–4. The number of propagules in a single dispersed unit was also recorded to allow quantification of the number of potential colonizing individuals that could arise from a single dispersal event. Australian Biological Fosberg & Resources Renvoize (1980) Study (1993) Australian Biological Fosberg & Sachet Resources (1967) Study (1994) Bridson (1985) Fosberg & Sachet (1981) Brooke et al. (1996) Fosberg (1973) Brown & Brown Fosberg et al. (1931) (1983) Brown (1931) Gothesson (1997) Brown (1935) Hallé (1980) Distribution data Species were assigned one of seven general distribution classes, based on their global distributions as follows: 1 ¼ Pantropic/Transpacific, 2 ¼ Old World, 3 ¼ IndoMalesian, 4 ¼ Australian, 5 ¼ Pacific, 6 ¼ Polynesian, 7 ¼ Endemic to Pitcairn group. Range data The world-wide range, as presence or absence in defined geographical units (Fig. 1), for each native vascular plant species and genus in the Pitcairn Island group was compiled. Species for which native synonymy or status was uncertain were included (e.g. Hibiscus tiliaceus L., Thespesia populnea (L.) Sol. ex Corrêa); in most cases these species were widespread pantropical Indo-Pacific species, and thus did not affect the resulting data analysis. For some endemic species, especially those for which the synonymy and status was uncertain, the distribution of closely related species was also compiled. Environmental data Environmental parameters for each island group in the Polynesian region were also collated, including: latitude and longitude, land area, altitude, maximum age (that the islands have been above sea level), island types (whether atolls, atoll and makatea, high volcanic islands or all island types), total species number, % endemism, ecosystem number (the number of broad terrestrial or marine ecosystem types, based where possible on an existing classification or estimated from the island description and structure, from Dahl, 1980; UNEP-Earthwatch, 1998) and invasive index (the threat represented by invasive introduced species rated on both on the number of such species and their aggressiveness in island situations, from Dahl, 1980; UNEP-Earthwatch, 1998). Data collation Data were sourced where possible from recently completed floras, but in other cases from taxonomic literature, websites and older floras. In addition, information was taken from various general books relating to the Pacific Islands. Field  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Royal Botanic Gardens Kew (1993) Sherff (1926) Skottsberg (1920–1956) Skottsberg (1922a) Skottsberg (1922b) Brownlie (1961) Brownlie (1965) Heads (1996) Heywood (1978) Brownsey (1977) Cheesman (1903) Holttum (1964) Holttum (1973) Christensen & Skottsberg (1920a) Christensen & Skottsberg (1920b) Clague (1996) Copeland (1932) Copeland (1938) Holttum (1976) Skottsberg (1951) St. John & Philipson (1962) St. John (1987) Stoddart & Sachet (1969) Stoddart (1975) UNEP-Earthwatch (1998) van Balgooy (1971) Holttum (1978) van Balgooy (1975) Copeland (1947) Doty (1954) Florence (1996) Florence (1997a) Florence (1987) Florence (1997b) Florence et al. (1995) Holttum (1985) Huguenin (1974) Hunt et al. (2000) van Balgooy (1984) van Balgooy (1993) van Steenis & van Balgooy (1966) Lam (1938) van Steenis (1963) McCormack (2000) Waldren et al. (1995) Merrill (1947) Waldren et al. (1999) Mueller-Dombois Wheeler & Carillet & Fosberg (1998) (1997) Oliver (1935) Wilder (1934) Rougerie (1995) Yuncker (1937) Royal Botanic Zizka (1991) Gardens Kew (1906–1996) Data analysis Exploratory analysis of the range data were carried out using cluster analyses with the Sørensen coefficient as a distance measure, and group averaging as the clustering procedure (an agglomerative clustering method, sometimes referred to as unweighted pair group method with arithmetic mean (UPGMA). The ordination technique of non-metric multidimensional scaling (NMDS) using the Sørensen coefficient was also carried out to further visualize the data set. In order to determine what environmental variables could explain the variation found in the NMDS dimensions, the final NMDS output for the regions was correlated with the environmental data set, using Spearman rank correlations. PC-ORD for windows ver. 3.2 (McCune & Mefford, 1997) was used for the data analysis. Maps were compiled using ArcView GIS ver. 3.1. 1314 N. Kingston et al. Eurasia North America East Asia South-east Asia Africa Philippines Malesia Hawaiian Islands Central America Micronesia Central Polynesia West Melanesia East Melanesia South America West Polynesia Easter Island Australia Juan Fernandez Islands New Zealand N 5000 0 5000 Kilometers N Central Polynesia Marquesas Islands West Polynesia Cook Islands King George Dissapointment Islands North-west Islands group Pallisier Islands Centre-west Society East group group Islands Centre-east group Duke of Gloucester Islands Austral Islands South Acteon Islands group Gambier Islands Pitcairn Islands Rapa New Zealand 1000 0 1000 Kilometers Figure 1 Geographical units to which species were assigned presence or absence in order to compile their world-wide ranges. The geographical units are delimited based on a combination of van Balgooy (1971) for the larger continents and landmasses, Mueller-Dombois & Fosberg (1998) for the Pacific islands, and UNEP-Earthwatch (1998) for the Tuamotu island subdivisions.  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Phytogeographical model for south-eastern Polynesia 1315 Table 2 shows the dispersal mechanism for species in each of the distribution classes, and some patterns can be established. Pantropic species are mostly water dispersed, with all the wind-dispersed species being pteridophytes. Old world species are also most likely to be water dispersed, but there are an increasing number of zoochorous species. Pacific, Polynesian and endemic species however, are more likely to be dispersed by an animal agent, in particular through bird ingestion. Dispersal using a single propagule dominates all distribution classes, but increasing numbers of species in the Pacific, Polynesian and Endemic classes contain more than one propagule in their dispersed unit. This explains the increased number of Pacific-wide species on Pitcairn, as dispersal is more likely to be successful when the distance is reduced, and a population is most likely to establish if more than one propagule arrives in a dispersal event. As complete species lists are available for all of the islands in the Pitcairn group, it is possible to further analyse distribution and dispersal within the island group. Figure 2 shows the percentage of the total flora in each of the seven worldwide distribution classes. Pitcairn and Henderson show similar patterns, with over 45% of the species being distributed within the Pacific region (Pacific, Polynesian and endemic distribution categories). The atolls of Oeno and Ducie however differ in that the dominant element in their flora is Old World and pantropic. van Balgooy (1960) and Sachet (1967) also conclude that the highest percentages of widely distributed species are found on atolls. Each of the islands in the Pitcairn group show differing patterns in terms of dispersal mechanisms (Fig. 3). The dominant dispersal mechanism for the Pitcairn flora is anemochory (39.5%), explained by the large number of pteridophyte species in the flora, as is typical for high islands. Zoochory is the most common method found in the Henderson flora (59.4%), and hydrochory the most common for both Oeno and Ducie (52.9% and 100% respectively). This pattern is expected due the nature of pantropic and Old World species being dispersed by hydrochory, and the fact that these atolls are dominated by species with such distributions. The smaller atoll floras are also more likely to contain species that can be dispersed in more than one way. This accounts for the success of these species in colonizing even the remotest island (i.e. Ducie), where 66.6% of the flora is dispersed in more than one RESULTS & DISCUSSION Colonization of the flora An assessment of the dispersal mechanisms employed by the flora, found that of the four dispersal mechanisms assigned to the species, zoochory is the most frequent mechanism for dispersal, with active rather than passive methods prevailing (Table 1). Anemochory is also frequent but rarely found in the angiosperms. Zoochorous mechanisms involving a single propagule are by far the most frequent, being found 35 times in the data set (29.7% of cases). Both hydrochory and endozoochory are the mechanisms in which more than one colonizer may be dispersed in a single dispersal event, but in many cases of ectozoochory more than one propagule will attach to the dispersal agent. In theory a single propagule seems unlikely to result in successful dispersal events, however in practice some species are capable of self-fertilization, and many pteridophyte species can reproduce vegetatively by apogamy and apospory (Miller, 1968). Thus even if a single propagule reaches an island it may result in successful colonization. Species with a Pacific-wide distribution dominate the Pitcairn flora (21.1%; see Table 2), with Old world, pantropic and endemic species also forming a large component. There were no species recorded with an American distribution. Pteridophytes tend to have more widespread distributions and the Pitcairn flora is typical, with eight of the 31 pteridophyte species having pantropic distributions. Table 1 Occurrence of the four dispersal mechanism in the Pitcairn flora, and the number of propagules dispersed in each case Propagule number Mechanism* 1 2 3–6 Endozoochory Ectozoochory Hydrochory Anemochory 14 21 14 34 7 1 8 0 Total 83 16 7–10 10þ Total 4 3 3 0 6 0 1 0 1 0 1 0 32 25 27 34 10 7 2 *Eleven species employ more than one mechanism. Thirty-one pteridophytes and three angiosperms (Taeniophyllum fasciola, Metrosideros collina and Senecio stokesii). Table 2 The number of Pitcairn group species occurring in each distribution and dispersal class. The percentage of the total native flora is in parentheses after the species number Dispersal class Distribution class World-wide/ Transpacific Old World Indo-Malesian Australian Pacific Polynesian Endemic  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Total species (%) 19 (16.7) 23 8 6 24 15 19 (20.2) (7.02) (5.26) (21.1) (13.1) (16.7) Endozoochory Ectozoochory Hydrochory Anemochory 1 0 11 8 4 2 2 13 7 10 5 0 1 2 4 6 11 1 0 5 4 1 6 5 3 7 4 2 1316 N. Kingston et al. Geographical origins of the flora 100 % Total species 80 Endemics Polynesian Pacific Australia Indo-Malesian Old World Pantropic 60 40 20 0 Pitcairn Henderson Oeno Ducie Figure 2 Percentage of species in each of the even distribution classes, calculated for each of the islands in the Pitcairn group. % Total species 100 80 2 methods Anemochory Hydrochory Ectozoochory Endozoochory 60 40 20 0 Pitcairn (81) Henderson Oeno (17) (64) Ducie (3) Figure 3 Percentage of species in each of the five dispersal classes, calculated for each of the islands in the Pitcairn group. Total number of species from each island is indicated in brackets. way (the two species Lepturus repens (G.Forst.) R.Br. & Argusia argentea). A cluster analysis of the Pitcairn flora using the range data set is shown in Fig. 4, with the dispersal mechanisms for each species overlaid. The species endemic to the Pitcairn group form cluster one, and these species are predominately dispersed by zoochory. The second cluster contains all species with narrow distributions in Polynesia, but with various dispersal methods. The third cluster contains a large proportion of water dispersed (hydrochorous) species that dominate the strand lines of all Polynesian islands and the low-lying wooded interiors of atolls. Cluster four contains species that are found in the interior of high islands, with some separation between the taxa found on both volcanic and makatea islands, from those confined to volcanic islands only in the lower part of the cluster. The cluster diagram thus shows putative ecological groupings, suggesting that the species occurring on an island strongly relate to the habitats available, and to a lesser extent to the dispersal ability of the species. It also suggests that species evolve a dispersal mechanism appropriate to habitat, with strand species being water dispersed, while inland and higher altitude species are wind and bird dispersed. A cluster analysis of the range data (Fig. 5) places the Cook, Society and Austral Islands together with East Melanesia, and West Polynesia, the Pitcairn, Marquesas and Gambier Islands and Rapa on branches from this. The atoll cluster grouped all of the Tuamotu islands together, and close to Central Polynesia. At the top of the cluster diagram is a group of the regions to the west of Polynesia, and below is a group of the islands and continents to the east of Polynesia. This shows that the Pitcairn group has the least similarities with Eurasia, which is not represented as it has no common species with the Pitcairn Island group, and America, and Pacific islands to the east of Pitcairn (i.e. Easter, Juan Fernandez and East Pacific Islands), and the highest similarities with the other south-east Polynesian island groups. To further explore these relationships within southeastern Polynesia, Venn diagrams were drawn to show the relationships between the floras of the Society, Austral and Cook Island groups based on the Pitcairnese element in their floras (Fig. 6a). Only seven species were not found on the Austral Islands, but six species are only found in the Austral group. This suggests that the flora of the Pitcairn group is most closely related to the flora of the Austral Islands. Figure 6b shows the comparable relationship for the Austral, Gambier Islands and Rapa, and while Rapa and the Gambier are geographically closer to Pitcairn, they are less closely associated to Pitcairn floristically than the Austral Islands. The strongest associations with the Society, Austral and Cook Islands are because these island groups also contain all of the island types found in the Pitcairn islands and are found at comparable latitudes. However, they have larger floras as they are closer to the main floristic source (the IndoMalesian region) and have a larger area. The relationship is, therefore, due the Pitcairn flora being essentially a subset of the larger flora of these island groups. This is supported by several species being endemic to the Austral Islands, Rapa and the Pitcairn Islands (e.g. Hibiscus australense Fosberg, Peperomia rapensis F.Br., Senecio stokesii F.Br.). Within the Pitcairn group itself, Pitcairn and Oeno showed the lowest similarity, having only 10 shared species, while the highest similarity was between Henderson and Pitcairn with 33 common species (Fig. 6c). However, 17 species were common to Henderson and Oeno, which is over 90% of the Oeno flora. Ducie was not included here as only three widespread species have been ever recorded from the island, and in the most recent survey only two species (Pemphis acidula JR. & G.Forst. and Argusia argentea), both of which are recorded from Pitcairn and Henderson. The 23 species common to Henderson and Pitcairn, but not found on Oeno, are inland species, not supported on the low lying atoll habitats found on Oeno. The 48 species found only on Pitcairn are high volcanic island species found on the high slopes and volcanic soils of Pitcairn. These species would not be tolerant of the limestone and associated dry habitats found on Henderson (e.g. various pteridophytes which generally require a humid habitat). Similarly, the 25 species found only on Henderson are adapted to the dry  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Phytogeographical model for south-eastern Polynesia 1317 Figure 4 Cluster diagram of Pitcairn species based on their distribution in Polynesia. Differences in formatting relate to dispersal mechanism with: Endozoochory (bold); Ectozoochory (small caps); Hydrochory (underlined); Anemochory (italicized). For species dispersed by more then one mechanism the branch label is formatted twice. The clustering procedure uses Sørensen distance and the group average linkage method.  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 1318 N. Kingston et al. Figure 5 Cluster diagram of regions based on the distribution of Pitcairn species. The clustering procedure uses Sørensen distance and the group average linkage method. limestone habitat and absent from the damper or more shaded high island habitats. In addition, several strand species are found on Henderson and not on Oeno, owing to the larger area of the island and the presence of coastal cliff habitats. If the origins of the endemic flora of the Pitcairn group are also considered, the link with the Cook, Society, Austral chain is reinforced. Table 3 shows the island groups with species closely related to the Pitcairn endemics and thus islands from where these endemics may have originated. It should be noted that these relationships are largely speculative and have not been tested phylogenetically, being based both on generic distributions in the Pacific and reviews in the literature. The strongest floristic affinities are found with Rapa and the Austral Islands, as well as the Society and Cook Islands, thus reinforcing the other analyses. The close association with Rapa is unexpected based on the previous analyses, but there are actually only 12 species from the Pitcairn group not found on Rapa. Considering the strength of the association between the Austral and Pitcairn groups, however, the affinities with Rapa may not be due to a direct link, but rather both Pitcairn and Rapa may have been separately colonized by ancestral species from the Austral Islands, which in some cases then speciated separately. An example is demonstrated in the genus Peperomia which has a high degree of genetic plasticity, and thus an ability to radiate and speciate into different habitats. Molecular analyses have indicated that P. pitcairnensis (Lauterb.) C. DC is related to P. rapensis, a species also found in the Australs and Rapa. Peperomia hendersonensis Yuncker and an as yet undescribed species from Pitcairn are closely related to P. pallida (G.Forst.) A.Dietr. (distributed from Tonga in the west to French Polynesia in the east), and detailed molecular studies have revealed that P. hendersonensis and P. sp. nov. are most closely related to populations on P. pallida on the Austral Islands of Rurutu and Raivavae (Bradley, 2002). In addition, there are five other species common between Rapa and the Pitcairn group with closely related species in the Austral Islands, so there is undoubtedly a strong biogeographical link between the islands. Causal factors of these distribution patterns In order to determine if environmental gradients could account for the distribution patterns found, an NMDS ordination of the range data set for the Polynesian island groups was carried out, and the environmental data for each island group was correlated with the dimension scores. A five-dimensional solution gave the lowest stress values, with 23% of the variation explained by dimension 1, 66% by dimension 2, and < 5% by each of axes 3–5. A Monte Carlo test of 100 runs was carried out with the NMDS to test for robustness in the data set and showed P < 0.01 for all dimensions. The Pitcairn group had a score of almost zero on dimension 1 and ordinates closely to all of the island groups with the exception of the Hawaiian group (Fig. 7). Longitude and total number of plants were the only environmental variables to correlate with dimension 1 (P < 0.05 in both cases; Table 4) and they would explain the separation of the Hawaiian islands, as they have a much higher total plant number and a higher longitude than the other island groups. Dimension 2 separated the Tuamotu islands with negative scores from the other island groups with positive scores, and correlated with altitude, island type, invasive index, % plant endemism (P < 0.01 in all cases) and total plants  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Phytogeographical model for south-eastern Polynesia 1319 (a) Society Islands Austral Islands 8 1 6 66 5 2 1 Cook Islands (b) Austral Islands Gambier Islands 19 12 1 34 17 2 4 Rapa & Marotiri rocks (c) Henderson Islands Pitcairn Islands 23 48 25 10 0 7 1 Oeno Atoll Figure 6 Venn diagrams to show the relationship between the Pitcairnese element in floras of: (a) the Society, Austral and Cook Islands; (b) the Austral and Gambier Islands and Rapa; (c) the islands in the Pitcairn group. Each circle represents the flora of the island group, with the overlap representing the number of common species in the floras of the islands.  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 (P < 0.05). Much more of the variation is explained by dimension 2 (66%) and so this may explain why so many environmental factors correlate. This division would be expected as the Tuamotu islands only contains the island type of atolls which are by nature low in altitude and have few endemic species. In addition, they tend to be less inhabited and so have fewer invasive species. As climatic factors, sea currents, sea levels and island geologies do not remain static, it is necessary to look at past environments in discussing the plant colonization of the Pitcairn group. In the period since the formation of the islands in the Pitcairn group, sea levels have probably not been lower than c. 150 m below current sea levels (Menard, 1964; Spencer, 1989; Pirazzoli, 1996). However, during periods of high sea levels before the formation of Pitcairn, Oeno, Ducie and Henderson (then an atoll) may have been totally submerged during certain periods, thus removing all of the terrestrial biota. Since the formation of Pitcairn and subsequent uplift of Henderson, there have been terrestrial habitats above water even during periods of high sea levels. During periods of low sea level there are several seamounts that may have been above water level and supported a terrestrial biota. Figure 8 shows the locations of seamounts and reefs that may have formed islands during such periods, and there would have been a number of emergent islands in the vicinity of Pitcairn, most notably to the west. There are also a number of seamounts and reefs around the Austral Islands, Rapa, the Cook Islands and across towards Fiji (in East Melanesia and the source of most of the Polynesian flora). Aside from a volcanic seamount close to Pitcairn, and another near Henderson, there are no seamounts or reefs between Pitcairn and Easter Island. The presence of these potential islands during cooler, windier and stormier periods would have allowed them to act as additional stepping stones for species to migrate to and colonize Pitcairn. Stepping stone islands are important to species whose propagules tend to be dispersed by zoochory, hydrochory or on floating rafts (e.g. reptiles, small mammals and arthropods can be carried on floating vegetation), and less important to anemochorous species (MacArthur & Wilson, 1967). In van Balgooy’s (1960, 1971) biogeographical regional analyses of the Pacific, Rapa is segregated into its own region, separate to the Polynesian region, to which it is geologically more associated. This division is based on the fact that the flora is more closely associated with that of Australia and New Zealand and forms a more southern element than that of Polynesia. It is interesting to note that there are several submerged reefs and banks to the south-west of Rapa towards New Zealand that could have acted as stepping stone islands for non-Polynesian species and genera to colonize Rapa (e.g. Corokia – found in New Zealand, Australia and Rapa; Hebe – found in New Zealand, Australia, Rapa and the Falklands). This theory might also account for some of the southern transpacific element found in the Pitcairn flora, notably Samolus repens, found in Australia, New Zealand, Pitcairn, Easter Island and South America, and Asplenium obtusatum G. Forst. found in Australia, New Zealand, Polynesia, Easter Island and South America. 1320 N. Kingston et al. Table 3 Locations of species that may be closely related to the endemic species found in the Pitcairn group Genus Abutilon Alyxia Angiopteris Bidens Coprosma Ctenitis Geniostoma Glochidion Haloragis* Homalium Ixora Meryta Myrsine Nesoluma Peperomia Total East Melanesia New Zealand West Polynesia Cook Island 4 4 4 4 4 4 4 4 4 4 4 4 Society Island Austral Island 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 11 11 13 4 4 4 4 3 5 Rapa 4 4 4 4 4 4 4 4 Marquesas Island Tuamotu Island Gambier Island 4 4 4 4 4 4 4 4 4 4 4 Hawaiian Island 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 13 9 7 7 4 3 *Putative endemic, the taxonomy of the Pitcairn material is uncertain. Table 4 Spearmans rank correlation for environmental variables with NMDS dimensions 1 and 2 Latitude Longitude Area (km2) Altitude Maximum age (Myr) Island type Ecosystem number Invasive index Total plants % Plant endemism Dimension 1 Dimension 2 0.391n.s. 0.706* 0.418n.s. 0.454n.s. 0.248n.s. 0.470n.s. 0.195n.s. 0.387n.s. 0.602* 0.397n.s. 0.356n.s. 0.150n.s. 0.153n.s. 0.798** )0.395n.s 0.791** 0.332n.s. 0.758** 0.679* 0.839** n.s. > 0.05; *P < 0.05; **P < 0.01. Figure 7 NMDS ordinations, with dimension 1 plotted against dimension 2, for Polynesian islands. Key: Cent.-E, Centre east group; Cent.-W, Centre west group; Diss., Disappointment islands; King, King George islands; Marq., Marquesas islands; Pall., Pallisier islands. Arrows indicate direction of environmental variables. Approximately 40 birds have been recorded from either the islands of the Pitcairn group or crossing the ocean area between the islands, and at least four landbirds became extinct during the period of Polynesian habitation (Brooke, 1995; Wragg, 1995). Most of the sea birds and migrants are opportunists, feeding primarily on fish or squid, but the landbirds are commonly either omnivores or frugivores (Brooke & Jones, 1995; Imber et al., 1995; Jones et al., 1995; Trevelyan, 1995). Studies on the feeding biology of the Henderson fruit dove (Ptilinopus insularis North) show that it feeds on the fruits of 19 species (Brooke & Jones, 1995). Two extinct frugivores, a ground dove (Gallicolumba sp.) and a pigeon (Ducula sp.), would have browsed from a similar number of species, but also probably from species with larger fruits (e.g. Santalum insulare Bertero) (Brooke &  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Phytogeographical model for south-eastern Polynesia 1321 Figure 8 Locations of seamounts and reefs that may have emergent islands during periods of lowered sea levels ()150 m). Potential islands are marked with grey crosses. Figure is derived from charts 4607 and 4061 of the Admiralty series. Jones, 1995). In addition, the migrant Bristle-thighed curlew (Numenius tahitensis Gmelin) and the Stephen’s Lory (Vini stepheni North) have been observed feeding on fruits (Brooke & Jones, 1995). These frugivores would have acted as dispersers for the plant species through the Pitcairn islands and farther afield, and there are almost certainly other migrant and vagrant birds that feed on fruit, but have not been observed doing so in the Pitcairn group. Discrepancies in the phytogeographical patterns Looking at the distribution of species within the Pitcairn group also raises some questions as to why certain species are not more widely dispersed within the archipelago. Homalium taypau is an example of an endemic species that is confined to the volcanic interior slopes of Pitcairn Island. Closely related species occur in the Cook and Austral Islands, but in the Austral Islands a Homalium species is found growing on both volcanic and makatea substrates. Homalium taypau could in theory, therefore, occur on Henderson Island. It seems unusual that a species so dominant on Pitcairn does not occur on nearby Henderson, especially as the species is probably bird dispersed. The explanation may lie in the fact that many Homalium species require a disturbance event (such as a severe storm or cyclone) to flower and set seed (G. McCormack, pers.  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 comm., 2000), and as such disturbances are not common on Pitcairn, this dispersal has not had a chance to occur. Other species that occur on Henderson, but could also occur on Pitcairn include Senecio stokesii, Geniostoma hendersonensis H.St. John and Nesoluma st-johnianum Lam & B. Meeuse (based on observations of these or closely related species occurring on volcanic substrates in the Austral Islands). As well as considering species that could be more widespread within the archipelago, it is also necessary to examine the many species and genera that do not occur in the Pitcairn group but could do so based on their distributions and habitat requirements. Cyrtandra is one such genus occurring in Malesia and the Pacific islands including the Cook, Society, Austral and Marquesas Islands, and has speciated widely on these islands. It is a genus of herbs and shrubs that occur in forest undergrowth, and may be dispersed by both endozoochory and hydrochory (van Steenis & van Balgooy, 1966). It would be ideally suited to forest habitats on Pitcairn Island, but is not found there. Colubrina asiatica Brongn. is a pantropic species, found extensively in Polynesia, and typical of seashore, rocky and forest habitats at low elevations. It is dispersed by flotation and this mechanism has proved very successful in its widespread colonization of remote locations. It would be suited to all of the Pitcairn Islands, 1322 N. Kingston et al. especially Oeno and Henderson. Byttneria is a genus made up mostly of American and Madagascan species. It has an unusual distribution in the Pacific, not being found in Melanesia or western Polynesia, but in the Society, Marquesas and Gambier Islands. It is thus one of the few species in Polynesia derived from recent American origin. It is a genus of forest shrubs and climbers that are dispersed by ectozoochory. As it has a limited distribution in the eastern Pacific, it is perhaps not surprising that Byttneria is not found in the Pitcairn group, but due to the proximity of the Gambier high islands to Pitcairn, it would seem reasonably likely that it would be dispersed there. Because of its remoteness, the flora of the Pitcairn group is a subset of the larger flora of south-east Polynesia, notably the Austral, Cook and Society Islands. Therefore, the fact that these genera and species do not occur may be due to the simple fact that they have not yet colonized. The Pitcairn group is also out of the line of cyclones which affect the Cook, Society and Austral Islands, and thus aid dispersal between these groups. The most recent records found for the Pitcairn group are in many cases coastal species that occur in very small population numbers (e.g. Samolus repens, Ipomoea littoralis Blume, Haloragis sp.) which may be very recent colonizers, or may have been overlooked by previous surveys. This supports the idea that species such as Colubrina asiatica may either have not yet arrived in the group, or may have been overlooked. Similarly these species may have existed in the past on Pitcairn and become extinct due to the natural species turnover cycle which has been suggested for islands which are at equilibrium (MacArthur & Wilson, 1967). Problems with the data set There is a large potential for error in compiling and analysing data of this sort. In many cases, the floras are unknown or under collected, especially as many early collectors were mainly interested in useful and food plants and a full floristic survey has not been carried out for many of the Pacific islands. In addition many collections have incomplete location information, often simply citing an island group or region. These problems are discussed further in van Balgooy (1971), Fosberg (1984) and Chown et al. (1998). Misidentified specimens, and errors in the published record are all too common and seriously affect analyses of this sort (Spellerberg & Sawyer, 1999). This is less of a problem when dealing at the genus level, but may be a major source of error at the species level (van Balgooy, 1971). Examples of this include a record of a Coprosma sp. from Tubuai which was a Psidium sp. Coprosma spp. are some of the most biogeographically important in the Pacific, while Psidium spp. are introduced invasive species. Obsolete synonymy also causes confusion and many of the species from the Pacific region require taxonomic revisions. Because of the nature of island floras containing narrow endemic species, there is also a danger that species may have become extinct since human occupation of the island, and so are not recorded. There are many examples of taxa, in particular useful plants, whose distribution would have a major affect on this analysis, and may have become extinct on many Pacific islands (e.g. Santalum spp. now extinct on Juan Fernandez, remaining in the Austral islands on only one motu). In this data set the presence/absence method has been used, but without accounting for pseudo-absences (taxon exists but has not yet been recorded) or reversal-absences (because of the extinction of the taxon from a geographical region). When compiling this data set it was often difficult to assign a species to a distribution class based on its current area of occupancy, and the low number of Australian and Indo-Malesian species may be due to misclassification. Only vascular plants are considered in this analysis as it is the relationships and origins of the Pitcairn group vascular flora that are of interest here. Further studies should undoubtedly consider other taxonomic groups. However, Carson (1996) suggested that biogeography could not exist without considering plants, as their distributions are not only more stable than in animals, but also frequently serve as specific determinant substrates of various animal species. van Balgooy (1971) argues strongly for the use of the genus as a working unit in biogeographical analyses, mainly because at the genus level one is less likely to encounter problems with synonymy or poorly defined species. He does, however, concede that the species distributions give a clearer and more detailed Ôphytogeographical pictureÕ. Tryon (1986) in his biogeographical analysis of fern species does not consider genera, as he considers the basic biogeographical features to be those of species. Using the genus as a working unit also means that widespread genera (e.g. Asplenium; Peperomia) will show no biogeographical patterns, but the same genera have many species which show biogeographically interesting and useful distributions in the Pacific and Polynesia. Thus studies at the level of genus may fail to pick up some biogeographical patterns at the large scale, the scale of interest in this study. Other disadvantages inherent to the generic method are that genera, whether large or small, are treated at the same level, and also genera are not uniformly known with some being the subject of recent revisions, and some not (van Balgooy, 1960). Ideally a study at this scale would consider island by island distributions and thus be able to pick up more differences by island type, as recommended by Stoddart (1992). Unfortunately, however, as such island by island floristic data are not yet available, an analysis such as that would increase the effect of pseudo-absences in the data set. Steps are being taken to rectify this situation in initiatives such as the ongoing ÔFlore de la Polynésie françaiseÕ (J. Florence, pers. comm., 1997) and the ÔCook Island Natural Heritage ProjectÕ (G. McCormack & E. Saul, pers. comm., 2000). In contrast to Stoddart (1992), however, Whittaker (1998) argues that if island by island data are  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Phytogeographical model for south-eastern Polynesia 1323 used there is a complicating effect of within-archipelago rather than island-mainland distance effects. Implications of this analysis in terms of Pacific biogeography Stoddart (1992) discusses the difficulties associated with existing biogeographical schemes, and suggests taking island type into consideration as the best solution. This study supports this idea, but finds that it is not island type alone, but also habitat availability that may be the most important factors affecting species distributions in the Pacific region. Thus the results here echo MacArthur & Wilson (1967) who also suggested that ultimately habitat availability would be found to be the dominant factor affecting the assembly of island biotas. The results presented here show that as the Pitcairn group contains all of the island types, it provides a simplified model of species distribution patterns. Differences are clearly seen between the floristics of a species poor remote atoll (Ducie), a less remote and slightly more diverse atoll (Oeno), a makatea island (Henderson) and a high island (Pitcairn). The results also suggest that analyses in future should look at the distribution of habitats to explain floristic distributions. However, the same habitats on different islands do not necessarily contain the same species, as some islands have higher species diversity due to other factors. An example would be a comparison between Metrosideros collinadominated cloud forest on Tahiti and Pitcairn. On Tahiti this forest type occurs at high altitudes (1000 mþ; MuellerDombois & Fosberg, 1998), and with many diverse plant species in the community, while on Pitcairn it occurs at lower altitudes (200 mþ) and with fewer plant species diversity. Direct comparison of the similarity between the habitat types, is difficult and would not explain phytogeographical relationships between the islands. However, the consideration of habitat types in conjunction with the actual species present, may prove to resolve many of the problems associated with phytogeographical regionalization systems. CONCLUSIONS The flora of the Pitcairn Islands is derived from the flora of the other island groups in the south-eastern Polynesian region, notably those of the Austral, Society and Cook Islands. The flora would seem to be predominantly derived from that of the Austral Islands. The Polynesian flora is in turn derived largely from the flora of the Indo-Malesian region. The migration methods of the flora are closely linked to the habitats in which the taxa occur, with strand species arriving through hydrochory, while taxa that occur farther inland arriving through zoochory and anemochory. Therefore, it is not simply dispersal of propagules to islands which limits their occurrence, but also the availability of a vacant niche in which they can establish a viable population before events that might lead to their extinction are experienced. In terms of regionalization of the Pacific, the results of this analysis using Pitcairn group floristic data only suggest that  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 south-east Polynesia is a valid biogeographical unit. This unit should contain the Cook, Austral, Society, Marquesas, Gambier, Tuamotu and Pitcairn Islands with Rapa, but exclude Easter Island, Tonga and Samoa. This disagrees with most biogeographical regionalizations published to date, notably in the inclusion of the Cook Islands and exclusion of Easter Island (see van Balgooy, 1971; Stoddart, 1992). However, the closeness of the association between the Pitcairn group and the Cook Islands found in this analysis, closer even than with the Marquesas Islands, suggests that this inclusion is valid. The study presented here contradicts the conclusions of Brown (1935) that the flora of the region is of American origin; no features of the flora of Pitcairn are of American origin. In addition Brown describes the flora as being 87% confined to the Polynesian region, with only two pantropic species. This study also shows this not to be the case. Brown’s (1935) research was based on limited knowledge of the Polynesian floras (his work was confined to the Marquesas Islands, and most of the other island groups had not yet been surveyed), and was at a time when new species were being described without proper comparisons between collections from other islands (Spellerberg & Sawyer, 1999). Even so, his was the first analysis to note both the high degree of endemism in the flora, and the floristic affinities between Rapa and New Zealand (van Balgooy, 1971). Stoddart (1992) discussed the limitations of previous regional schemes, but did not carry out an analysis or add to the schemes already in place. He did however recommend that analysis should take island type into consideration when developing regional schemes, a factor taken into consideration in this study and indeed shown to have an effect. Having struggled with the problem of the phytogeographical regionalization of the Pacific (van Balgooy, 1960, 1971), van Balgooy et al. (1996) supports Stoddart’s view that a hierarchical system of regions is impractical and classifications should be based on the patterns shown in analyses based on island type. Although the flora of the Pitcairn group is essentially a subset of the flora of the Polynesian region, the results of this analyses still show the biogeographical relationships found when using a larger data set for the whole of Pacific. This demonstrates that although the Pitcairn Island group has a impoverished and disharmonic flora, due to its age and location, the fact that the group contains all of the oceanic island types found in the region, and thus a representative sample of the habitats available in the region, means it can serve as a useful model for Pacific biogeographical analyses. ACKNOWLEDGMENTS We thank the Pitcairn Island Council and Pitcairn Islands Commission for logistical support in the field. Support was received from UK Foreign and Commonwealth Office, Linnaean Society of London, Royal Geographic Society, Trinity College Dublin Association and Trust, Royal Horticultural Society, Systematics Association, Merlin Trust, Percy Sladen Memorial Fund, Oleg Polunin Trust, Air New 1324 N. Kingston et al. Zealand, Air Tahiti and Skye Instruments. Thanks to Dr Mike Brooke and Dr Jacques Florence for species information, and to Dr Jane Stout for comments on this manuscript. REFERENCES Australian Biological Resources Study (1993) Flora of Australia: Vol. 50. Oceanic Islands 2. AGPS Press, Canberra. Australian Biological Resources Study (1994) Flora of Australia: Vol. 49. Oceanic Islands 1. AGPS Press, Canberra. van Balgooy, M.M.J. (1960) Preliminary plant-geographical analysis of the pacific. Blumea 10, 385–430. van Balgooy, M.M.J. (1971) Plant-geography of the Pacific. Blumea Supplement 6, 1–222. van Balgooy, M.M.J. (ed.) (1975) Pacific Plant Areas – 3. Rijksherbarium, Leiden. van Balgooy, M.M.J. (ed.) (1984) Pacific Plant Areas – 4. Rijksherbarium, Leiden. van Balgooy, M.M.J. (ed.) (1993) Pacific Plant Areas – 5. Rijksherbarium, Leiden. van Balgooy, M.M.J., Hovenkamp, P.H. & van Welzen, P.C. (1996) Phytogeography of the Pacific – floristic and historical distribution patterns in plants. The origin and evolution of Pacific island biotas, New Guinea to Eastern Polynesia: Patterns and Processes (ed. by A. Keast & S.E. Miller ), pp. 191–213. SPB Academic Publishing, Amsterdam. Bradley, U. (2002) Biogeography and speciation of the genus Peperomia Ruiz and Pavon in Eastern Polynesia. Unpublished PhD Thesis, Trinity College, Dublin. Bridson, D.M. (1985) The reinstatement of Psydrax (Rubiaceae, subfam. Cinchonoideae, tribe Vanguerieae) and a revision of the African species. Kew Bulletin, 40, 687–725. Brooke, M.D.L. (1995) The modern avifauna of the Pitcairn Islands. Biological Journal of the Linnean Society 56, 199–212. Brooke, M.D.L. & Jones, P.J. (1995) The diet of the Henderson fruit dove Ptilinopus insularis I. Field observations of fruit choice. Biological Journal of the Linnean Society 56, 149–165. Brooke, M.D.L., Jones, P.J., Vickery, J.A. & Waldren, S. (1996) Seasonal patterns of leaf growth and loss, flowering and fruiting on a subtropical Central Pacific Island. Biotropica 28, 164–179. Brown, F.B.H. (1931) Flora of south-eastern Polynesia: I Monocotyledons. Bulletin of the Bernice P. Bishop Museum 84, 1–194. Brown, F.B.H. (1935) Flora of south-eastern Polynesia: III Dicotyledons. Bulletin of the Bernice P. Bishop Museum 130, 1–386. Brown, E.D.W. & Brown, F.B.H. (1931) Flora of south-eastern Polynesia: II Pteridophytes. Bulletin of the Bernice P. Bishop Museum 89, 1–123. Brownlie, G. (1961) Studies of Pacific ferns, Part IV: the pteridophyte flora of Pitcairn Island. Pacific Science 15, 297–300. Brownlie, G. (1965) The geographical affinities of the south Pacific island fern floras. Pacific Science 19, 219–223. Brownsey, P.J. (1977) A taxonomic revision of the New Zealand species of Asplenium. New Zealand Journal of Botany 15, 39–86. Carson, H.L. (1996) Pacific Basin biotas and evolutionary theory. The origin and evolution of Pacific island biotas, New Guinea to Eastern Polynesia: patterns and processes (ed. by A. Keast & S.E. Miller ), pp. 7–18. SPB Academic Publishing, Amsterdam. Cheesman, T.F. (1903) The flora of Rarotonga, the chief island of the Cook Group. Transactions of the Linnean Society of London (1) Botany 7, 261–313. Chown, S.L., Gremmen, N.J.M. & Gaston, K.J. (1998) Ecological biogeography of Southern Ocean Islands: speciesarea relationships, human impacts, and conservation: The American Naturalist 152, 562–575. Christensen, C. & Skottsberg, C. (1920a) The ferns of Easter Island. The natural history of Juan Fernandez and Easter Island, Vol. 2: botany (ed. by C. Skottsberg), pp. 47–53. Almqvist & Wiksells, Uppsala. Christensen, C. & Skottsberg, C. (1920b) The Pteridophyta of the Juan Fernandez Islands. The natural history of Juan Fernandez and Easter Island, Vol. 2: botany (ed. by C. Skottsberg), pp. 1–45. Almqvist & Wiksells, Uppsala. Clague, D.A. (1996) The growth and subsidence of the Hawaiian-Emperor volcanic chain. The origin and evolution of Pacific island biotas, New Guinea to Eastern Polynesia: patterns and processes (ed. by A. Keast & S.E. Miller), pp. 35–50. SPB Academic Publishing, Amsterdam. Copeland, E.B. (1932) Pteridophytes of the Society Islands. Bulletin of the Bernice P. Bishop Museum 93, 1–85. Copeland, E.B. (1938) Ferns of south-eastern Polynesia. Occasional Papers of the Bernice P. Bishop Museum 14, 45–101. Copeland, E.B. (1947) Genera Filicum. Chronica Botanica Company, Waltham, MA. Dahl, A.L. (1980) Regional ecosystems survey of the South Pacific Area, pp. 1–99. South Pacific Commission, Noumea, New Caledonia. Doty, M.S. (1954) Floristic and ecological notes on Raroia. Atoll Research Bulletin 33, 1–41. Florence, J. (1987) Endémisme et évolution de la flore de la Polynésie française. Bulletin de la Societe Zoologique de France 112, 369–380. Florence, J. (1996) Gallicae Polynesiae florae Praecursores: 1. Nouveautes taxonomiques dans les Euphorbiaceae, Piperaceae et Urticaceae. Adansonia 18, 239–274. Florence, J. (1997a) Flore de la Polynésie française. Editions de l’Orstom, Paris. Florence, J. (1997b) New species of Alyxia Banks ex. R. Brown (Apocynaceae) and Glochidion J.R. & J.G. Forster (Euphorbiaceae) from the Pitcairn Islands (South East Pacific). Novon 7, 27–31. Florence, J., Waldren, S. & Chepstow-Lusty, A.J. (1995) The flora of the Pitcairn Islands: a review. Biological Journal of the Linnean Society 56, 79–119. Fosberg, F.R. (1973) Caesalpinia major a legitimate name. Taxon 22, 162–163. Fosberg, F.R. (1984) Phytogeographic comparison of Polynesia and Micronesia. Biogeography of the tropical Pacific (ed. by F.J. Radovsky, P.H. Raven & S.H. Sohmer), pp. 33–44. Bernice P. Bishop Museum, Honolulu. Fosberg, F.R. & Renvoize, S.A. (1980) The flora of Aldabra and neighbouring islands. Kew Bulletin Additional Series. Her Majesty’s Stationery Office, London. Fosberg, F.R. & Sachet, M.-H. (1967) Plants of south-eastern Polynesia. Micronesica 2, 153–159.  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Phytogeographical model for south-eastern Polynesia 1325 Fosberg, F.R. & Sachet, M.-H. (1981) Polynesian plant studies 6–18. Smithsonian Contributions to Botany 47, 1–38. Fosberg, F.R., Sachet, M.H. & Stoddart, D.R. (1983) Henderson Island (south-eastern Polynesia): summary of current knowledge. Atoll Research Bulletin 272, 1–47. Gothesson, L.-A. (1997) Plants of the Pitcairn Islands including local names and uses. University of New South Wales, Sydney. Gressit, J.L. (ed.) (1963) Pacific Basin biogeography. Bishop Museum Press, Honolulu. Hallé, N. (1980) Les Orchidees de Tubuai (archipel des Australes, Sud Polynésie). Cahiers de l’Indo-Pacifique 2, 69–130. Heads, M.J. (1996) Biogeography, taxonomy and evolution in the Pacific genus Coprosma (Rubiaceae). Candollea 51, 381–405. Heywood, V.H. (1978) Flowering plants of the world. Batsford, London. Holttum, R.E. (1964) The tree-ferns of the genus Cyathea in Australasia and the Pacific. Blumea 12, 241–274. Holttum, R.E. (1973) Studies in the family Thelypteridaceae V.: the genus Pneumatopteris Nakai. Blumea 21, 293–325. Holttum, R.E. (1976) The genus Christella Leveille, sect. Christella; studies in the family Thelypteridaceae, XI. Kew Bulletin 31, 293. Holttum, R.E. (1978) The morphology and taxonomy of Angiopteris (Marattiaceae) with description of a new species. Kew Bulletin 32, 587–594. Holttum, R.E. (1985) Studies in the fern genera allied to Tectaria Cav. IV. The genus Ctenitis in Asia, Malesia and the Western Pacific. Blumea 31, 1–38. Huguenin, B. (1974) La Végètation des ı̂les Gambier: rélevé botanique des éspeces introduites. Cahier Pacifique 18, 459–471. Hunt, E., Carillet, J.-B., Galbraith, K., Jones, R., Keller, N., Lyon, J., McKinnon, R., O’Byrne, D., Pinheiro, L. & Wheeler, T. (2000) South Pacific. Lonely Planet Publications, Australia. Imber, M.J., Jolly, J.N. & Brooke, M.d.L. (1995) Food of three sympatric gadfly petrels (Pterodroma spp.) breeding on the Pitcairn Islands. Biological Journal of the Linnean Society 56, 233–240. Jones, P., Schubel, S., Jolly, J., Brooke, M.d.L. & Vickery, J. (1995) Behaviour, natural history, and annual cycle of the Henderson Island Rail Porzana atra (Aves: Rallidae). Biological Journal of the Linnean Society 56, 167–183. Kay, E.A. (1979) Little worlds of the Pacific – an essay on Pacific Basin biogeography. University of Hawaii Harold L. Lyon Lectures 9. Kingston, N.E. & Waldren, S. (2003) The plant communities and environmental gradients of Pitcairn Island: the significance of invasive species and the need for conservation management. Annals of Botany 92, 31–40. Lam, H.J. (1938) Monograph of the genus Nesoluma (Sapotaceae): a primitive Polynesian endemic of supposed Antarctic origin. Occasional Papers of the Bernice P. Bishop Museum 14 (9), 127–165. MacArthur, R.H. & Wilson, E.O. (1967) The theory of island biogeography. Monographs in Population Biology. Princeton University Press, Princeton, NJ. McCormack, G. (2000) List of Cook islands native plants. Rarotonga. Unpublished Report, Cook Islands Natural Heritage Project.  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 McCune, B. & Mefford, M.J. (1997) Multivariate analysis of ecological data. MjM Software, Gleneden Beach, OR. Menard, H.W. (1964) Marine geology of the Pacific. McGrawHill, New York. Merrill, E.D. (1947) A botanical bibliography of the islands of the Pacific. Contributions from the United States National Herbarium 30, 1–403. Miller, J.H. (1968) Fern gametophytes as experimental material. Botanical Review 34, 361–440. Mueller-Dombois, D. & Fosberg, F.R. (1998) Vegetation of the tropical Pacific islands. Springer-Verlag, New York. Oliver, W.R.B. (1935) The genus Coprosma. Bulletin of the Bernice P. Bishop Museum 132, 1–207. Pirazzoli, P.A. (1996) Sea-level changes – the last 20,000 years. John Wiley & Sons, Chicester. Rougerie, F. (1995) Nature and functioning of the atolls of Tuamotu archipelago (French Polynesia). Oceanologica Acta 18, 61–78. Royal Botanic Gardens Kew (1906–1996) Index Filicum. Royal Botanic Gardens Kew, London. Royal Botanic Gardens Kew (1993) Index Kewensis. Oxford University Press, Oxford. Sachet, M.-H. (1967) Coral islands as ecological laboratories. Micronesica 3, 45–49. Schouw, J.F. (1823) Gründzuge einer allegemeinen Pflanzengeographie. G. Reimer, Berlin. Sherff, E.E. (1926) Studies in the genus Bidens. Botanical Gazette. 70, 89–109. Skottsberg, C.J.F. (ed.) (1920–1956) The natural history of Juan Fernandez and Easter Island. Almqvist & Wiksells, Uppsala. Skottsberg, C. (1922a) The phanerograms of Easter Island. The natural history of Juan Fernandez and Easter Island, Vol. 2: botany (ed. by C. Skottsberg), pp. 61–83. Almqvist & Wiksells, Uppsala. Skottsberg, C. (1922b) The Phanerograms of the Juan Fernandez Islands. The natural history of Juan Fernandez and Easter Island, Vol. 2: botany (ed. C. Skottsberg), pp. 95–238. Almqvist & Wiksells, Uppsala. Skottsberg, C. (1951) A supplement to the pteridophytes and phanerograms of Juan Fernandez and Easter Island. The natural history of Juan Fernandez and Easter Island, Vol. 2: botany (ed. C. Skottsberg), pp. 763–792. Almqvist & Wiksells, Uppsala. Spellerberg, I.F. & Sawyer, J.W.D. (1999) An introduction to applied biogeography. Cambridge University Press, Cambridge, MA. Spencer, T. (1989) Tectonic and environmental histories in the Pitcairn group, palaeogene to present: reconstructions and speculations. Atoll Research Bulletin 322, 1–41. St. John, H. (1987) An account of the flora of Pitcairn island with new Pandanus species. Pacific Plant Studies 46. St. John, H. & Philipson, W.R. (1962) An account of the flora of Henderson Island, South Pacific Ocean. Transactions of the Royal Society of New Zealand 1, 175–194. van Steenis, C.G.G.J. (ed.) (1963) Pacific plant areas – 1. Monographs of the National Institute of Science and Technology. National Institute of Science and Technology, Manila. van Steenis, C.G.G.J. & van Balgooy, M.M.J. (eds) (1966) Pacific plant areas – 2. Rijksherbarium, Leiden. Stoddart, D.R. (1975) Vascular plants of Aitutaki. Atoll Research Bulletin 190, 87–122. 1326 N. Kingston et al. Stoddart, D.R. (1992) Biogeography of the tropical Pacific. Pacific Science 46, 276–293. Stoddart, D.R. & Sachet, M.-H. (1969) Reconnaissance geomorphology of Rangiroa Atoll, Tuamotu Archipelago, with a list of vascular flora of Rangiroa. Atoll Research Bulletin 125, 1–44. Trevelyan, R. (1995) The feeding ecology of Stephen’s lory and nectar availability in its food plants. Biological Journal of the Linnean Society 56, 185–197. Tryon, R. (1986) The biogeography of species, with special reference to ferns. Botanical Review 52, 117–156. UNEP-Earthwatch (1998) Island Directory. http://www. unep.ch/islands/isldir.htm. Waldren, S., Florence, J. & Chepstow-Lusty, A.J. (1995) Rare and endemic vascular plants of the Pitcairn islands: a conservation appraisal. Biological Conservation 74, 83–98. Waldren, S., Weisler, M.I., Hather, J.G. & Morrow, D. (1999) The non-native vascular plants of Henderson Island, South Central Pacific Ocean. Atoll Research Bulletin 463, 1–14. Wheeler, T. & Carillet, J.-B. (1997) Tahiti and French Polynesia. Lonely Planet Publications, Australia. Whittaker, R.J. (1998) Island biogeography: ecology, evolution and conservation. Oxford University Press, Oxford. Wilder, G.P. (1934) The flora of Makatea. Bulletin of the Bernice P. Bishop Museum 120, 1–55. Wragg, G.M. (1995) The fossil birds of Henderson Island, Pitcairn Group: natural turnover and human impact, a synopsis. Biological Journal of the Linnean Society 56, 405–414. Yuncker, T.G. (1937) Revision of the Polynesian species of Peperomia. Bulletin of the Bernice P. Bishop Museum 143, 1–73. Zizka, G. (1991) Flowering plants of Easter Island. Palmarum Hortus Francofurtensis 3, 1–108. BIOSKETCHES Naomi Kingston is currently a postdoctoral researcher in the Department of Botany of Trinity College Dublin working on disjunct phytogeographical patterns in the European flora. Her PhD (2001) involved research in a broad number of areas including floristics, vegetation analysis and mapping, conservation genetics and island biogeography, using the study site of the Pitcairn Islands. Steve Waldren is Curator/Administrator of Trinity College Botanic Gardens. His current research interests include biogeography of the Irish and Polynesian floras, the physiological ecology of turlough (seasonal lakes) plants and plant conservation biology. Una Bradley recently completed her PhD on biogeography and speciation of Peperomia in the islands of southeastern Polynesia. Appendix List of the native vascular plants species for the Pitcairn group of islands, which are the species used in this analysis. The range column relates to the range of the species within the Pitcairn group as follows: P, Pitcairn; H, Henderson; O, Oeno; D , Ducie Family Genus Species Authority Range Adiantaceae Aizoaceae Apiaceae Apocynaceae Apocynaceae Apocynaceae Araliaceae Aspidiaceae Aspidiaceae Aspleniaceae Aspleniaceae Aspleniaceae Aspleniaceae Aspleniaceae Aspleniaceae Aspleniaceae Asteraceae Asteraceae Asteraceae Blechnaceae Boraginaceae Boraginaceae Boraginaceae Brassicaceae Capparidaceae Convolvulaceae Convolvulaceae Adiantum Sesuvium Apium Alyxia Alyxia Cerbera Meryta Ctenitis Diplazium Arachniodes Asplenium Asplenium Asplenium Asplenium Lastreopsis Loxoscaphe Bidens Bidens Senecio Doodia Argusia Cordia Heliotropium Lepidium Capparis Ipomoea Ipomoea hispidulum portulacastrum prostratum fosbergii scandens manghas brachypoda cumingii harpeodes aristata nidus obtusatum polyodon shuttleworthianum pacifica gibberosum hendersonensis mathewsii stokesii media argentea subcordata anomalum bidentatum cordifolia littoralis macrantha Sw. (L.) L. Labill. Florence Roem. aeiou Schult. L. Harms Holttum T.Moore (G.Forst.) Tindale L. G.Forst. G.Forst. Kunze Tindale T.Moore Sherff Sherff F.Br. R.Br. L. f. Lam. A.Gray Montin Lam. Blume Roem. aeiou Schult. P P, H P H, P P H P P P P, H, O P, H H P P P H, O P H P P, H, O, D H H P, H, O P, H P P, H  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 Phytogeographical model for south-eastern Polynesia 1327 Appendix continued Family Genus Species Authority Range Convolvulaceae Convolvulaceae Cyatheaceae Cyperaceae Davalliaceae Davalliaceae Davalliaceae Davalliaceae Euphorbiaceae Euphorbiaceae Euphorbiaceae Flacourtiaceae Flacourtiaceae Gleicheniaceae Goodeniaceae Gramineae Haloragaceae Hernandiaceae Hernandiaceae Hymenophyllaceae Hymenophyllaceae Lauraceae Leguminosae Leguminosae Leguminosae Liliaceae Loganiaceae Lycopodiaceae Lythraceae Malvaceae Malvaceae Malvaceae Malvaceae Marattiaceae Menispermaceae Myrsinaceae Myrsinaceae Myrtaceae Myrtaceae Nyctaginaceae Nyctaginaceae Nyctaginaceae Oleaceae Ophioglossaceae Ophioglossaceae Orchidaceae Pandanaceae Piperaceae Piperaceae Piperaceae Piperaceae Piperaceae Pittosporaceae Poaceae Poaceae Polypodiaceae Polypodiaceae Polypodiaceae Polypodiaceae Ipomoea Operculina Cyathea Fimbristylis Davallia Nephrolepis Nephrolepis Nephrolepis Chamaesyce Glochidion Glochidion Homalium Xylosma Dicranopteris Scaevola Cenchrus Haloragis Hernandia Hernandia Trichomanes Trichomanes Cassytha Caesalpinia Canavalia Senna Dianella Geniostoma Lycopodium Pemphis Abutilon Hibiscus Hibiscus Thespesia Angiopteris Cocculus Myrsine Myrsine Eugenia Metrosideros Boerhavia Pisonia Pisonia Jasminum Ophioglossum Ophioglossum Taeniophyllum Pandanus Peperomia Peperomia Peperomia Peperomia Peperomia Pittosporum Lepturus Thuarea Phymatosorus Phymatosorus Phymatosorus Pyrrosia pes-caprae turpethum medullaris cymosa solida biserrata cordifolia hirsutula sparrmannii comitum pitcairnense taypau suaveolens linearis sericea calyculatus sp. sonora stokesii endlicherianum tahitense filiformis major rosea glanduligera intermedia hendersonense cernuum acidula pitcairnense australense tiliaceus populnea chauliodonta ferrandianus hosakae aff. niauensis reinwardtiana collina tetrandra grandis umbellifera didymum nudicaule reticulatum fasciola tectorius blanda hendersonensis pitcairnensis rapensis sp. aff. arborescens repens involuta commutatus powellii scolopendria serpens L. (L.) S. Manso (G. Forst.) Sw. Hillebrand (G. Forst.) Sw. (Sw.) Schott (L.) C.Presl (G. Forst.) C. Presl (Boiss.) Hurus. Florence (F. Br.) H.St. John H.St. John (J.R. Forst. aeiou G. Forst.) G.Forst. (Burm.) Underw. Vahl Cav. Indet. L. (F.Br.) Kubitzki C.Presl Nadeaud L. (Medik.) Dandy aeiou Exell (Sw.) DC. (H.St. John) A.C.Sm. Endl. H.St.John (L.) Pic.Serm. J.R. Forst. aeiou G. Forst. Fosberg Fosberg L. (L.) Sol. ex Corrêa Copel. Gaudich H.St. John P H P H P, H P, H, P P, H P, H P P, H P P, H P, H P, H, P P P H P P H, O P, H H H P, H H, P P, H P P P, H, P, H, P P H P P, H P H, O H, O P P, H P P P P, H, P H, P P P H P, H, H P P P, H, P, H  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328 (Blume) DC (J.R. Forst. aeiou G. Forst.) A.Gray G. Forst. R.Br. (J.R. Forst. aeiou G. Forst.) Seem. G. Forst. L. L. (G. Forst.) Rchb. Parkinson ex Z Kunth. Yuncker (Lauterb.) C.DC F.Br. Indet. W. Rich ex A. Gray (G. Forst.) R.Br. (G. Forst.) R.Br. ex Roem. aeiou Schult. (Blume) Pic.Serm. (Baker) Pic.Serm. (Burm.) Pic.Serm. (G.Forst.) Ching O, D O O O O O, D O 1328 N. Kingston et al. Appendix continued Family Genus Species Authority Range Portulacaceae Primulaceae Psilotaceae Rosaceae Rubiaceae Rubiaceae Rubiaceae Rubiaceae Rubiaceae Rubiaceae Rubiaceae Rubiaceae Santalaceae Sapindaceae Sapotaceae Solanaceae Surianaceae Thelypteridaceae Thelypteridaceae Thelypteridaceae Tiliaceae Ulmaceae Urticaceae Urticaceae Verbenaceae Viscaceae Viscaceae Vittariaceae Portulaca Samolus Psilotum Osteomeles Coprosma Cyclophyllum Guettarda Hedyotis Ixora Morinda Psydrax Timonius Santalum Allophylus Nesoluma Lycium Suriana Christella Macrothelypteris Pneumatopteris Triumfetta Celtis Pilea Procris Premna Korthalsella Korthalsella Vittaria lutea repens nudum anthyllidifolia benefica barbatum speciosa romanzoffiensis fragrans myrtifolia odorata polygamus insulare rhomboidalis st-johnianum sandwichense maritima parasitica torresiana costata procumbens pacifica sancti-johannis pedunculata serratifolia platycaula rubescens elongata Sol. ex Seem. Pers. (L.) Beauv. (Sm.) Lindl. W.R.B.Oliver (G.Forst.) N.Hallé aeiou Florence L. (Cham. aeiou Schlech.) Fosberg (Hook. aeiou Arn.) A.Gray A.Gray (G.Forst.) A.C.Sm. aeiou S.P. Darwin (G.Forst.) Robinson Skottsberg (Nadeaud) Radlkofer Lam. aeiou Meeuse A.Gray L. (L.) Lév. (Gaudich) Ching. Holttum G.Forst. G.Planch. Florence (G.Forst.) Wedd. L. (Tieghem) Engler (Tieghem) Lecomte Sw. P, H P P, H P P P, H P, H H, O H P, H P, H H H H H P, H H, O P P P H, O P, H P H H H H P  2003 Blackwell Publishing Ltd, Journal of Biogeography, 30, 1311–1328