Congeneric species that have evolved on oceanic islands rarely grow with one another. This spatial sorting is thought to be the result of niche pre-emption, where the first species to occupy a given habitat would exclude its relatives through competition. The evidence that competition shapes local species distributions is scant. I propose that hybridization thwarts the invasion of congeners. We know that crossing barriers between products of recent adaptive radiations typically are weak. Colonization of occupied patches would be sporadic, because invaders would be moving from habitats where they are adapted to those where they are not. Nearly all progeny of the scant invaders would be ill-adapted hybrids, because the pollen pool would consist almost exclusively of indigenous pollen. Only if aliens were predominantly inbreeders or apomicts would they have a chance of escaping the deluge of local pollen and becoming established in the occupied patch. The spatial reach of hybridization is much greater than that of competition.

Here I present a cladistically based monograph of a group of closely related mosses known as the Syrrhopodon involutus complex. Monographs remain a critical and indispensable component of systematics, but traditional monographic practices require some revisions if modern monographs are to truly reflect and incorporate phylogenetic understanding. I summarize the methods used to address issues relating to sampling and the formation of OTUs for this monograph, and emphasize the importance of incorporating morphological character data in a total evidence phylogenetic analysis. In addition to a traditional Linnaean taxonomic treatment, I provide rank-free definitions of the terminal taxa in the S. involutus complex. Following the PhyloCode recommendations for the designation and naming of more inclusive taxa, I present a novel method for naming terminal taxa as least-inclusive clades. This monographic treatment represents a first attempt at applying the PhyloCode at the terminal level in any plant group, and recognizes five distinct taxa in the S. involutus complex, which I have designated with the rank-free clade name Leucophanella: banksii, borneensis, involutus, rufescens, and revolutus.

Evidence for a southern South American lineage of polygrammoid ferns was inferred from analyses using four cpDNA markers. The lineage corresponds to the genus Synammia, which has usually been treated as a synonym of the polyphyletic genus Polypodium. Seven specimens of two of the three recognized species were sampled to infer the phylogenetic relationships of this lineage to other polygrammoid ferns and the diversity within this lineage. All approaches found Synammia clearly separated from other polygrammoid ferns, but the exact relationships are still unclear. Model-based methods recovered Synammia as sister to a mainly Neotropical clade of polygrammoid ferns, but support for this hypothesis was provided only by Bayesian inference of phylogeny. The deletion of nearly the whole rps4-trnS IGS is a synapomorphy of Synammia. The lineage is separated from other extant polygrammoid ferns, but very low to zero sequence diversity was found among the seven samples of Synammia. The results strongly support an independent lineage of polygrammoid ferns in southern South America.

The monophyletic Eurasian clade of Amaryllidaceae was analyzed using plastid ndhF and rDNA ITS sequences for 33 and 29 taxa, respectively; all genera were represented by at least one species. Both maximum parsimony and Bayesian analysis were used on each data set and the combined data. Both sequence matrices resolve the Central and East Asian tribe Lycorideae as sister to the Mediterranean-centered genera of the clade, and recognize two large subclades within the greater Mediterranean region: Galantheae, consisting of Acis, Galanthus and Leucojum; and Narcisseae (Narcissus and Sternbergia)/Pancratium. However, there are areas of incongruence between the ndhF and ITS trees. When three predominantly monotypic genera, Hannonia, Lapiedra, and Vagaria, centered in North Africa, are removed from the alignments, the two sequence matrices produce fully congruent topologies with increased support at many of the nodes, with ILD between partitions rising from P = 0.07 to 0.96. We hypothesize that lineage sorting took place after the divergence of Galantheae and Narcisseae/Pancratium from a common genepool with Hannonia, Lapiedra, and Vagaria retaining a mosaic of the ancestral haplotypes. We also performed dispersal-vicariance analysis to reconstruct biogeographic scenarios on several of the generic level phylogenies found with and without these three genera included, as well as on a species-level phylogeny of Galantheae. After the vicariant divergence of the Asian Lycorideae, North Africa and the Iberian Peninsula are the most likely areas of origin for the rest of the clade. The results of the dispersal-vicariance analysis are discussed in the context of the complex biogeographic history of the Mediterranean basin.

The systematic placement of the little-known species Ptychosperma micranthum (Arecaceae/Palmae: Arecoideae: Areceae) from far western New Guinea has been repeatedly disputed, resulting in recombinations in both Heterospathe and Rhopaloblaste. However, comparative morphological and molecular phylogenetic studies provide strong evidence against the placement of the species within any of these three genera, or indeed in any other accepted genus. Thus, a new genus, Dransfieldia, is herein described and a new combination, Dransfieldia micrantha, is made. Morphological character analyses demonstrate that the combination of character states that defines Dransfieldia is highly distinctive, despite the fact that many of the states are homoplasious. Dransfieldia micrantha is a slender, unarmed palm possessing a well-defined crownshaft, strongly ridged leaflets with entire, acute apices, an infrafoliar inflorescence with a persistent prophyll that is split apically by the exertion of the peduncular bract, a peduncle that is longer than the rachis, bullet-shaped multistaminate male flower buds in which the filaments in the outer whorl are irregularly inflexed in bud, and fruit with apical stigmatic remains. Molecular phylogenies strongly support the position of Dransfieldia within tribe Areceae, placing it within a clade of genera from the western Pacific.

Carex subg. Vignea is characterized by sessile bisexual spikes, distigmatic flowers, and the lack of cladoprophylls. Phylogenies reconstructed using nrDNA internal and external transcribed spacer (ITS and ETS 1f) sequences for 100 vignean taxa support this subgenus as monophyletic. The atypical C. gibba is sister to all remaining taxa. Many clades in the remainder of the subgenus do not correspond to easily defined morphological groups, with species representative of several disparate sections frequently contained within a single clade. Many traditionally recognized sections are not supported, although others such as sects. Ovales, Stellulatae, and Glareosae are monophyletic. Tree topologies indicate that gynaecandry has evolved multiple times in subg. Vignea. Species of uncertain subgeneric affinity are variously placed in our analysis. Carex fecunda, previously linked to subg. Vignea, is positioned within the outgroup composed of species traditionally placed in subg. Carex and Vigneastra. While species with highly compound inflorescences are often regarded as ancestral in Carex, our results indicate that this inflorescence type has evolved multiple times and is derived within subg. Vignea.

During a revision of Andean Poa, three new species, one from Colombia and two from Peru, were recognised as new and are described here. Poa soderstromii from the Páramo of Chisacá, Colombia, is similar to the Peruvian Poa brevis from which it differs by having the glumes nearly as long as the spikelet and the lemmas conspicuously 5-nerved and scabrous on the keel. Poa oscariana, from Cusco, is similar to P. gilgiana and P. huancavelicae; from both it can be separated by the spikelets broadly elliptic in outline, 4.5–5.5 mm long and dull pale green, and glumes and lemmas asperulous, without hyaline margins. Poa pilgeri from Junín, Huancayo, resembles P. candamoana, but differs by its filiform and convolute leaves and small spikelets. The species are described, illustrated and compared with similar species. A key as well as commentaries on the subgeneric affinities of the new species are included.

A phylogenetic analysis of molecular (ITS, trnL-F, trnK including the matK coding region) and morphological data is presented for the pantropical monocot family Costaceae (Zingiberales), including 65 Costaceae taxa and two species of the outgroup genus Siphonochilus (Zingiberaceae). Taxon sampling included all four currently described genera in order to test the monophyly of previously proposed taxonomic groups. Sampling was further designed to encompass geographical and morphological diversity of the family to identify trends in biogeographic patterns and morphological character evolution. Phylogenetic analysis of the combined data reveals three major clades with discrete biogeographic distribution: (1) South American, (2) Asian, and (3) African-neotropical. The nominal genus Costus is not monophyletic and its species are found in all three major clades. The Melanesian genus Tapeinochilos is monophyletic and included within the Asian clade. Monocostus and Dimerocostus are sister taxa and form part of the South American clade. The African-neotropical clade is composed entirely of the genus Costus; moreover, there is support for previously recognized subgeneric groupings within the Costus clade. Evolutionary trends in floral morphology show that close associations with pollinators have evolved several times from an ancestral generalist pollinator floral form. Bee pollination has evolved once in the family, arising in Africa from an open-flowered (generalist) ancestor. Bird pollination has evolved multiple times: once from an open-flowered ancestor in Southeast Asia and multiple times from a bee-pollinated ancestor in the neotropics. Additional morphological characters not traditionally used to define taxonomic groups, but having high consistency in the current phylogenetic analysis, are discussed.

Evidence is presented for phylogenetic relationships in pantropical Melastomataceae, subfamily Olisbeoideae based on combined exon and intron sequences of the nuclear glyceraldehyde 3-phosphate dehydrogenase gene. Parsimony and maximum-likelihood analyses yielded a well-supported ingroup tree consistent with earlier morphologically based concepts of six genera—Memecylon, Mouriri, Votomita, Spathandra, Lijndenia, and Warneckea. The position of the root node in Olisbeoideae remains equivocal. Outgroup-rooted maximum parsimony suggests a deep divergence between Old and New World lineages, while the maximum-likelihood rooting resolved paleotropical genera as a paraphyletic grade basal to the neotropical taxa. The Fitch optimization method for estimating character evolution consistently inferred the strongly acrodromous leaf venation pattern as ancestral in Olisbeoideae, reinforcing the conclusion that the superficially uninervate or brochidodromiform venation pattern of Memecylon, the neotropical subclade, and some Lijndenia and Warneckea species is best interpreted as a series of independent losses of the strongly acrodromous condition. Genomic GapC sequences may have phylogenetic utility at intergeneric level in other angiosperm families, particularly those that have low apparent rates of chloroplast DNA sequence evolution.

A phylogenetic analysis of the Celastrales was performed using nuclear (18S, ITS 1, 26S rDNA) and plastid (atpB, matK, rbcL, trnL-F spacer) genes. In contrast to most previous studies, Celastrales and Malpighiales are resolved as being more closely related to one another than either are to Oxalidales. The Huaceae are well supported as the sister group to Oxalidales, not Celastrales, as had been previously proposed. The Lepidobotryaceae are unambiguously supported as sister to the clade consisting of Celastraceae and Parnassiaceae. The Parnassiaceae are well supported as members of an early branching lineage within Celastraceae, rather than as its sister group. Likewise, Pottingeria appears to be part of an early derived lineage of Celastraceae. Empleuridium is unambiguously supported as a derived member of Celastraceae, as are Brexia, Canotia, Siphonodon, Stackhousia, and Tripterococcus. Bhesa is unambiguously supported as a member of Malpighiales, though its relationships within the order remain unclear. Perrottetia should be transferred out of Celastraceae and into eurosids II, being closely related to Tapiscia and Dipentodon. These results help delimit the Celastrales and Celastraceae as morphologically more homogeneous taxa.

The phylogenetic relationships and chromosomal evolution of the diverse tropical genus Passiflora (Passifloraceae) are explored using data from two chloroplast markers: the rpoC1 intron and the trnL/trnT spacer region. A survey of the presence or absence of the rpoC1 intron in 136 species representing 17 of Killip's (1938) 22 subgenera of Passiflora and four other genera in the Passifloraceae revealed intron losses in 46 taxa. A minimum of two losses were confirmed by a parametric bootstrap approach on sequence data from the trnL/trnT chloroplast non-coding region for 61 taxa. The results of phylogenetic analyses of the trnL/trnT sequence data support the reduction of Killip's 22 subgenera to four as proposed in a new classification system by Feuillet and MacDougal (2004). The monophyly of the ‘n=6’ and ‘n=9’ chromosomal and morphological groups is strongly supported. In addition, these data indicate that Passiflora biflora, or closely related species, is the likely continental sister to the red-flowered Caribbean taxa, while P. auriculata is weakly supported as the New World sister to the Old World Passifloras. Finally, character optimization of chromosome numbers on the phylogenetic tree supports x=12 as the base chromosome number for Passiflora.

The genus Leavenworthia (Brassicaceae) has long been a focus of research on mating system evolution, due to the presence of both self-incompatible and self-compatible species within the genus. A phylogenetic hypothesis invoking multiple transitions between mating systems has been proposed for Leavenworthia, but this hypothesis has not been subject to molecular phylogenetic analysis. DNA sequence variation from four non-coding chloroplast regions (the trnL intron; and the trnT-trnL, trnL-trnF, and psbA-trnH intergenic spacers) was used to reconstruct the generic phylogeny, to test the validity of several proposed species, and to assess the number of mating system transitions within the genus. The strict consensus tree largely reflected the long-standing phylogenetic hypothesis for Leavenworthia, although additional data are needed to fully validate the recognition of L. crassa and L. alabamica. Unexpected results included the placement of L. uniflora as sister to the rest of the genus, and the apparent hybridization between L. exigua and L. torulosa. Finally, our data strongly supported a minimum of three mating system transitions within Leavenworthia.

A parsimony analysis of ndhF nucleotide sequences representing 24 species and 13 genera of Sterculioideae strongly supports (100% bootstrap) the monophyly of the group. Within the Sterculioideae clade four major clades are recognized with good bootstrap support but relationships among them are not resolved. This analysis suggests the recognition of Argyrodendron as separate from Heritiera, supports Acropogon as separate from Sterculia, and Tarrietia as part of Heritiera. The current circumscriptions of Hildegardia and Firmiana are not supported. The use of fruit characters in the delimitation of genera within Sterculioideae may not be appropriate in some cases and other morphological characters need to be found.

Balsaminaceae are a morphologically diverse family with ca. 1,000 representatives that are mainly distributed in the Old World tropics and subtropics. To understand the relationships of its members, we obtained chloroplast atpB-rbcL sequences from 86 species of Balsaminaceae and five outgroups. Phylogenetic reconstructions using parsimony and Bayesian approaches provide a well-resolved phylogeny in which the sister group relationship between Impatiens and Hydrocera is confirmed. The overall topology of Impatiens is strongly supported and is geographically structured. Impatiens likely originated in South China from which it colonized the adjacent regions and afterwards dispersed into North America, Africa, India, the Southeast Asian peninsula, and the Himalayan region.

Phylogenetic analysis of 74 species of Symplocos was conducted with DNA sequence data from the chloroplast trnC-trnD intergenic region. Maximum parsimony and Bayesian inference trees are consistent with those of a previous study based on combined data from the nuclear ribosomal ITS region and the chloroplast regions rpl16, matK, and trnL-trnF. The inclusion of 21 phylogenetically informative indel characters from trnC-trnD resulted in greater maximum parsimony resolution and clade support than with these characters excluded. A combined five-gene-region (trnC-trnD, ITS, rpl16, matK, and trnL-trnF) analysis based on a data set with complete sequence data is almost completely concordant with that of the combined four-gene data set, and displays higher resolution and overall clade support (in some cases 20 to > 30 bootstrap percentage points). The data indicate monophyly for only one subgenus (Epigenia) and four sections (Barberina, Cordyloblaste, Neosymplocos, and Urbaniocharis) of Symplocos; the other three subgenera (Symplocos, Hopea, and Microsymplocos) and tested sections (Bobu, Palaeosymplocos, and Symplocastrum) will require recircumscription to accord with the principle of monophyly. Supermatrix and supertree analyses were conducted to provide phylogenetic estimates based on all 107 available Symplocos samples with sequence data from at least two of the five gene regions. The supermatrix consensus tree is completely consistent with that from the combined five-gene estimate based on complete sequence data, whereas the supertree consensus differs from these trees in two regions of the topology. The supermatrix consensus also displays greater clade resolution than the supertree consensus. The aligned trnC-trnD data set exhibits a lower percentage of parsimony-informative characters than those based on matK and rpl16 sequences. The length, organization, and relative sequencing ease of trnC-trnD in Symplocos nonetheless confirm the utility of this region for phylogenetic estimation in flowering plants, as previously described.

Conradina (Lamiaceae) consists of six allopatric species endemic to the southeastern United States, four of which are federally endangered or threatened. The limits and status of several taxa have been contested based on morphological grounds, and clarification of these limits is necessary for the design and implementation of effective and fiscally responsible protection and management plans. The objectives of this study were to investigate the monophyly of Conradina and its relationship to other endemic mints of the southeastern United States, to understand the patterns of diversification in Conradina, and to clarify species relationships. A molecular phylogeny was inferred by sequencing ITS and plastid regions from multiple accessions of each species of Conradina (except for a single accession of C. verticillata) and multiple individuals from species of Clinopodium, Dicerandra, Piloblephis, Stachydeoma, Monarda, Pycnanthemum, and Mentha. ITS sequence data strongly support the monophyly of Conradina, in agreement with evidence from morphology. In contrast, plastid sequence data do not support a monophyletic Conradina and place the genus as paraphyletic to Clinopodium, Stachydeoma, and Piloblephis. Similar plastid haplotypes are shared by different genera, perhaps due to shared ancestral polymorphisms, or more likely, introgression that occurred recently or during the Pleistocene. Within Conradina, ITS sequence data do not resolve species relationships, while plastid sequence data do not support the monophyly of most traditionally defined species of Conradina that are distinguishable morphologically. Species relationships in the plastid data set may also be obscured by introgression or ancestral polymorphism. More rapidly evolving sequence data from nuclear markers will be necessary to clarify relationships in Conradina and related mints from the southeastern United States.

The wild species associated with the Andean domesticate, the pepino, and closely associated with the wild and cultivated potatoes, are morphologically and ecologically variable. We studied 10 of the 11 known species, represented by 35 accessions, of Solanum section Basarthrum, plus material of two putative new species. Given the morphological variability, and cryptic species, molecular studies were appropriate. Amplified Fragment Length Polymorphisms (AFLP) were utilized because they are highly polymorphic, and cover most of the genome. Some 98% of the 292 fragments recovered proved informative. A neighbour joining cluster analysis and principal coordinates analysis largely supported previous taxonomic distinctions based on decades of morphological and biosystematic study. However, two new distinct molecular elements were identified, one autogamous, the other self incompatible, that will be described as new species, and that mimic a species pair of wild tomatoes native to the same region of Peru. The most diverse taxonomic group, the series Caripensia, also proved to be the most diverse genetically (85% of loci polymorphic), allowing morphologically similar species to be distinguished. The AFLP data and species distributions in this series support an hypothesis of rapid evolution and peripheral isolation as evolutionary mechanisms in the geographically and ecologically diverse series Caripensia.