Sean Graham

I used various approaches to investigate mycoheterotrophic and carnivorous monocots, which obtain essential nutrients from soil fungal partners or animal prey, respectively. Fully mycoheterotrophic plants are unable to photosynthesize, and can be difficult to place in plant phylogeny due to substantial morphological modification and elevated substitution rates in all three plant genomes. Mitochondrial genes are expected to be less susceptible to long-branch artefacts in phylogenetic inference than plastid or nuclear genes, as they generally evolve more slowly. I used a phylogenomic approach considering all 37 protein-coding mitochondrial genes to infer the phylogenetic placements of mycoheterotrophic lineages of monocots. I infer that Thismiaceae (excluding Afrothismia) are distantly related to Burmanniaceae, but that fully mycoheterotrophic Afrothismia (Thismiaceae) is the sister group of photosynthetic Taccaceae and core Thismiaceae. The latter finding supports a further independent loss of photosynthesis in Dioscoreales. Scattered distribution of the cox1 intron in distantly related lineages of monocot mycoheterotrophs supports convergent intron gains, potentially consistent with repeated invasions from soil fungal genomes. I demonstrate that Triantha occidentalis (Tofieldiaceae, Alismatales) is a previously overlooked carnivorous lineage with a sticky-trap inflorescence. Field experiments, isotopic data and mixing models demonstrate significant N transfer from prey, with an estimated 68% of leaf N obtained from capture, comparable to levels for co-occurring sundew. Glandular hairs on flowering stems secrete phosphatase, a digestive enzyme seen in other carnivorous plants. Triantha is nearly unique among carnivorous plants in capturing prey on its inflorescence axis close to flowers; however, its glandular hairs capture only small insects. I also studied Triantha phylogeography across its mostly North American range, by surveying 11 plastid-encoded ndh genes from 75 populations. All three North American species are likely monophyletic, although T. occidentalis monophyly requires recognizing T. japonica as a synonym, which is consistent with its nested position in T. occidentalis, as the sister group of populations in Haida Gwaii (Canada). Plastid ndh genes have experienced various degrees of loss or reading frame interruption within T. glutinosa and T. occidentalis, and a strong geographic signal is evident in patterns of ndh gene loss/pseudogenization across the range of T. occidentalis.

View record

The sister orders Pandanales and Dioscoreales are the last two major clades of monocots that remain to be examined in depth using phylogenomic approaches. Previous studies produced basic phylogenetic outlines for both orders, but left multiple higher-order relationships unresolved, including precise placements of non-photosynthetic mycoheterotrophic lineages. Species-level relationships in the Dioscorea yams, the largest Dioscoreales genus, are also poorly known despite the economic importance of this clade. Here I develop and apply phylogenomic methods to reconstruct the evolutionary histories of Pandanales, Dioscoreales and Dioscorea, using the resulting frameworks to address various phylogenetic and evolutionary hypotheses in each group. I employed a bi-organellar phylogenomic approach to resolve higher-order relationships in Pandanales, with mostly strong support. Mitogenomes are generally slowly evolving, but nevertheless permit inference of well-supported relationships. They also exhibit sporadic order-wide rate accelerations that are decoupled from plastome rate variation. I estimated clade ages to address several biogeographic hypotheses, and uncovered plastome structural variants that may define individual families. In Dioscoreales, I used plastid phylogenomic data and morphology to infer higher-order relationships. The molecular analyses resolve higher-order relationships, but are complicated by extreme substitution rate elevation observed in several mycoheterotrophic lineages. Nevertheless, my analyses confirm the non-monophyly of Burmanniaceae and Dioscoreaceae as currently circumscribed in angiosperm classification schemes. An updated morphological data set supports the local phylogenetic placement of several mycoheterotrophic genera for which plastid data are lacking, and ancestral-state reconstructions predict morphological synapomorphies relevant to the revision of family-level classification schemes. For phylogenomic studies of Dioscorea, I developed a customized bait panel comprising 260 low-to-single-copy nuclear genes. I assessed the utility of the panel with a pilot taxon sampling, supplemented with transcriptome data, that comprises representatives of all major Dioscorea clades, including multiple crop yams. The panel enables resolution of phylogenetic relationships at both deep and recent scales in Dioscorea, mostly with strong support, and provides insights into relationships between crop plants and putative crop wild relatives. Overall, my dissertation offers new insights into a broad diversity of evolutionary questions, made possible by resolution here of recalcitrant phylogenetic relationships in Pandanales, Dioscoreales and the Dioscorea yams.

View record

I use organellar phylogenomic approaches to infer overall land-plant phylogeny and higher-order relationships in mosses, liverworts and hornworts, and perform more focused studies of relationships in two bryophyte clades, the granite mosses (Andreaeaceae), and the liverwort genus Herbertus. Using plastid and mitochondrial phylogenomic data sets, I infer conflicts in overall embryophyte relationships across different analyses. These conflicts mask an underlying unrooted tree that is also consistent with most published studies: [(liverworts, mosses), (hornworts, vascular plants)]. Inferred relationships within the major bryophyte clades (mosses, liverworts, hornworts) largely agree across analyses, and between trees inferred from the two organellar genomes. I show that combining genomic and transcriptomic data sources can mislead phylogeny locally for heavily RNA-edited taxa, but that excluding putative edit sites restores cases of unexpected non-monophyly to monophyly. I find that relaxed purifying selection affects multiple plastid genes in a mycoheterotrophic liverwort (Aneura mirabilis) but not a putatively mycoheterotrophic moss, Buxbaumia. Plastid genome structure is nearly invariant across autotrophic bryophytes, but the tufA locus, presumed lost in embryophytes, is unexpectedly retained in several mosses. I reconstruct the broad phylogeny of Andreaeaceae, one of a handful of non-peristomate moss lineages, using a variety of phylogenetic and nuclear phylogenomic data sets and approaches. I infer largely congruent relationships across disparate molecular data sources. These inferences conflict with traditional infrageneric classification, likely reflecting an overreliance on homoplasious morphological characters. I confirm that Andreaea wilsonii is deeply nested in Andreaea, refuting its circumscription as a distinct genus by several authors. I propose an updated infrageneric classification for Andreaea, including three new sections, providing a framework for future systematic studies of the genus. I also use various phylogenetic and nuclear phylogenomic approaches to address species diversity and relationships in Herbertus, a leafy liverwort genus with a complex taxonomic history. My analyses demonstrate an incomplete understanding of species diversity in Herbertus, and reveal widespread misapplication of names due to imperfect morphological species concepts. A reassessment of morphological, ecological, and distributional data in Herbertus, made in light of the new molecular inferences, helps delineate taxa and highlights overlooked species diversity and cryptic species.

View record

Plastid genomes (plastomes) of fully mycoheterotrophic plants (which obtain nutrition from fungi and have lost photosynthesis) may exhibit accelerated substitution rates, gene losses and structural rearrangements compared to their more stable photosynthetic relatives. Distantly related lineages provide independent data points to study plastome degradation. I used Sanger sequencing to assess the utility of three nonphotosynthetic plastid genes in phylogenetic inference of seven monocot families that include mycoheterotrophic taxa. I also assembled full plastome genomes for multiple mycoheterotrophic monocots, a heterotrophic conifer (Parasitaxus, Podocarpaceae) and autotrophic relatives for comparative analysis. Phylogenomic inferences are robust to different likelihood approaches and often extensive gene loss, are generally congruent with the few-gene analyses, and are insensitive to long branches, in contrast to parsimony. Patterns of gene loss and retention are largely in agreement with hypothesized trajectories, starting with plastid NAD(P)H dehydrogenase, followed by the loss of other photosynthesis-related genes, and ending in gradual loss of transcriptional apparatus and other non-photosynthesis related genes. I observed retention (delayed loss?) of genes encoding subunits of plastid-encoded RNA polymerase (Parasitaxus and some species in Petrosavia, Petrosaviaceae), plastid ATP synthase (Petrosavia, perhaps Parasitaxus in modified form) and Rubisco (Petrosavia), consistent with secondary non-photosynthetic functions of the latter two complexes. Some group IIA introns appear to be retained despite the loss of the plastid intron maturase, matK. Retained open reading frames are generally under strong purifying selection in Sciaphila (Triuridaceae). Genome contraction is the major mode of genome rearrangement, with severe reduction seen in some lineages (e.g., Apteria in Burmanniaceae is reduced to ~16 kb). Some mycoheterotrophs are nearly or completely colinear with autotrophic lineages (Geosiris in Iridaceae, at ~123 kb). Others have multiple minor or major rearrangements, which may be unrelated to the presence or absence of an inverted repeat (IR). Four independent IR losses were observed (in Burmanniaceae, Corsiaceae, Petrosaviaceae and Triuridaceae), an extra IR copy evolved in Campylosiphon (Burmanniaceae), and an entire IR re-evolved in Parasitaxus. Shifts in IR boundaries were also found in all mycoheterotrophs. Within-taxon comparisons (e.g., in Corsiaceae and Petrosavia) also underline that idiosyncratic evolutionary changes may occur following each loss of photosynthesis.

View record

In my thesis I aim to improve our phylogenetic and evolutionary knowledge of two ancient and distantly related groups of aquatic flowering plants, Hydatellaceae and Alismatales. While the phylogeny of monocots has received fairly intense scrutiny for two decades, some parts of its diversification have been less frequently investigated. One such lineage is the order Alismatales, which defines one of the deepest splits in monocot evolution. Many families of Alismatales are aquatic or semi-aquatic, and they have been implicated in historical discussions of monocot origins. I evaluate inter-familial relationships in the order, considering a suite of 17 plastid genes for 31 Alismatales taxa for all 13 recognized families. This study improves on our understanding of, and confidence in, higher-order Alismatales relationships. I also uncovered convergent gene loss of plastid-encoded subunits for the NADH dehydrogenase complex. I then expand monocot coverage outside Alismatales by including unpublished and newly sequenced data for otherorders. This large-scale sample facilitated a re-evaluation of monocot phylogeny and molecular dating, the latter using 25 fossil constraints. Previously included in the monocot order Poales, Hydatellaceae are a small family of ephemeral aquatics relatively recently found to be the sister group of water lilies (Cabombaceae and Nymphaeaceae). I present the first molecularphylogeny of the family and evaluate aspects of the family's morphological evolution. I show how sexual system shifts are associated with shifts in otherreproductive traits. I also infer a temporal scale for Hydatellaceae diversification using a two-step Bayesian approach. I use the resulting dated tree to address biogeographic patterns and aspects of niche evolution. I show that its "Gondwanan" distribution is the result of long-distance dispersal and not continental rifting, and demonstrate strong phylogenetic niche conservatism in the family. These studies expand our understanding of evolution in Hydatellaceae, and provide a substantial update to our understanding of Alismatales (and more generally monocot) phylogeny and divergence times.

View record

The unifying theme of this thesis is adaxial-abaxial or dorsiventral patterning in leaves. The adaxial-abaxial axis sets the thickness of a leaf and without the appropriate juxtaposition of the adaxial and abaxial domains, radialized leaves develop. The underlying genetic mechanisms of the development of these polarity defects started to be elucidated only over the past 20 years in the model Arabidopsis, in particular. I investigated this patterning in a variety of non-model species. Firstly, I investigated the variability of dorsiventral polarity in plants with naturally occurring radialized leaves including Allium, Nepenthes, Krishna fig, Pelargonium, several Cactaceae species, and popREVOLUTA mutant of a poplar hybrid. Subsequent chapters aimed to incorporate morphology and anatomy with molecular genetics in order to elucidate the underlying basis of the phenotype of interest in species that have not been used as model systems for leaf development, including canola and poplar. A novel mutant (lamina epiphylla, lip) was identified in canola, which has adaxialized leaves and leaf-derived organs. Some of the HD-ZIPIII candidate genes were sequenced in canola, but I was unable to determine the location of the LIP mutation. The rest of this thesis focuses on the abaxial greening and unifacial petiole phenotypes seen in some species of poplar that have isobilateral leaves (others in the genus have bifacial leaves). YABBY, KANADI, and HD-ZIPIII genes are some of the major contributors to setting proper adaxial-abaxial polarity and I investigated the relationships of these genes by identifying the orthologs in Arabidopsis, poplar, and eucalyptus (a genus that shares the abaxial greening phenotype with poplar). Further, I studied the species relationships within the genus Populus in order to establish the ancestral state of leaf type. I determined that bifacial leaves are likely derived within the genus. Finally, two poplar species (black cottonwood with bifacial leaves and hybrid aspen with isobilateral leaves) were compared on the basis of morphology, anatomy, and molecular genetics in order to determine the underlying basis of the abaxial greening and unifacial petiole phenotypes in hybrid aspen. I identified a subset of genes that may be involved in determining these phenotypes, but further investigation is needed.

View record

Substantial ambiguities still remain concerning the broad backbone of moss phylogeny. I surveyed 17 slowly evolving plastid genes from representative taxa to reconstruct phylogenetic relationships among the major lineages of mosses in the overall context of land-plant phylogeny. I first designed 78 bryophyte-specific primers and demonstrated that they permit straightforward amplification and sequencing of 14 core genes across a broad range of bryophytes (three of the 17 genes required more effort). In combination, these genes can generate sturdy and well-resolved phylogenetic inferences of higher-order moss phylogeny, with little evidence of conflict among different data partitions or analyses. Liverworts are strongly supported as the sister group of the remaining land plants, and hornworts as sister to vascular plants. Within mosses, besides confirming some previously published findings based on other markers, my results substantiallyimprove support for major branching patterns that were ambiguous before. The monogeneric classes Takakiopsida and Sphagnopsida likely represent the first and second split within moss phylogeny, respectively. However, this result is shown to be sensitive to the strategy used to estimate DNA substitution model parameter values and to different data partitioning methods. Regarding the placement of remaining nonperistomate lineages, the [[[Andreaeobryopsida, Andreaeopsida], Oedipodiopsida], peristomate mosses] arrangement receives moderate to strong support. Among peristomate mosses, relationships among Polytrichopsida, Tetraphidopsida andBryopsida remain unclear, as do the earliest splits within sublcass Bryidae. A Funariidae, [Timmiidae, [Dicranidae, Bryidae]]] arrangement is strongly supported, as are major relationships within subclasses Funariidae and Dicranidae. I also reconstructed the phylogeny of the nonperistomate moss family Andreaeaceae, with a focus on costate taxa, using two complementary sets of plastid markers and taxa. The major subgenera (Andreaea and Chasmocalyx) and sections of Andreaea (Andreaea and Nerviae) are rejected as monophyletic. Well-supported lineages include clades comprising: (1) Andreaea nivalis and A. rigida (northern hemisphere members of subgenus Chasmocalyx) and A. blyttii (section Nerviae); (2) most of the remainder of Nerviae; (3) a mixture of costate and ecostate species from Chasmocalyx, Nerviae,all sampled members of section Andreaea, and subgenus Acroschisma. Relationships among the major lineages, including the root of the family, are all well supported.

View record

No abstract available.