Hawaiian angiosperm radiations of North - Annals of Botany

31 Pages·2010·1.64 MB·English

Annalsof Botany105:849–879, 2010
doi:10.1093/aob/mcq052,availableonlineat www.aob.oxfordjournals.org
INVITED REVIEW
Hawaiian angiosperm radiations of North American origin
Bruce G.Baldwin1,* and Warren L.Wagner2
1Jepson Herbarium and Department of Integrative Biology, University of California, Berkeley, CA 94720-2465, USA and
2Department of Botany, National Museum of Natural History, Smithsonian Institution, Washington 20013-7012, DC, USA
*For correspondence. E-mail [email protected]
Received:15November2009 Returnedforrevision:10December2009 Accepted:1February2010 Publishedelectronically:9April2010 D
o
w
n
lo
†BackgroundPutativephytogeographicallinksbetweenAmerica(especiallyNorthAmerica)andtheHawaiian a
d
IslandshavefiguredprominentlyindisagreementanddebateabouttheoriginofPacificflorasandtheefficacyof e
d
long-distance(oversea)plantdispersal,giventheobstaclestoexplainingsuchmajordisjunctionsbyvicariance. fro
†Scope Review of past efforts, and of progress over the last 20 years, toward understanding relationships of m
Hawaiian angiosperms allows for a historically informed re-evaluation of the American (New World) contri- h
butionto Hawaiian diversityandevolutionaryactivityofAmericanlineages in an insularsetting. ttps
†Conclusions Temperate and boreal North America is a much more important source of Hawaiian flora than ://a
suggested by most 20th century authorities on Pacific plant life, such as Fosberg and Skottsberg. Early views ca
d
ofevolutionastooslowtoaccountfordivergenceofhighlydistinctiveendemicswithintheHawaiiangeological e
m
timeframeevidentlyimpededbiogeographicalunderstanding,asdidlackofappreciationfortheimportanceof ic
rare,oftenbioticallymediateddispersaleventsandecologicalopportunityinislandecosystems.Molecularphy- .o
u
logeneticevidenceforNorthAmerican ancestryofHawaiianplantradiations,suchasthe silverswordalliance, p.c
mints,sanicles,violets,schiedeasandspurges,underlinesthepotentialoflong-distancedispersaltoshapefloras, o
m
inaccordancewithhypotheseschampionedbyCarlquist.Characteristicsimportanttocolonizationoftheislands, /a
o
suchasdispersibilitybybirdsandancestralhybridizationorpolyploidy,andecologicalopportunitiesassociated b
wexittehns‘sivkeydisivlaenrdsisfi’coaftiotenmopfereantdeeomricbolrineaelagcelismoafteNionrtthheAtmroepriiccaalnHoariwgainiiatnhaatrcahreipaemlaognogmthaye mhaovsetbspeeenciekse-yrictho /article
clades of Hawaiian plants. Evident youth of flowering-plant lineages from North America is highly consistent -a
b
with recent geological evidence for lackof high-elevation settings in the Hawaiian chain immediately prior to s
formation ofthe oldest,modern high-elevation island,Kaua‘i. tra
c
t/1
Keywords: Angiosperms, adaptive radiation, disjunctions, flora, island biogeography, long-distance dispersal, 05
ecologicalopportunity,HawaiianIslands,NorthAmerica, Pacific, phytogeography, sky islands. /6
/8
4
9
/9
3
7
INTRODUCTION high-elevation islands, such as Kaua‘i, compared with the 1
4
eastern high-elevation islands, such as Hawai‘i, was reflected b
y
Extreme isolation and endemicity of the Hawaiian flora, with by more single-island endemics and more extensive morpho- g
u
90% of (approx. 1030) native angiosperm species found logical differences between congeneric species on the older e
s
nowhere else (Wagner et al., 1999, 2005a), have made the islands than on the younger islands. His questions about the t o
n
effort to resolve original sources of founder lineages bothfas- relative importance of habitat diversity, evolutionary lability 2
4
cinating and complex (Fig. 1). Engler (1879–1882; Fig. 2A), and paucity of founders in contributing to autochthonous N
o
in his early analysis of Hawaiian phytogeography, and development of the angiosperm flora still remain to be fully v
e
Hillebrand (1888a; Fig. 2B), who produced the first major answered and demonstrate that his comprehension of island m
b
account of the flora of the Hawaiian Islands, concluded that biology was well ahead of its time. er 2
long-distance dispersal of great magnitude, although unlikely, Drawingfromearlyfloristicwork(e.g.Mann,1869;Wawra, 0
1
must have been required for native plants to become estab- 1872–1875), Engler (1879–1882) presented a tabulation of 8
lished in the archipelago. They both noted the association Hawaiian vascular-plant taxa and the distributions of putative
between long-term isolation of the Hawaiian Islands and the extra-Hawaiian relatives, mostly from tropical or subtropical
exceptional proportion of endemic Hawaiian plants, as well areas, for well over one-third of the Hawaiian species and
asthe importance of birds in plant dispersal and of ecological genera. Only 42 (20%) of the indigenous Hawaiian angio-
factors in diversification of the flora. Engler (1879) reasoned sperm relationships proposed by Engler were exclusively
thattheevolutionarypotentialofplantsdispersedtothearchi- with New World (mostly tropical) species, including all
pelago was high because ‘...the newly developed varieties native Compositae and fleshy-fruited Campanulaceae (lobe-
always encountered open terrain, and they had to fight with lioids). Over half of those native angiosperms proposed to
only few competitors’ (p. 130). Hillebrand (1888a) discussed have exclusively New World relatives were noted as having
the contribution of elevational and climatic diversity to ‘... fleshy fruits, easily adhering seeds or fruits, or associations
increase the variety of forms’ (p. XV). Hillebrand (1888a) with wet habitats, all features favourable for dispersal by
also appreciated that the greater age of the western birds (see Carlquist, 1974, 1996).
#The Author2010.Publishedby OxfordUniversity Presson behalfofthe AnnalsofBotanyCompany.Allrights reserved.
ForPermissions, please email: [email protected]
850 Baldwin & Wagner — American origins of Hawaiian flora
Japan North
NORTH PACIFIC OCEAN America
30º 30º
Northern Hawaiian
Mariana Islands
Islands
15º 15º
D
o
Marshall w
n
Islands lo
a
d
e
d
0ºPaGpuuian eNaew Kiribati IsLlainneds 0º from
Solomon h
Islands Marquesas ttp
s
Samoa Islands ://a
Cook c
Vanuatu a
15º Fiji Islands 15º de
Society m
ic
Islands .o
u
p
.c
Australia o
SOUTH PACIFIC OCEAN m
30º Lord Howe 30º /a
o
Island 500 miles b/a
New Zealand rtic
le
-a
b
150ºE 165ºE 180º 165ºW 150ºW 135ºW 120ºW 105ºW stra
c
FIG.1. MapofthePacificBasin.ThedistancefromsouthernCaliforniatotheclosestHawaiianIsland(Hawai‘i)isapprox.3800kmtothesouth-west,andthe t/10
distancefromtheMarquesasIslands(Eiao)toHawai‘iisapprox.3400kmnorthacrosstheequator.MapcourtesyofD.T.Harbaugh. 5
/6
/8
Although Hillebrand (1888a) did not provide a synthetic among botanists about the extent of North American– 49
account of his thoughts on the original source areas for Hawaiian phytogeographical links. Attention to that disagree- /93
7
native Hawaiian plants in his flora (the introduction was ment serves to illuminate the intellectual obstacle during the 1
4
incomplete at the time of his death), he did suggest elsewhere early20thcenturyposedbythewidespreadconceptionofphe- b
y
thatmostHawaiianplantshaveAustralasianaffinity,excluding notypic evolution as too slow to account for highly divergent g
u
themajorityofHawaiianCompositae,whichappearedtohave endemics on young islands. Here we provide historical per- es
closest relatives in America (Hillebrand, 1888b). He also spective on some of those conflicting ideas and review t o
n
recognized that the ‘Hawaiian Islands (are) the only modern evidence for angiosperm dispersal across the vast 2
4
Polynesian Island group which contain a large proportion of (and deep) oceanic barrier between North America and the N
o
indigenous plants with American affinities’ (Hillebrand, Hawaiian Islands, with special focus on Hawaiian taxa from ve
m
1888a, p. XXIX). By ‘American’, Hillebrand (1888a) was temperate and boreal North America, and the ecological b
e
referring at least in part to South America, based on his obstacles and opportunities facing such plants in a tropical r 2
comment about ‘...the important American element of the archipelago. 0
1
8
Andine regions which is apparent in the Hawaiian flora’
(p. XIV). His reference to American sources for Hawaiian
flora evidently mostly did not refer to western North America DISPARATE EARLY IDEAS ON ORIGINS
northofMexico:‘...thenorthwestcoastofAmerica,including OF HAWAIIAN FLORA
California, has until quite recent times only contributed one
Endemism and antiquity of the Hawaiian flora
or two inhabitants of the highest mountains, and has besides
few so-called representative species’ (p. XIV). Early 20th century botanists presented a diversity of views on
Although various hypotheses were subsequently advanced originsoftheHawaiianfloraanddifferedabouttheimportance
to explain the arrival of native Hawaiian plants from source of long-distance dispersal and particular source areas –
areas to the west and south of the islands, the difficulty of especiallyAmerica – aswellastheageandtimingofevolution
explaining a North American element in the Hawaiian flora andassemblyoftheflora.Ingeneral,thehighlevelofHawaiian
bymeansotherthanlong-distancedispersalofintercontinental floral and faunal endemism and distinctiveness of the endemic
magnitude (approx. 3800km) led to major disagreement taxa was viewed as evidence for greater age of the islands
Baldwin & Wagner — American origins of Hawaiian flora 851
A B C
D
o
w
n
lo
a
d
e
d
fro
m
h
D E F ttp
s
://a
c
a
d
e
m
ic
.o
u
p
.c
o
m
/a
o
b
/a
rtic
le
-a
b
s
tra
c
t/1
0
5
/6
/8
G H I 4
9
/9
3
7
1
4
b
y
g
u
e
s
t o
n
2
4
N
o
v
e
m
b
e
r 2
0
1
8
FIG. 2. Scientists who proposed hypotheses on the origins of Hawaiian flora. (A) Heinrich Gustav Adolf Engler (1844–1930), (B) Wilhelm (William)
B. Hillebrand (1821–1886), (C) Alfred Russel Wallace (1823–1913), (D) Forest Buffen Harkness Brown (1873–1954), (E) Douglas Houghton Campbell
(1859–1953), (F) Carl Johan Fredrik Skottsberg (1880–1963), (G) David Daniels Keck (1903–1995), (H) Francis Raymond Fosberg (1908–1993), (I)
Sherwin Carlquist (1930–). A, B and D–H courtesy of the Hunt Institute for Botanical Documentation, Carnegie Mellon University. C courtesy of
SmithsonianInstitutionLibraries,fromWallace(1905).
and/or flora and fauna than is indicated by modern geological past land connections between the Hawaiian Islands and areas
and molecular data (see Price and Clague, 2002; Clague to the south (e.g. Campbell, 1918). Alfred Russel Wallace’s
et al., 2010) and, by some, as evidence for now-untenable (1881; Fig. 1C) perspective on the high level of Hawaiian
852 Baldwin & Wagner — American origins of Hawaiian flora
endemisminIslandLifemayhavebeeninfluential:‘Thesefacts Guppy’s (1906) ideas on origin and assembly of the
undoubtedly indicate an immense antiquity for this group of Hawaiian flora also somewhat resemble the later speculations
islands, or the vicinity of some very ancient land (now sub- of Brown (1921, 1922; Fig. 2D) and Brown and Brown
merged), from which some portion of their peculiar fauna (1933), although not in mode of oversea dispersal or in
might be derived’ (p. 298). timing of events. Brown (1922) proposed three dispersal
waves for Hawaiian and South Pacific angiosperms, with the
vastmajorityoffoundersarrivingdirectlyorindirectlyviaflo-
tation in sea currents from the Isthmus region of Central
Hypotheses of long-distance dispersal to the Hawaiian Islands
America.Heconcludedthattheearliestofthosesupposeddis-
Guppy(1906)championedtheideaofoversealong-distance persal waves included ancestorsofmodernendemic Hawaiian
dispersal by providing extensive evidence for dispersal ability genera (‘...characterized by Compositae and Lobelioideae’,
D
ofHawaiianandotherPacificplantsbasedonfirst-handexper- p. 222) and occurred during the Mesozoic, possibly as early o
w
imentation and detailed observations of fruit and seed charac- as the Jurassic and pre-dating the known angiosperm fossil n
lo
teristics. His conclusion that birds, in addition to physical record. Although Brown (1922) acknowledged a large a
d
forces, were an important agent of Pacific plant dispersal Indo-Malesian element in the Hawaiian flora, he believed ed
was corroborated and extended by subsequent studies (see that even those plants, which he regarded as the latest fro
m
Ridley, 1930; Carlquist, 1974, 1996), although Guppy’s idea natural arrivals, probably descended ultimately from Central h
that high levels of floristic and avian endemism in Hawaiian American ancestors that dispersed west by oceanic rafting. ttp
s
forests meant that birds long ago ceased to move propagules Ridley (1930), in considering dispersal of plants worldwide, ://a
to or from such habitats is at odds with documentation of fre- also noted a ‘distinct American element’ in the Hawaiian c
a
quent bird traffic to diverse sites in the archipelago from con- flora, although he found it ‘...difficult to see how such de
m
tinental areas (see Ballard and Sytsma, 2000). Hillebrand’s plants as have American affinities could have crossed over ic
(1888a) earlier idea that the Pacific golden plover has been 2,000 miles of sea’ (p. 687). .o
u
important in the distribution of Hawaiian plant species is p.c
o
still widely held (see Fosberg, 1951; Johnstone and m
McFarlane, 1967; Carlquist, 1974; Ballard and Sytsma, Vanished land connections and the American ‘stumbling-block’ /a
o
2000; Lowrey et al., 2005; Harbaugh et al., 2009). In contrast to the ideas of Bentham, Wallace, Guppy and b/a
Guppy’s (1906) suggestion that plant dispersal to the Brown, Campbell (1918, 1919, 1920, 1926, 1928, 1932, rtic
Hawaiian Islands occurred in distinct, successive waves, with 1933; Fig. 2D) and Skottsberg (1925, 1939, 1941; Fig. 2F) le
-a
an early and extremely ancient series of dispersals from regarded America as a relatively minor contributor to the b
s
America (during his early Tertiary ‘Age of Compositae’) and Hawaiian flora compared with the South Pacific, Australasia tra
c
a much more recent (post-Pleistocene) wave from and Indo-Malesia and did not believe that long-distance dis- t/1
0
Indo-Malesia was attributed in part to Bentham (1873), who persal alone could account for the high floristic richness of 5
/6
noted that the high level of endemism in Hawaiian the Hawaiian Islands. Bryan (1915) also concluded that ‘... /8
Compositaeshows‘...thattheancientconnexion,ofwhatever few indeed have been the representatives of the North 49
nature it may have been, with America...has been so long American flora that have been brought to the islands’ /93
7
severed as not to have left a single unmodified common form. (p. 191). Campbell subscribed to the Pacific snail expert 1
4
The species are all either descendants altered by long isolation Pilsbry’s (1916) hypothesis of once-continuous land connec- b
y
orpossibly,insomeinstances,preservedremnantsoftypeslong tions among the Hawaiian Islands and between those islands, g
u
since extinct elsewhere’ (p. 556). Guppy also may have been the South Pacific, Australasia and Indo-Malesia, and es
influenced by Wallace’s (1881) extension of Bentham’s Campbell suggested a southern connection with South t o
n
(1873)viewsontheantiquityandoriginofthemorphologically America as well. Subsidence of the hypothetical block con- 2
4
unusual,endemiclineagesofHawaiianCompositae:‘Thegreat necting those areas was suggested to have led to isolation of N
o
preponderance of American relations in the Compositae...is the Hawaiian volcanic peaks as islands, following occupation ve
m
very interesting and suggestive...We may...look upon the of those peaks by ancestors of most lineages of the modern b
e
Compositae as representing the most ancient portion of the flora. Skottsberg (1925) advanced an ‘Antarctic circuit r 2
existing flora of the Sandwich Islands, carrying us back to a theory’, based in part on evidence at that time for an ancient 0
1
8
very remote period when the facilities for communication southern super-continent, with south-to-north overland
with America were greater than they are now...(P)lants of migration from Antarctica into Polynesia and into South
North Temperate affinity may be nearly as old, but these may America to account for floristic similarities across those
have been derived from Northern Asia by way of Japan and regions. He indicated that extending that land connection to
the extensive line of shoals which run northwestward from the the Hawaiian Islands, as suggested by Campbell, was ‘...
Sandwich Islands...Those which exhibit Polynesian or rather daring, but it is the only manner in which to get away
Australian affinities, consisting, for the most part, of less from transpacific bridges, if we do not believe in oversea
highly modified species usually of the same genera, may have migration’ (Skottsberg, 1925, p. 34).
had their origin at a later, though still somewhat remote, Campbell(1919,1920,1926,1928,1933)didbelieveinthe
period, when large islands, indicated by the extensive shoals possibilityof oversea plant transport; he indicated that lackof
tothesouthandsouthwest,offeredfacilitiesforthetransmission NorthAmericandiversityintheHawaiianIslandswassurpris-
of plants from the tropical portions of the Pacific Ocean’ inginlightofrelativelyfavourableprospectsforlong-distance,
(Wallace, 1881, pp. 303–304). trans-Pacific dispersal from north-westAmerica as opposed to
Baldwin & Wagner — American origins of Hawaiian flora 853
over-water dispersal from areas to the south. Although 1879; Wallace, 1881; Hillebrand, 1888; Campbell, 1919;
Campbell (1933) did not believe that over-water dispersal Brown, 1922) or at least possibly from America (Skottsberg,
could account for the high representation of South Pacific, 1925; but see Skottsberg, 1931). Gray’s (1852) taxonomic
Malesian, Australasian or even South American plants in the placement of Argyroxiphium (and the closely related
Hawaiian flora, he believed that the relatively few founders Wilkesia) with the American tarweeds, in subtribe Madiinae,
from the Pacific coast of North America probably were ‘... had largely held sway until Keck’s (1936a) study, wherein
introduced...since the isolation of the islands’ (p. 181). he indicated that Argyroxiphium (including Wilkesia, in his
Skottsberg (1925), who rejected Bryan’s (1921) suggestion classification) was most closely related to the endemic
of a direct land connection across the deep Pacific abyss Hawaiian genus Dubautia (in which he included Railliardia)
between America and the Hawaiian Islands and was highly and that ‘...the affinities of these genera are to be found to
sceptical about the efficacy of oversea dispersal, indicated the south or southwest in the Pacific’ (p. 11), with genera of
D
that ‘...the representatives in (the Hawaiian Islands) of tribe Senecioneae such as Robinsonia (Juan Fernandez o
w
American fauna and flora, even if they are few, are a Islands)orBrachionostylum(NewGuinea),asearliersuggested n
lo
stumbling-block’ (p. 35). bySkottsberg(1931),orBedfordia(Australia)ratherthanwith a
d
American Compositae. Keck’s (1936a) suggestion that ed
Argyroxiphium/Wilkesia and Dubautia/Railliardia represent a fro
Shattering the American–Hawaiian floristic connection? m
commonHawaiiangroupimpactedtwolong-suggestedphyto- h
Keck (1936a; Fig. 2G), in his monograph of the Hawaiian geographical links between endemic Hawaiian genera and ttp
s
silverswords (Argyroxiphium, Compositae), re-evaluated the American taxa (Gray, 1852, 1865, 1870): Argyroxiphium/ ://a
evidence for American ancestry of Hawaiian vascular plants. Wilkesia to the mostly Californian tarweeds (Madiinae; c
a
He concluded that Skottsberg’s (1925) ‘stumbling-block’ of Fig.3B)andRailliardiatothemostlyCalifornianRaillardella de
m
trying to account for movement of plants between America sensu lato (s.l.), which was then widely treated in tribe ic
andtheHawaiianIslandscouldberesolvedindisfavourofphy- Senecioneae(followingBentham,1873)andnotyetrecognized .o
u
togeographicalaffinitiesbetweenthetwoareasexceptforasfew as a tarweed. Keck (1936a) concluded: ‘By thus divorcing p.c
as nine ‘very inconsequential’ late arrivals to the islands, with ArgyroxiphiumfromtheAmericangeneratowhichithasbeen om
closely related congeners on the American mainland. thought related, the most persistently proposed connection /a
o
Keck(1936a)wasprincipallyconcernedwiththeoriginsof between the ancient element in the Hawaiian flora and the b/a
endemic Hawaiian genera and the persistent suggestion that NewWorldhasbeenshattered’(p.11). rtic
some of those highly distinctive groups – especially in le
-a
Compositae – might be descendants of ancestors that arrived b
by long-distance dispersal, which was difficult to escape as Evolutionary inactivity of American lineages stra
in the Hawaiian flora? c
an explanation for phytogeographical ties with America, t/1
0
especially North America. If the endemic Hawaiian genera Fosberg’s(1948;Fig.2H)tabulationofthesourceareasfor 5
/6
of putatively American origin were instead found to be Hawaiian vascular-plant lineages was the first comprehensive /8
more closely related to plants from outside the New World, assessment of extra-Hawaiian affinities of the indigenous 49
then ‘...it is possible to follow one’s inclinations and agree florasinceEngler’s(1879)andprovidedadetailedperspective /93
7
with Campbell, Skottsberg, and others that the endemic ontheimportanceofAmericanfounders.Althoughhedrewin 1
4
element in the Hawaiian flora, at least so far as the flowering partonSkottsberg’s ideas,Fosberg (1948)recognized thatthe b
y
plants are concerned, is for the most part of great age, and Hawaiian flora was assembled by long-distance dispersal g
u
that it may have had its beginnings on nearby land masses in rather than by ancient land connections to other areas, and es
the Pacific even before the present archipelago became habit- he wryly noted: ‘If we resort to land bridges or continents to t o
n
able’ (Keck, 1936a, p. 5). Like otherearly 20th century bota- account for the presence of the Hawaiian flora, then we may 2
4
nists, Keck (1936a) evidently viewed the high degree of well have to build them in all directions’ (p. 119). N
o
endemism in the flora as indicative of greater antiquity than Of his estimated 272 angiosperm immigrants to the ve
m
the islands: ‘Even the flora of the youngest Hawaiian moun- Hawaiian Islands, Fosberg (1948) concluded that regions to b
e
tains is very old, with 91 percent of the species endemic. the west and south of the archipelago, widely believed at the r 2
This is not easily explained by those who regard the timetobethemostimportantsourcesforHawaiianflora,con- 0
1
8
Hawaiian islands as purely oceanic in origin and the flora as tributed the majority of founder species, as corroborated sub-
carried there from great distances across the ocean...’ sequently (e.g. Wright et al., 2001; Lowrey et al., 2001;
(p. 4). He similarly noted that the species of endemic Gemmill et al., 2002; Howarth et al., 2003; Nepokroeff
Hawaiian genera ‘...are usually well-marked, rather static, et al., 2003; Hao et al., 2004; Cronk et al., 2005; Costello
and senile in character’ (p. 5) and that ‘...many species and Motley, 2007; Harbaugh and Baldwin, 2007; Percy
appear to be on the verge of extinction because of their lack et al., 2008; Clark et al., 2009; Harbaugh et al., 2009). His
of plasticity’ (p. 4). table of flowering-plant founders included 109 from the
Keck’s(1936a,b)focusonrelationshipsofHawaiiansilver- ‘Indo-Pacific’(‘IndonesiaorsoutheasternAsiaandattenuating
swords (Argyroxiphium; Fig. 3A) was well chosen to test an out into the Pacific’) and 45 from the ‘Austral’ region (‘south
American–Hawaiian phytogeographical relationship; most Pacific, from Australiato Patagonia’). He also noted that tem-
botanists who had published on origins of the Hawaiian flora perateandtropicalregionsoftheAmericaswerethesourceof
had pointed to Argyroxiphium as an example of an endemic more founder species (50) than expected, based on most con-
genus of American origin (e.g. Bentham, 1873; Engler, temporaryviews,andslightlyexceededthenumberoffounder
854 Baldwin & Wagner — American origins of Hawaiian flora
A B
D
o
w
n
lo
a
d
e
d
fro
m
h
ttp
s
://a
c
a
d
e
m
ic
.o
u
p
.c
o
m
D /a
o
b
/a
rtic
le
-a
b
s
C tra
c
t/1
0
5
/6
/8
4
9
/9
3
7
1
4
b
y
g
u
e
s
t o
n
2
4
N
o
v
e
m
b
e
r 2
0
1
FIG. 3. Exemplars of Hawaiian angiosperm radiations (left) and American relatives (right) in Compositae–Madiinae (top) and Umbelliferae (bottom). 8
(A) Argyroxiphium sandwicense subsp. macrocephalum. Image by W. L. Wagner. (B) Madia elegans. Image by B. G. Baldwin. (C) Sanicula mariversa.
ImagecourtesyofJ.K.Obata.(D)S.arctopoides.ImagebyB.G.Baldwin.
speciesfromthe‘Austral’region,whichwasbelievedbysome had undergone minimal evolutionary change since arriving
to be the most important source area for Hawaiian plant in the archipelago and probably represented recent arrivals,
lineages (e.g. Copeland, 1948). as Keck (1936a) also concluded. Based on Fosberg’s (1948)
Fosberg (1948, 1951) reconciled the high number of tabulations, none of the vascular plant genera of exclusively
Hawaiian vascular-plant founders from temperate and tropical American (and/or boreal) origin in the native Hawaiian flora
regionsoftheNewWorld(50–51of272angiosperms;16–18 had produced more than two species or infraspecific taxa per
of135pteridophytes)withthewidelyheldviewofarelatively original immigrant except for Gunnera (Gunneraceae) and
minor American element in the Hawaiian flora by noting that the endemic Hawaiian genera Hesperomannia (Compositae),
most indigenous Hawaiian plant lineages from the Americas Isodendrion (Violaceae) and Nothocestrum (Solanaceae).
Baldwin & Wagner — American origins of Hawaiian flora 855
Founders of the three endemic genera and Hawaiian RaillardellaunitedthetwoNorthAmerican–Hawaiianphyto-
Psychotria (Rubiaceae) are the only plants of putatively geographical links that Keck (1936a) rejected (i.e.
American origin that he later noted as having ‘...given rise Argyroxiphium/Wilkesia with the tarweeds; Dubautia/
to any significant number of evolutionary offshoots that still Railliardia with Raillardella). Carlquist’s (1959) demon-
survive, and none has produced a large number. stration of a large suite of vegetative and reproductive charac-
Furthermore, all these genera are among the more doubtfully teristics uniting tarweeds and silverswords and his
American of the lot’ (Fosberg, 1951, pp. 204–205). point-by-point refutation of Keck’s (1936a) argument against
Molecular phylogenetic studies have since overturned the a tarweed–silversword relationship provided a convincing
hypotheses of American origins for Hesperomannia (Kim case for a western North American origin of a major
et al., 1998), Nothocestrum (Olmstead et al., 2008) and Hawaiianangiospermradiation.Inpart,Keck’s(1936a)rejec-
Hawaiian Psychotria (Nepokroeff et al., 2003), and upheld a tion of a tarweed ancestry of the silversword alliance was
D
tropical American ancestry for Isodendrion (Feng, 2005; based on the striking differences in habit between the mostly o
w
Tokuoka, 2008) and Hawaiian Gunnera (Wanntorp et al., herbaceous tarweeds and the woody or semi-woody n
lo
2002; Wanntorp and Wanntorp, 2003), which is now treated Hawaiian taxa. Carlquist (1959) suggested that those habital a
d
as including only two species (see Wagner et al., 1990). differences reflected evolution of woodiness in an insular ed
Fosberg’s (1948) suggestion of an American origin for the setting in response to opportunities for year-round growth. fro
m
two species of Hawaiian raspberries (Rubus, Rosaceae) was Among the shared features of the tarweeds and silversword h
confirmed by molecular phylogenetic studies, although his alliance, Carlquist (1965, 1974, 1980) noted that the sticky ttp
s
hypothesis of monophyly of Hawaiian Rubus was rejected bracts associated with tarweed ray fruits and the pappus ://a
(Howarth et al., 1997; Alice and Campbell, 1999; Morden characteristics of tarweed disc fruits allow for ready attach- c
a
et al., 2003); each of the two Hawaiian species evidently menttoanimals,andthatsuchexternaladhesiontobirdfeath- de
m
stems from distinct western North American ancestors, with erswouldhave allowed foroverseatransport from Americato ic
R. hawaiiensis most closely related to the salmonberry the Hawaiian Islands. Like St. John (1950), Carlquist (1959) .o
u
(R. spectabilis) and R. macraei most closely related to the could find no morphological support for Skottsberg’s (1931) p.c
California blackberry (R. ursinus). and Keck’s (1936a) suggestions of a close relationship of the om
silversword alliance to genera of tribe Senecioneae from /a
o
CONTIRNIVBEUSTTIOIGNATTOINHGATWHAEIIAAMNEFRLIOCWANERING A(1u9Ms5t8roaalle,acsubial,aro1r95tph9he)yJlhouygapenontFehteeircsniasnsodtufedzlioeIssnlga-ncddoissn.tfianrmceeddisCpearrslaqluiasnt’ds b/article
PLANT DIVERSITY -a
subsequent adaptive radiation of tarweeds in the Hawaiian b
s
Sincethe1950sandespeciallyoverthelast20years,intensive Islands by showing that the Hawaiian silversword alliance is tra
c
evolutionary studies have greatly improved understanding of a monophyletic group nested within the California tarweed t/1
the origins and relationships of Hawaiian angiosperms. lineage (Baldwin et al., 1991). Based on chloroplast DNA 05
Evidence fromthose efforts forancestrallyAmerican lineages (cpDNA) and nuclear rDNA trees, the Hawaiian clade is /6/8
in the Hawaiian flora is detailed here. more closely related to primarily Californian genera of the 49
‘Madia’ lineage than to other major lineages of Californian /93
7
Madiinae (Baldwin, 1996, 2003). Molecular evidence for 1
Revisiting the silversword alliance evolutionofwoodinessintheHawaiiansettingandecological 4 b
y
Carlquist’s(1958a,b,1959;Fig.2I)comparativeanatomical diversification of all modern members of the silversword alli- g
u
studies of the silverswords and relatives provided the first ancewithin the last 4–6 Myr (Baldwin and Robichaux, 1995; es
strong refutation of hypotheses of minimal evolution or lack Baldwin and Sanderson, 1998) conforms to Carlquist’s t o
n
of representation of American plants in the Hawaiian archi- hypothesis of adaptive radiation of the group and contrasts 2
4
pelago (e.g. Skottsberg, 1931; Keck, 1936a; Fosberg, 1951). with the once-prevalent view of endemic Hawaiian N
o
Like Keck (1936a), Carlquist recognized that the true Compositae genera as ancient relicts that reflect an early ve
m
Hawaiian silverswords and greenswords (Argyroxiphium) Tertiary or even Mesozoic age of the Hawaiian chain (e.g. b
e
belong to the same insular lineage as the Hawaiian na‘ene‘e Guppy, 1906; Brown, 1921, 1922). Production of vigorous r 2
or kupaoa group (Dubautia, including Railliardia), and not Hawaiian–Californianhybridsbetweenmembersofthesilver- 0
1
8
to a different tribe or subtribe of Compositae, as earlier sword alliance and tarweeds of the ‘Madia’ lineage (Baldwin
treated (e.g. by Gray, 1870; Bentham, 1873). Unlike Keck et al., 1991; Carr et al., 1996) and hybrids between members
(1936a), Carlquist (1957) recognized that the iliau of the two diploid Californian sublineages of the ‘Madia’
(Wilkesia), a bizarre rosette-plant from Kaua‘i, is not more lineage implicated in an allotetraploid ancestry of the silver-
closely related to the rosette-plants in Argyroxiphium than to sword alliance (Barrier et al., 1999) provided biological evi-
the trees, shrubs, mat-plants, cushion plants and vines in dence for a level of retained genetic similiarity between
Dubautia, and, most importantly, that the insular lineage rep- tarweeds and silverswords consistent with a limited time
resented by Argyroxiphium, Dubautia and Wilkesia (i.e. the frame for divergence from a common ancestor.
Hawaiian silversword alliance) belongs in subtribe Madiinae,
with the primarily Californian tarweeds, including
A Hawaiian offshoot of the Californian sanicles
Raillardella s.l. (Carlquist, 1958a, b, 1959).
Carlquist’s (1959) expansion of the tarweed subtribe The ecologically diverse Hawaiian sanicles (Sanicula,
Madiinae to include the Hawaiian silversword alliance and Umbelliferae; Fig. 3C) are another group judged to be of
856 Baldwin & Wagner — American origins of Hawaiian flora
‘obscure’ affinity by Fosberg (1948) that structural and V. langsdorffii and Hawaiian violets constituted a robust clade
molecular data have indicated are of western North to the exclusion of a Japanese sample of V. langsdorffii.
American origin. Wallace (1881) noted that Hawaiian ‘... Experimental hydrolysis of seeds of V. langsdorffii and other
Sanicula (is) allied to Oregon species’ (p. 302) and violets to simulate conditions in a bird digestive tract demon-
Hillebrand (1888) similarly indicated that the sole Hawaiian strated that violets could be avian dispersed from the
species of Sanicula recognized at the time, S. sandwicensis, American Arctic to the Hawaiian Islands without harm via
‘...comes near S. menziesii [¼ S. crassicaulis]...from internal transport. They also inferred a polyploid ancestry of
California and Oregon’ (p. 144). Froebe (1971) suggested Hawaiian violets based on chromosome counts consistent with
that the Hawaiian species (sect. Sandwicenses) descended hexaploidy, decaploidy and dodecaploidy of different popu-
from two distinct lineages of the mostly Californian sect. lations of V. langsdorffii and with octoploidy of Hawaiian
Sanicoria (Fig. 3D), based in part on presence of a taproot violets.
D
in both groups. Subsequent molecular phylogenetic analyses In addition to providing the first compelling evidence of a o
w
indicated that the Hawaiian taxa constitute a monophyletic boreal origin for a diverse lineage of Hawaiian plants, n
lo
(rather than biphyletic) group that is nested within sect. BallardandSytsma(2000)pointedtoevidentnicheconserva- a
d
Sanicoria (Vargas et al., 1998). Three Californian or mostly tism of some Hawaiian violets to underline the importance of ed
Californian species share a large (24-base pair) deletion in ecologicalopportunityincolonizationoftheHawaiianIslands fro
m
the nuclear rDNA internal transcribed spacer 2 (ITS-2) with by plants from higher latitudes. They noted that lack of well- h
the Hawaiian species; all of those taxa that share the deletion documented representation of boreal plant lineages in the ttp
s
and their North American sister-group have prickly fruits that Hawaiian flora ran counter to the frequent opportunities for ://a
adhere readilytofurorfeathersandwerethereforeinferredto such dispersal, given the wide diversity of Arctic- and c
a
havereachedtheHawaiianIslandsbyexternalattachmenttoa sub-Arctic-breeding birds that migrate regularly to the de
m
bird(Vargasetal.,1998),assuggestedearlierforthetarweed islands, and suggested that transport to a suitable habitat ic
ancestor of the silversword alliance. Unlike silverswords, within the tropical island setting may be a more important .o
u
Hawaiian sanicles remain herbaceous and descended from limitation for establishment of northern plants than crossing p.c
diploid (not polyploid) continental ancestors, with diversifica- the oceanic expanse between the islands and boreal regions. om
tion from a common ancestor less than 1 Mya based on an That argument is consistent with earlier ideas expressed by /a
o
externally calibrated, rate-constant ITS tree. Wallace (1881), Setchell (1935) and Carlquist (1974), who b/a
concluded that transport of plants to islands may be less of a rtic
limiting factor in long-distance dispersal than ecological con- le
Violets and boreal American plant diversification -a
siderations upon arrival. Ballard and Sytsma also noted that b
in the Hawaiian Islands insufficient rigorous evidence about origins of other stra
c
BallardandSytsma(2000)providedmolecularevidencefor Hawaiian plants of possible northern ancestry may have t/1
a boreal, western North American origin of the Hawaiian resulted in underestimation of the importance of boreal 05
violets (Viola, Violaceae; Fig. 4A), another group regarded lineages in the Hawaiian flora. /6/8
by Fosberg (1948) to be of ‘obscure’ extra-Hawaiian affinity. 49
Prevalence of woodiness and multi-flowered inflorescences /93
Hawaiian mints and American hedge nettles: another North 7
among Hawaiian violets led Skottsberg (1940) to suggest 1
that the Hawaiian members of Viola represent the most American link to a major Hawaiian radiation 4 b
y
ancient lineage in this highly diverse, worldwide genus of Lindqvist and Albert (2002) demonstrated that the highly g
u
mostly herbaceous, single-flowered taxa. Skottsberg (1940) diverse Hawaiian mints (approx. 59 species in Haplostachys, es
andothers(e.g.St.John,1989;Wagneretal.,1990)suggested Phyllostegia and Stenogyne, Lamiaceae; Fig. 4C), previously t o
n
a closest relationship of Hawaiian violets to Latin American regarded as Austral or south-east Asian in origin (e.g. 2
4
woody violets of sect. Leptidium or sect. Rubellium, although Campbell, 1919; Fosberg, 1948), are phylogenetically nested N
o
St. John (1989) regarded Hawaiian violets as polyphyletic, within Stachys (‘hedge nettles’; Fig. 4D) and most closely ve
m
with separate introductions accounting for woody and herba- related to temperate North American members of the nearly b
e
ceous taxa in the islands. Ballard and Sytsma (2000) instead worldwide genus. Fleshy fruits of two of the three Hawaiian r 2
found that the Hawaiian violets constitute a young clade mintgenera(PhyllostegiaandStenogyne)wereinpartrespon- 0
1
8
nested among lineages of boreal, mostly North American her- sible for their earlier association with other fleshy-fruited,
baceous taxa of boggy or otherwise wet habitats, like those mostly south-east Asian members of Lamioideae (Prasium,
where some of the Hawaiian species occur. As in the silver- BostrychantheraandGomphostemma)intribePrasieaeorsub-
swordalliance,moleculardataweredecisiveindemonstrating family Prasioideae (e.g. Bentham, 1832–1836; Briquet,
that woodiness was derived in the Hawaiian setting from an 1895–1897; Wunderlich, 1967), rather than with dry-fruited
ancestrally herbaceous condition. lamioids. Pericarp analysis led Ryding (1994) to conclude
More specifically, Ballard and Sytsma (2000) showed that instead that fleshy fruits evolved repeatedly in Lamioideae
Hawaiian violets probably descended from an Arctic or far- and that the Hawaiian mints, although probably monophyletic
north temperate American ancestor referable to Viola langs- (including the dry-fruited Haplostachys), are not as closely
dorffii s.l. (‘Alaskan violet’; Fig. 4B), which occurs from related to Prasium or Gomphostemma as earlier suggested, in
northern California to Alaska in western North America and accordance with the subsequent molecular data.
Kamchatka in the Russian Far East to northern Japan. They In addition to representing an exceptionally diverse
found that the sampled boreal American representative of Hawaiian lineage of vascular plants in number of recognized
Baldwin & Wagner — American origins of Hawaiian flora 857
A B
D
o
w
n
lo
a
d
e
d
fro
m
h
ttp
s
://a
c
a
d
e
m
ic
.o
u
p
.c
o
m
/a
o
b
/a
rtic
C le
-a
b
s
tra
c
D t/1
0
5
/6
/8
4
9
/9
3
7
1
4
b
y
g
u
e
s
t o
n
2
4
N
o
v
e
m
b
e
r 2
0
1
8
FIG. 4. ExemplarsofHawaiianangiospermradiations(left)andAmericanrelatives(right)inViolaceae(top)andLabiatae(bottom).(A)Violachamissoniana
subsp. robusta. Image courtesy of G. D. Carr. (B) V. langsdorffii. Image courtesy of M. Goff. (C) Phyllostegia floribunda. Image courtesy of J. K. Obata.
(D)Stachyschamissonis.ImagecourtesyofA.Brousseau,copyright#1995SaintMary’sCollegeofCalifornia.
species, Hawaiian mints have radiated across highly contrast- American Stachys; among the closest relatives of Hawaiian
ing environmental settings, including dry shrublands and wet mints in North American Stachys based on the molecular
forests, and exhibit floral diversity associated with both trees are the insect-pollinated S. quercetorum and the bird-
insect pollination (in Haplostachys and Phyllostegia) and pollinated S. chamissonis (Lindqvist and Albert, 2002).
birdpollination(inStenogyne).Thepresenceofsuchdiversity Polyploidy, possibly of hybrid origin, was inferred by
in pollination systems in the Hawaiian mints may reflect evo- Lindqvist and Albert (2002) to have been ancestral within
lutionarylabilitypresentintheircommonancestorwithNorth the American founder lineage that gave rise to Hawaiian
858 Baldwin & Wagner — American origins of Hawaiian flora
mints (2n¼64, 66) and suggested it to possibly account for sevenrecognizedwoodyspeciesthatbelongtotwolineagesof
the magnitude of their morphological diversification (see sufficiently different habit, ecology and floral morphology to
also Lindqvist et al., 2003), as proposed earlier for the silver- have been thought to represent descendants of distinct
sword alliance (Barrier et al., 1999). Hawaiian founders (see Wagner et al., 1990). One of these
two groups contains subshrubs of often mesic habitats with
white, notched petals (e.g. S. perlmanii; Fig. 5C); the other
Hawaiian lineages of the carnation family group comprises shrubs of dry habitats, with narrower leaves
and green or cream-coloured to reddish, divided petals (e.g.
Schiedea. The most diverse lineage of Hawaiian
S. hawaiiensis). Skottsberg (1941) included Hawaiian Silene
Caryophyllaceae also appears to have probably descended
in a group of temperate genera (including Rubus, Sanicula
from a North American founder, based on recent molecular
and Viola) that he regarded as being ‘...so distinct and so
results (Wagner et al., 2005b; Harbaugh et al., 2010). The
D
endemic Hawaiian genus Schiedea (Fig. 5A) consists of 34 distant from everything else that their history here (in the ow
species of herbs, shrubs and vines that arguably represent the Hawaiian Islands) must have been a long one’ (p. 697). nlo
Fosberg (1948) regarded all members of Hawaiian Silene as a
greatest diversification of reproductive traits in angiosperms d
of the islands, with lineages that are monomorphic and of ‘obscure’ affinity. Chowdhuri (1957) classified Hawaiian ed
insect- or bird-pollinated (sometimes autogamous) or are members of the genus in a common section with two poorly fro
m
dimorphic and wind-pollinated. A bird-pollinated lineage known, putatively Japanese species, S. japonica and h
within Schiedea is so distinctive morphologically that it was S. tanakae, possibly implying an East Asian origin of the ttp
s
until recently (Wagner et al., 2005b) treated as a distinct Hawaiian taxa. ://a
genus, Alsinidendron (Fig. 5A). Based on molecular phylogenetic data from cpDNA and ca
Discerning relationships of Schiedea to other members of nuclear DNA (ITS) sequences, Eggens et al. (2007) found de
m
subfamily Alsinoideae outside the Hawaiian Islands has been that the two morphological groups of Hawaiian Silene consti- ic
tute a clade that is sister to S. antirrhina (Fig. 5D), a wide- .o
complicated by extensive morphological divergence of the u
group from all living representatives of the subfamily. spread, weedy annual of temperate North America and South p.c
Skottsberg (1939) hesitantly suggested a boreal ancestry of America, with minimal molecular divergence between the om
American and Hawaiian taxa. Some uncertainty remains /a
Schiedea (including Alsinidendron) before concluding that o
the group is instead a southern (‘Austral’) element in the about the geographical affinities of Hawaiian campions b/a
Hawaiian flora, derived ultimately from Antarctic ancestors because of the occurrence of S. antirrhina on both American rtic
continents and because of a sister-group relationship between le
(Skottsberg, 1941), as he concluded also for the the Hawaiian+American clade and a diverse Silene clade -ab
silversword-alliance genera (see above). Fosberg (1948) s
treated Schiedea as of ‘obscure’ phytogeographical affinity. from Eurasia (specifically, the Mediterranean and Middle tra
c
Extensive molecular phylogenetic sampling across East), which conceivably could have been the source area t/1
(A2l0s1in0o)ildeedaetobthyeWdiascgonveerryetthaalt.S(c2h0i0e5dbea) iasnsdistHerartboatuwgohmeotnaol-. fHoarwaiinidanepIesnladnednst. PdriesvpaelresnacleeovfehnetsrbatcoeouAsmneesrsician thaendconthtie- 05/6/8
typic genera, Honckenya and Wilhelmsia (Fig. 5B), that are nental lineages that are closely related to Hawaiian campions 49
boreal, rather than austral, endemics. Although both boreal indicates that woodiness of the Hawaiian species probably /93
7
evolved in the insular setting, as inferred for Schiedea, the 1
genera occur in Eurasia as well as North America, the 4
silversword alliance and some lineages of Hawaiian mints b
founder of Schiedea appears most likely to have dispersed y
and Viola. g
fromNorthAmerica,wherethreeofthefourtaxa(subspecies) u
e
ocofnHtionnecnkteanlyarealraetirveesstricotfedHanodncwkehneyreathaenmdosWtcillhoeslemlysiraelataerde WoodyspurgesfromNorthAmerica.Thespectaculardiversityof st on
endemic (Harbaugh et al., 2010). Based on the phylogenetic trees, shrubs and subshrubs in Hawaiian Euphorbia subg. 24
findings, woodiness in Schiedea appears to be secondarily Chamaesyce (16+ species; Fig. 6A) has been long suggested No
derived from an herbaceous habit, which is common to to represent an endemic lineage of Indo-Pacific origin ve
m
Honckenya, Wilhelmsia and their closest relatives (see (Degener and Croizat, 1936; Fosberg, 1948; Koutnik, 1987). b
e
Carlquist, 1995). Wagner et al. (2005b) noted that successful Carlquist (1970, 1974, 1980) regarded the Hawaiian taxa as r 2
colonization of the Hawaiian Islands by the founder of exemplary of adaptive radiation, based on anatomical, mor- 01
8
Schiedea was not complicated by dioecy and the associated phological and ecological considerations. Studies by Pearcy
need for dispersal of multiple individuals from the mainland; and colleagues (e.g. Robichaux and Pearcy, 1980; Pearcy
floralhermaphroditismisunequivocallyestimatedasancestral et al., 1982; Pearcy and Fransceschi, 1986) demonstrated
inSchiedea,based onoccurrence ofthatconditionthroughout major ecophysiological differences among the ancestrally
the closest continental relatives except Honckenya, which and uniformly C4 Hawaiian chamaesyces that correspond to
appears to have independently evolved a dimorphic sexual the wide range of habitats occupied by different taxa, such
system. as exposed, dry beaches and shaded, wet forests. Morden
and Gregoritza (2005) noted levels of molecular variation
Silene. The other endemic lineage of Hawaiian andlineagedivergencewithintheHawaiiancladethatareunu-
Caryophyllaceae was recently resolved as the sister group to sually high for an oceanic island group and that warrant
a temperate American species (Eggens et al., 2007). The species-level treatment for at least some of the ten endemic
mostlyNorthernHemispheregenusSilene(campionsorcatch- varieties that have been recognized recently (Koutnik, 1987,
flies),withapprox.650species,isrepresentedintheislandsby 1999).