A taxonomic synopsis of Altingiaceae with nine new combinations

Abstract A taxonomic synopsis of the Altingiaceae is presented, including the taxonomic enumeration and distribution of 15 recognized species based on studies of 1,500 specimens from 24 herbaria throughout the distributional range of the taxa. Previous phylogenetic analyses based on several molecular markers have shown that Altingia and Semiliquidambar are nested within Liquidambar. All Altingia and Semiliquidambar species are now formally transferred to Liquidambar, which has the nomenclatural priority. The following nine new combinations are herein made: Liquidambar cambodiana(Lecomte) Ickert-Bond & J. Wen, Liquidambar caudata (H. T. Chang) Ickert-Bond & J. Wen, Liquidambar chingii (Metcalf) Ickert-Bond & J. Wen, Liquidambar gracilipes (Hemsl.) Ickert-Bond & J. Wen, Liquidambar multinervis(Cheng) Ickert-Bond & J. Wen, Liquidambar obovata (Merrill & Chun) Ickert-Bond & J. Wen, Liquidambar poilanei (Tardieu) Ickert-Bond & J. Wen, Liquidambar siamensis (Craib) Ickert-Bond & J. Wen, and Liquidambar yunnanensis (Rehder & Wilson) Ickert-Bond & J. Wen.


Introduction
Th e Altingiaceae (the sweet-gum family) are a small family of trees that have been traditionally classifi ed into members with a predominantly temperate distribution ( Liquidambar L.) and those with a largely tropical to subtropical distribution ( Altingia Noronha, Semiliquidambar H. T. Chang). Th e family is valued worldwide for its timber and fragrant resin (styrax) and is locally highly prized for the roots and bark used in traditional Chinese medicine (Vink 1957;Zhang et al. 2003;Ickert-Bond et al. 2007). Most noteworthy to biologists, Altingiaceae show a fascinating intercontinental disjunction in temperate regions of North America, W Asia and some higher elevational montane areas in subtropical Asia and Mexico (Ickert-Bond and Wen 2006). Furthermore, deep molecular divergence coupled with a high level of morphological similarity suggests a conserved morphology of some species, i.e., morphological stasis, an evolutionary phenomenon that has been proposed for many animal groups as well as some plants (Ickert-Bond and Wen 2006). Other members of the Altingiaceae exhibit morphological divergence in response to habitat diversity in the subtropics of eastern Asia. One member of the sweet-gum family, Semiliquidambar, has puzzled scientists since its discovery in the 1960s, due to its rarity and morphological intermediacy between the other two genera (Ferguson 1989).
Generally, the Altingiaceae were considered closely related to the Hamamelidaceae (see below). Th e family is recognized by the solitary capitate woody infructescences with many bicarpellate fruits, and male infl orescences in heads aggregating into racemes. Other systematic characters that diff erentiate the two families are less well known and understood. It is now generally agreed that the Altingiaceae are a distinct family ; for detailed comparisons see our other contributions (APG II 2003;Pigg et al. 2004;Ickert-Bond et al. 2005Ickert-Bond and Wen 2006;APG III 2009).
Th e family name Altingiaceae is based on Altingia, fi rst named by Noronha (1790) in honor of the former General Governor Alting of the East Indian colonies of the Netherlands (Hayne 1830). Th e family was formally designated in 1843 by Horaninow (Hoogland and Reveal 2005). Th e scientifi c name for Liquidambar L. is a combination of the Latin and Arabic words Liquidus and Amber meaning fragrant liquid or balsam (Yaltrik and Efe 2000). Most authors have recognized Altingia and Liquidambar to be members of the Hamamelidaceae s.l., most often at the subfamily level, while Endlicher in his Genera Plantarum (1840) segregated the Altingiaceae [s. Balsamifl uae] from the Hamamelidaceae s. str and placed Altingiaceae in Julifl orae , between Platanaceae and Salicaceae, while the Hamamelidaceae were placed in Discanthae , between Loranthaceae and Bruniaceae. Subfamily Altingioideae was recognized by J. Williams in his revision of Balfour's Manual of Botany in 1855. Bentham and Hooker (1883) treated Liquidambar and Altingia as distinct genera alongside other typical genera of Hamamelidaceae, without mention of subfamilies, but they recognized two categories (Abteilungen): (1) with the ovary containing 2-many ovaries, and (2) with the ovary containing a single ovary. Reinsch (1890) retained the traditional one family concept, but departed from the general consensus by splitting the Hamamelidaceae based on morphological and anatomical characters, that he considered to be more fundamental than the fruit, into three subfamilies: (1) Altingioideae including Altingia and Liquidambar and (2) Bucklandioideae including Exbucklandia R.W. Brown and Rhodoleia Champ. ex Hook., and (3) Hamamelidoideae including Corylopsis Siebold & Zucc. , Dicoryphe Th ouars , Distylium Siebold & Zucc., Eustigma Gardner & Champ. , Fothergilla L. , Hamamelis L. , Loropetalum R. Br. , Parrotia C.A. Mey. , Sycopsis Oliv. and Trichocladus Pers. , ( Semiliquidambar was not known at the time).
Baillon (1871) explicitly excluded Liquidambar and Altingia from the Hamamelidaceae, contrary to Bentham and Hooker, who included these two genera in the family. Baillon placed them in an intermediate position between Hamamelidaceae and Platanaceae. Traditionally, the Hamamelidaceae s.l. (including Altingiaceae) have been considered as a member of the Hamamelididae Takht. (Cronquist 1981;Takhtajan 1997). Recent molecular studies have shown this assemblage to be polyphyletic and support Altingiaceae and Hamamelidaceae s. str. as members of the saxifragoid clade within a larger rosid clade (Chase et al. 1993;Magallón et al. 1999;Soltis et al. 2000;APG II 2003;Fishbein and Soltis 2004;Soltis et al. 2007;APG III 2009;Soltis et al. 2011). Furthermore, relationships among some Saxifragales, and the remaining families (Altingiaceae, Cercidiphyllaceae, Daphniphyllaceae, Hamamelidaceae, and Paeoniaceae) also remain unclear (Feng et al. 1998;Qiu et al . 1998;Hoot et al. 1999;Savolainen et al . 2000;Fishbein et al. 2001;Soltis et al. 2000). In maximum likelihood analyses of a fi ve-gene data set, Fishbein et al. (2001) recovered an optimal topology with Daphniphyllaceae and Hamamelidaceae sister to the remaining members of the clade; however, the precise branching order of these two early-diverging members of Saxifragales was unclear. Following Daphniphyllaceae and Hamamelidaceae, Altingiaceae, Cercidiphyllaceae, and Paeoniaceae appeared as successive sisters to a core clade of Saxifragaceae, Haloragaceae and Crassulaceae. Fishbein et al . (2001) also showed that the poor resolution obtained in Saxifragales is not due to violations of assumptions or to combining data partitions having confl icting histories or processes. Rather, their analyses suggest instead that the initial diversifi cation of Saxifragales was indeed rapid. Within Saxifragales molecular phylogenetic results have rarely supported a sister relationship between Altingiaceae and Hamamelidaceae s. str. (e.g., Hoot et al. 1999;Fishbein et al. 2001, but see Fishbein and Soltis 2004). Most recently, based on over 50,000 bp Jian et al. (2008) have found strong support for a clade composed of the Paeoniaceae + woody clade (Cerdiphyllaceae, Daphnipyllaceae, and Hamelidaceae) Altingiaceae)) to be sister to the rest of the Saxifragales. Th e sister group relationship of Altingiaceae with Hamamelidaceae plus Cerdiphyllaceae and Daphniphyllaceae was also strongly supported in a supermatrix approach by Soltis et al. (2013).
Altingia and Liquidambar are each defi ned by several morphological characters and have been maintained as separate genera in modern taxonomic treatments (Vink 1957;Tardieu-Blot 1965;Zhang et al. 2003). Analyses based on several molecular markers suggest that Altingia is nested within Liquidambar (Shi et al. 1998;Shi et al. 2001;Ickert-Bond et al. 2005;Ickert-Bond and Wen 2006;Ickert-Bond et al. 2007, Wu et al. 2010) and that Semiliquidambar is of intergeneric hybrid origin between L. formosana -L. acalycina and A. obovata or A. chinensis . Yet our morphological analysis supports Altingia and Liquidambar as mutually exclusive sister clades (Ickert-Bond et al. 2005. Th e apparent incongruence of these phylogenies appears to be due to morphological convergence. Characters that distinguish Liquidambar from Altingia are related to an open wind pollination syndrome and may represent convergences to temperate habitats, particu-larly, the presence of anthers borne on long fi laments and the loss of stomium bifurcations would facilitate the wind dispersal of pollen (Huff ord and Endress 1989), while long narrow styles on exserted fruits (Fig. 2D) may aid in the capture of pollen on the broad stigmatic surfaces in open habitats of temperate Liquidambar . Furthermore, additional synapomorphies for Liquidambar may also represent adaptations for a temperate distribution. Th ese characters (elongate and tapered carpel shape, seeds with distal wings, and more tightly constructed infructescences) are related to seed rather than pollen dispersal. Several other families (e.g., Platanaceae) show a similar convergence among temperate members (Tiff ney 1986;Crane 1989).
Character-state changes in Altingia seem to correlate with tropical and subtropical environments in eastern Asia and Indochina, whereas changes in Liquidambar correlate with temperate sites, where the genus is found today. Of the eight characters defi ning Altingia ( Fig. 2A, B), four are reversals (characters 2-5: ratio of leaf length to width, leaf division, venation, and stipule size) (see fi g. 86 of Ickert-Bond et al. 2007). Th e availability of diverse habitats in tropical and subtropical eastern Asia and Indochina facilitated the diversifi cation of Altingia species in response to recent active uplifts of mountains in eastern Asia since the Tertiary (Morley 1999;Wen 1999Wen , 2001Ickert-Bond and Wen 2006).
To maintain the monophyly of the group in question (Potter and Freudenstein 2005), we place all taxa of Altingiaceae in Liquidambar (the earliest available name), and maintain the conserved name Altingiaceae for the family. Appropriate new combinations are provided below.

Methods
We evaluated all currently recognized taxa within Altingia , Liquidambar , and Semiliquidambar . Our study is based on: (1) fi eld observations from throughout the distributional range of the taxa, with fi eld visits to sites in Mexico (Veracruz), Vietnam, Cambodia, Indonesia and Guangdong, Hainan, Hong Kong, Hubei, Giangxi, and Zheijang provinces in China, and (2) the analysis of specimens in 24 herbaria (including available types): A, BK, BM, C, E, F, FI, FN, FUS, GH, HGAS, HN, IBSC, ISTO, K, LINN, LU, N, NFU, NY, P, PE, SYS, US. We previously provided detailed examination of the fruit anatomy and morphology (Pigg et al. 2004;Ickert-Bond et al. 2005;Ickert-Bond et al. 2006 and here provide an overview of some of the features that have been used to characterize the genera both for ovulate and staminate infrutescences. Measurements were made with an electronic caliper (Mitutoyo mod. CD-6"CS).
One genus and ca. 15 species: E, W, and SE Asia, Central, and North America. Pollen morphology appears uniform throughout the family with spherical, pentaporate grains (Fig. 3A, E, I), that show scabrae and irregularly shaped perforations on the tectum ( Fig. 3B-C, F-G, J-K) and a tectate-columellate exine (Fig. 3D, H, L; Ferguson 1989). Zavada and Dilcher (1989) found slight diff erences in the breadth of the columellae between Liquidambar styracifl ua and Altingia obovata based on TEM imagery. Our analysis of the exine using freeze-fracturing and SEM shows slight difference of this characters between Liquidambar excelsa (= Altingia excelsa , Fig. 3D)   and L. styracifl ua (Fig. 3L), but the columellae in Liquidambar gracilipes (= A . gracilipes , Fig. 3H) appear to be of equal width to those in the ones examined from L. styracifl ua (Fig. 3L).
Key to the species of Liquidambar s.l.*

1
Leaves consistently palmately-lobed 2 Leaves of mature trees with three lobes, sometimes 5 in juvenile condition 3 Infructescence subglobose, with stout styles 4-6 mm long, curved; seeds with circular fl ange . Type. Liquidambar styracifl ua L.    Note. A few specimens from Hunan and Guangdong exhibit pubescent stems, leaf surfaces and lobing attributable to young plants, previously annotated by Merrill (1934) as " L. formosana var.  Distribution. China (Fujian, Guangdong, Hong Kong, and Zhejiang).  ( Note. Specimens of L. multinervis show long-petioled papery leaves, with 10 lateral veins, and serrate margin, closely resembling L. siamensis or L. excelsa . Broken infructescences of the specimen at N suggest few fruits per infructescence, thus underscoring the closeness to L. siamensis, while Cheng (1947) described its close affi nity with L. yunnanensis . Th is species is poorly understood, we have only seen the type collection that consists of a branch with multiple leaves and a crushed infructescence. Th e distributional discontinuity of L. multinervis in N Guizhou from those of L. siamensis and L. excelsa much further south may warrant specifi c status, but more material is needed. N Guizhou is not well explored botanically and with more exploration of this area, we might be able to better characterize this taxon in the future. Distribution. China, N Guizhou. ( -1 sheet). When comparing material from Turkey (ISTO), the specimens collected by Aksoy 5202 (2 sheets) have margins that are sometime lobed beyond the typical 5 -lobing, also observed in Aksoy 5201 (ISTO-1 sheet), and Aksoy 5203 (ISTO -4 sheets), while specimens identifi ed as L. orientalis var. integriloba lack such lobing, as seen in Aksoy 5204 (ISTO-3 sheets) also from Turkey. Th is specimen ( Murray 1020 -GH, MO) lacks the typical lobing of var. orientalis and could thus be considered var.  Boratynska et al. 15 (K); between Malona and Archangelos, old trees along small stream, very frequent, K. Boratynska 164 (K); Salakos, hedges near stream, Davis 40317 (K); SE of Salakos, along stream, below orchard, K. Boratynska 15 (K).