Advanced image data analyzer. APS. Ammonium persulfate. ASA ..... these immature oligodendrocytes (Burne et al., 1996). Axon-oligodendrocyte ..... Hyper Cassette. Amersham. Ice machine AFâ10. Scottsman. Imaging plates BAS IP MS 2325. Fuji Photo fil
Computer-mediated discourse. CofP. Community of Practice. CP. Cooperative Principle. D. âDouble complaintâ, i.e. a user complains for two reasons at the same time. D_Statis. The German Federal Statistics Office and the Statistical Offices of the
does not answer the question how the equilibrium can be reached. ...... Note that equations (17), (19) and (20) can be interpreted as a model where fiscal ..... Proposition 1.4.4 and 1.4.5 show that the MSV solution may be E-stable even under ......
recent research shows, e.g. in the Ferlo North Faunal Reserve, Senegal, and in Free State, ... Chapter 1. 2 status of this cat, as there is no information about these aspects, except the Ngorogoro. Crater, Tanzania (GEERTSEMA 1981 & 1985) and on Sout
Dec 22, 2011 - 1.2.3 Linear Approximation . ..... that in a two-country set-up the focus has to be either on the effect of trade openness or the effect of bilateral ...
3.10. US firm-level yearly data: Inflation, liquidity holdings and R&D ...... by an entrepreneurial moral hazard problem and gives rise to a well-defined corpo- ...... 0.81. 1.34. 1.11. 17.16. 1.12. 34.88. 0.58. 0.55. 0.73. 0.3. 6. 0.73 con tem p . c
witnessed Ghanaian businesses monopolizing markets and many foreign industries pulling out of the ... population growth is shifting considerably from rural to urban areas. ..... industrial operations such as Coca-cola, Fan Ice, or Star Beer are locat
Pollination of Medicinal Plants (Nigella sativa and Coriandrum sativum) and ... Matthias Schindler for his dedicated help for the classification of bees. I thank Dr. ...... The study answers the following questions: a) What are the ...... 20.2Â±1.57
Dec 1, 2010 - of Banking and Finance of the University of Bonn: Sven Balder, Michael Brandl, An. Chen ...... implemented in a commercial model package marketed by KMV corporation. ... bankruptcy costs and taxes, and all securities in the market are a
read out from a bit pattern from the EDDA data acquisition and we assume that we have taken the correct sign ...... bit definitions for PLX interrupt status register.
May 6, 2011 - patible with EUT, they are widely assumed in economic and financial modeling due to their simplicity and ..... is a stronger property, not only in theory, but also in practice. In particular, it is not ...... the system of equations (2)
6. Fortunately, this way of viewing. âpsychologicalâ happiness in philosophy has changed over the last several years, ... 8 M. E. P. Seligman and M. Csikszentmihalyi, "Positive Psychology [Special Issue]," American. Psychologist 55, no. ...... is
Example formats for sheet music are BMP, TIFF and PDF. Having specified the meaning of music ...... Piano Sonata Opus 2 No. 1, Allegro, by Ludwig van Beethoven played by Alfred Brendel ...... In the sheet music books âBeethoven Piano Sonatas Volume
pe antagonis greement wi sults from th hich could b rently, Val8 olved in th located at th ffinity of A e252, Met27 nsible for th developed contributio ide-chains, ...... 2.24 Corre ng of agonis sulting bind. 2.9) implie en bonding all adenosin lved
der Rheinischen Friedrich-Wilhelms-UniversitÃ¤t. Bonn vorgelegt von ..... premiums. In response, the economy is forced to repay parts of its external debt (financial deleveraging). ...... âPtrtKtâ1(i) â WtLt(i) â PtACHt(i) â PtAC. â. Ht(i
error-free independent variables (e.g., time, dose, covariates), the fixed ..... parameters by one (degree of freedom [df] = 1) a ÎOFV of at least 3.84 would be ..... transfer compartment into the model until the objective function value (OFV) did n
ii. Sustainable Management of Fruit Orchards in the Soconusco,. Chiapas, Mexico - Intercropping Cash and Trap Crops. SUMMARY. South Mexico belongs to the ...... system. NÂ°*. Rain season. Dry season Modification NÂ°*. Cintalapa. (Mango). 1 Maize stra
zur. Erlangung des Grades. Doktor der Agrarwissenschaften. (Dr. agr.) der ... respectively. Thirty-six h after the injection, half of the BSA injected group (n=30) and half of the saline group ...... Biochemie Clinique en pathologie aviaire. Introit
representations include simple chemical composition formula and more complex notations, such as the ..... different ways, for example, as logarithmic potency difference (e.g., âpIC50 or. âpKi) or normalized potency .... mal inhibitory concentrati
eral neither symmetric (indeed, we normally have W[O,Ï] = W[Ï(O),Ïâ1]), nor does ..... the segmentation of previous images to guide the edge detection in consecutive ..... partial differential equation with a CrankâNicolson finite difference s
Feb 2, 2011 - Schibler U, HagenbÃ¼chle 0, Wellauer PK, Pittet AC (1983) Two ..... Most gratitude to Almighty God and to our Lord Jesus Christ for giving me life ...
Nov 21, 2007 - 4.11 B). This confirmed juxtanuclear subcellular distribution of. CB2-eCFP fusion protein within golgi compartment. Cannabinoid receptors are known to undergo glycosylation (Song and Howlett 1995; Gonsiorek, Fan et al. 2007), which pre
world's appurtenances, none has a more profound potential to improve quality ... from transportation systems to potable water and sanitary sewer services, from ...
more energy at the new time step is approximated based on an inner, secondary variational .... we consider for a test function Ï â C1 the disturbed curve c + ÎµÏ. .... CHAPTER 4. NATURAL TIME DISCRETIZATION FOR ISOTROPIC WILLMORE FLOW. For a test
Chapter 3 How much does it take to resolve relationships and to identify species with molecular markers? An example from the epiphytic Rhipsalideae (Cactaceae) Summary The taxonomic units and species limits in the Cactaceae have been difficult to define and molecular phylogenetic studies so far yielded largely unresolved trees, so relationships within Cactaceae remain insufficiently understood. This study focuses on the predominantly epiphytic tribe Rhipsalideae and evaluates the utility of a spectrum of rapidly evolving and non-coding plastid genomic regions. The study including 51 of the 52 accepted species, and 11 of 13 of the infraspecific taxa. Six plastid regions were sequenced, comprising two group II introns (trnK, rpl16), three intergenic spacers (rps3-rpl16, psbA-trnH, and trnQ-rps16) and matK, totalling c. 4200 nucleotides per sample. These regions were evaluated for their phylogenetic signal and for their species discrimination power for DNA based species recognition based on beforehand defined operational taxonomic units (OTUs). A well resolved and supported specieslevel tree could be inferred. The Rhipsalideae were found to be monophyletic and to contain five major clades that correspond to the genera Rhipsalis, Lepismium, Schlumbergera, Hatiora, and Rhipsalidopsis. The species-level tree was well resolved and supported and the rpl16 and trnK introns yielded the best phylogenetic signal and the best OTU identification potential while matK, psbA-trnH and trnQ-rps16 were less effective in both ways. The highest OTU identifications rate of 97% was found using c. 2500 nt. The phylogenetic performance of the markers was not determined by the level of sequence variability and the species discrimination power did not necessarily correlate with the phylogenetic utility of the markers.
3.1 INTRODUCTION Cactaceae are one of the major floristic components of the New World’s arid as well as seasonally moist tropical regions and at the same time one of the most popular plant families in horticulture. While there is little doubt that Cactaceae are a natural group considering morphological and molecular synapomorphies (Barthlott & Hunt 1993, Nyffeler 2002, Wallace & Gibson 2002), the recognition of tribes, genera and species within the family has always been difficult. Many cacti look similar due to convergent evolution, which is frequent in the family – large columnar forms, small globular cacti and epiphytes with flattened, leaf-like stems are suspected to have evolved each several times (Barthlott & Hunt 1993, Wallace & Gibson 2002). Until now, relationships within the Cactaceae are insufficiently understood and fairly few molecular phylogenetic studies have been conducted, contrary to other popular plant families such as orchids or bromeliads. So far, only major clades of Cactaceae have been identified but their interrelationships remained largely unresolved (Nyffeler 2002, Wallace & Gibson 2002, Hernández-Hernández & al. 2011; Bárcenas & al. 2011). But many tribes and genera were shown to be either poly- or paraphyletic, indicating that they had either been based on plesiomorphic or convergent morphological characters (Applequist & Wallace 2002, Butterworth & Wallace 2004, Arias & al. 2005, Edwards & al. 2005, Ritz & al. 2007, Korotkova & al. 2010). Besides, species-level trees for Cactaceae hitherto remained largely unresolved or weakly supported statistically due to low sequence divergences or insufficient data and sampling. Strongly increased taxon sampling (666 taxa) did not improve on this (Bárcenas et al., 2011). Attempting to resolve a Cactaceae tree, especially at the species level, seems therefore challenging and a combined analysis of genomic regions selected for their high phylogenetic utility and putative performance at species-level was therefore tempting. A recent comparison of the mutational dynamics of non-coding chloroplast regions (introns and spacers) indicated differences in phylogenetic structure even among highly variable non-coding DNA (Borsch & Quandt, 2009). At lower distance levels, i.e. between genera and species, the addition of more chloroplast intron and spacer sequences into combined matrices has generally resulted in increased resolution and support for the inferred trees (e.g. Barfuss & al. 2005, Löhne & al. 2007, Tesfaye & al. 2007). However, phylogenetic structure per informative site has not been compared in detail and the combined dataset of six markers in this survey provides a good case for study. This study focuses on the tribe Rhipsalideae DC., which is one major group of in total four lineages of epiphytic cacti (Korotkova & al. 2010). The Rhipsalideae occur mainly in South American tropical and subtropical rainforests, with a center of diversity in the Mata Atlântica. A few species are also found in the Northern and
Chapter 3 Central Andes. All Rhipsalideae are predominantly epiphytic and/or epilithic and only rarely terrestrial; exhibiting mostly a pendent or semi-erect, shrubby habit with terete, angular or flattened and sometimes almost leaf-like stems. Flower morphology ranges from medium-sized colored bird-pollinated flowers in Schlumbergera Lem. to small insect pollinated white flowers in Rhipsalis Gaertn. and Lepismium Pfeiff. Rhipsalis is the largest and most widely distributed genus of epiphytic cacti and Rhipsalis baccifera (Mill.) Stearn is the most widespread of all Cactaceae species. Besides it is the only cactus with a natural distribution area extending beyond the Americas into tropical Africa, Madagascar and Sri Lanka (Barthlott 1983). Rhipsalideae is the oldest name for any epiphytic Cactaceae group at higher rank, and was established by A. P. de Candolle (1828). The tribe in its initial circumscription contained only Rhipsalis; other genera were yet to be described. Following the addition of more and more species and genera, generic limits became controversial. Establishing sound generic concepts was difficult due to intergrading vegetative characters, phenotypic plasticity and the largely uniform flower morphology. The two main kinds of treatments were either combining most of the small flowered taxa in an expanded genus Rhipsalis (Schumann 1899, Vaupel 1925-1926, Hunt 1967) while recognizing the larger-flowered taxa as generically distinct, or to accept several small genera (e.g. Britton & Rose 1923, Buxbaum 1962). The total number of genera recognized in the past has consequently varied from two (Vaupel, 1925-1926, Hunt 1967) to nine (Backeberg 1959, 1966), reflecting differing emphases on similarities or on differentiating characters. The Rhipsalideae currently comprise four genera Lepismium, Rhipsalis, Hatiora Britton & Rose and Schlumbergera, totaling 52 accepted species (Hunt 2006). That treatment is largely based on the nomenclatural proposals of Barthlott (1987a) and the commented checklist of Barthlott & Taylor (1995), but molecular data subsequently revealed Lepismium as polyphyletic and a part of it is now excluded from the Rhipsalideae (Nyffeler 2002, Korotkova & al. 2010). Leaving aside the species excluded from Lepismium, a clade that could be referred to as “core Rhipsalideae” was resolved with 100% bootstrap support, but this finding was based on sampling only a single species for each genus (Nyffeler, 2002; Hernández-Hernández et al., 2011; Bárcenas et al., 2011). More detailed hypotheses on Rhipsalideae relationships based on sequence data of trnQ-rps16, rpl32-trnL, psbA-trnH and ITS have been recently published, focussing on Schlumbergera and Hatiora (Calvente & al. 2011). The Rhipsalideae and the genera besides Hatiora were found as monophyletic, but only based on the plastid data. ITS trees depicted a basal polytomy and the relationships between genera and especially between species remained largely resolved or weakly supported.
Chapter 3 Unstable generic limits and constant movement of species between Rhipsalideae genera has resulted in instability of names. Species boundaries have also been controversial, and often gradual variation in morphological characters fostered extreme divergence of “lumping” and “splitting” treatments. As a result, there are about 450 names for the currently accepted 52 Rhipsalideae species (listed by Barthlott & Taylor, 1995). To give one example: Lepismium cruciforme, the type species of Lepismium has been described under more than 30 names (Britton & Rose 1923). Although DNA barcoding has emerged as a new tool to recognize and later identify species (Hebert & al. 2003), no such approach has yet been attempted for the Cactaceae, albeit necessary. Due to the problems described above, Cactaceae taxonomy is still far from reliable. A high proportion of cacti are believed to be threatened with extinction, and most are CITES-listed (Hunt 1999). An accurate understanding of species limits and the availability of reliable identification tools is therefore desirable for Red Listing and conservation planning. In addition to phylogenetics, we will therefore also examine our data sets with respect to species identification power of different plastid regions. The Rhipsalideae are well-suited for this purpose: they are a comparatively small group and most of the taxa are well known morphologically and thus allow for the clear determination of Operational Taxonomic Units (OTUs). In addition to that, Rhipsalideae are among the best-collected Cactaceae groups and well represented in botanical collections so that enough documented material exists and all but one species were available for inclusion in this study. Only few DNA barcoding studies in flowering plants so far used a full taxonomic setting of all known species of a group and also multiple individuals to assess intraspecific variation. Examples include Paeonia sect. Moutan (Paeoniaceae, Zhang & al. 2009), Crocus (Iridaceae, Seberg & Petersen, 2009) and Psiguria (Cucurbitaceae, Steele & al. 2010). One of the major challenges of such barcoding approaches is to find the most effective markers that allow as many species as possible to be distinguished. This requires a large number of sequence characters in order to accumulate enough variable sites, especially in recently diverged groups with low levels of sequence divergence. Seberg and Petersen (2009) concluded that about 5800 bp would be necessary to identify all Crocus species, which corresponds to 8-9 chloroplast regions and Steele & al. (2010) found at least four regions were required for Psiguria. Two chloroplast markers, the rbcL gene and the fast evolving matK gene, have been recently adopted as plant barcodes by the Consortium for the Barcoding of Life (CBOL Plant Working Group 2009). Both markers had been among the most frequently proposed barcoding regions, among with the psbA-trnH spacer (Kress & al. 2005, Cowan & al. 2006, Kress & Erickson, 2007), although various other markers had also been evaluated for barcoding purposes (Taberlet & al. 2007, Fazekas & al. 2008,
Chapter 3 Ford & al. 2009). Usually, these were suggested in view of their simple similaritybased discrimination utility (BLAST approach) irrespective of their phylogenetic signal. For our study of Rhipsalideae, we have selected six structurally different rapidly evolving plastid regions: two group II introns (trnK, rpl16), three intergenic spacers (psbA-trnH, trnQ-rps16, and rps3-rpl16) and matK. All regions were known to be highly variable at low taxonomic levels and/or have been proposed as candidate regions for DNA barcoding. In addition, trnK/matK, rpl16 and psbA-trnH have already been successfully applied within Cactaceae, offering possibilities to compare phylogenetic performance or patterns of molecular evolution and combining datasets. The trnK/matK region is one of the best established phylogenetic markers. It provides a high number of informative characters, even at low taxonomic levels, exhibits high phylogenetic structure (Müller & al. 2006, Borsch & Quandt, 2009) and, as stated above, matK is among the most promising candidates for a barcode (e.g. Chase & al. 2007, Lahaye & al. 2008). The psbA-trnH spacer is among the most variable chloroplast spacers. Although there are some problems limiting its usage, such as frequent indels, microsatellites, inversions and a high degree of homoplasy (Borsch & Quandt, 2009, Devey & al. 2009, Whitlock & al. 2010), psbA-trnH may still be a successful barcode marker due to its high intraspecific variability (Cowan & al. 2006, Chase & al. 2007, Kress & Erickson 2007, Seberg & Petersen 2009). The rpl16 intron is the most variable chloroplast intron (Kelchner 2002) and is one of the most frequently used markers in phylogenetics. It has so far shown high intraspecific variability and yielded good phylogenetic signal between closely related taxa, compared to other chloroplast markers in the same taxon set (Löhne & al. 2007, Tesfaye & al. 2007, Sánchez del-Pino & al. 2009). Although rarely used so far, the trnQ-rps16 spacer is expected to be informative at low taxonomic levels as well. Evidence for this comes from the high percentage of potentially informative characters (PICs) as found by Shaw & al. (2007) and the results of Calviño & Downie (2007) and Fleischmann & al. (2010). Phylogeny reconstruction and barcoding are different approaches. Even if the sequence data would not resolve the evolutionary relationships due to lack of information or conflict among informative sites, the same markers may provide enough autapomorphic substitutions to distinguish between species. Nevertheless, it is likely that markers which contain sufficient information to resolve phylogenetic relationships will be valuable DNA barcodes as well. We were therefore interested to examine if there is a correlation between overall variability of a genomic region (useful for barcoding) and phylogenetic structure (required for tree inference). Our approach is twofold: Using the same data set, we first aim at resolving phylogenetic relationships at species level. Secondly, we evaluate which are the best suited markers for DNA-
Chapter 3 based species recognition within Rhipsalideae, either alone or in combination. Moreover, we will discuss the impact of molecular characters for delimitations of genera and species within Rhipsalideae also in light of the evolution of morphological characters.
3.2 MATERIAL AND METHODS 3.2.1 Plant material and taxon sampling The plant material used in this study was largely obtained from the living collections of the Botanical Gardens of the University of Bonn, where the world’s probably most comprehensive living collection of the Rhipsalideae has been established over three decades by W. Barthlott. Further material was obtained from the Rhipsalideae collections of the Botanical Garden Berlin-Dahlem and the Royal Botanic Gardens, Kew, as well as from the Sukkulenten-Sammlung Zürich. We have sampled 52 species including all the infraspecific taxa. Taxon sampling followed the most up-to-date reference work for the Cactaceae (Hunt, 2006) where 53 species are accepted in Rhipsalideae. Rhipsalis goebeliana Backeb. was sampled additionally. Lepismium incachacanum (Cárdenas) Barthlott, classified as Rhipsalideae therein, was not sampled since we recently found it not to belong therein (Korotkova et al., 2010). No material was available of Rhipsalis ormindoi N.P. Taylor & Zappi and the recently described Rhipsalis aurea M. F. Freitas & J. M. A. Braga (de Fatima Freitas et al., 2009). Morphologically variable and widely distributed species such as R. micrantha, R. teres and R. baccifera were represented by specimens from different countries or collection sites, thus covering some of their intraspecific variation. In total, our analysis contains 110 ingroup and 5 outgroup taxa. All taxa sampled with their origins and voucher information are listed in Appendix 1.
3.2.2 Isolation of genomic DNA Isolation of DNA from cacti is troublesome due to the high mucilage content of the tissue. Initial attempts using a commercial DNA extraction Kit (Plant Genomic DNA Mini Kit, Avegene Life Science Corp., Taiwan) yielded poor results because columns were easily clogged, DNA yield was low (c. 5-30 ng/μl) and the DNA was impure (A260/A280 values were usually between 2.5 and 3). For efficient isolation of DNA we removed most of the water-storing tissue as soon as possible after collection and dried the remaining cortex tissue over silica-gel in a drying chamber for one or two days at 50°C. This treatment significantly lessened the amount of mucilage during extraction. The dried plant material was homogenized (Retsch mixer mill MM200,
Chapter 3 Haan, Germany), incubated for 20 minutes at 65°C with 700 μl of extraction buffer containing 2% CTAB, 1% PVP, 100 mM Tris (pH 8), 20 mM EDTA, 1.4 M NaCl, and 0.2 vol% mercaptoethanol. Further steps followed the procedure described by Borsch & al. (2003). Only two extractions were carried out, since measurements of DNA concentration showed a very low amount of DNA (less that 5 ng/μl) in the third fraction. Concentration and purity of the DNA (A260/A260 as well as A260/A230 ratio) were measured using a spectrophotometer (NanoDrop. peqLab, Erlangen, Germany). This isolation method yielded a high amount (120 to 1000 ng/μl) of clean DNA, with an A260/A280 value between 1.7 and 2.1. Original genomic DNA was stored at -30°C and working dilutions with a standard concentration of 10ng/μl were made for use in PCR.
3.2.3 Amplification and sequencing Amplification conditions and primers used were the same as described in Chapter 2. All primers used for amplification and sequencing are listed in the Appendix 2. All PCR products were stained with 100x SybrGreen nucleic acid stain and electrophoresed on a 2% agarose gel, excised and purified using the Gel/PCR DNA Fragment Extraction Kit (Avegene Life Science Corp., Taiwan) and sequenced via Macrogen Inc. (Seoul, South Korea). All chloroplast regions were easily amplified and sequencing was also straightforward. All regions were sequenced using the amplification primers, additional internal sequencing primers (see Appendix 2) were used if reads were short. Pherograms were edited and sequences were assembled using PhyDe v. 995 (Müller & al. 2005+, www.phyde.de).
3.2.4 Sequence alignment, coding of length mutational events Sequences were aligned manually using PhyDe v. 0995 (Müller & al. 2005+). Rules for the alignment of length variable DNA followed Kelchner (2000) and Löhne & Borsch (2005). All sequences could be aligned unambiguously and only homonucleotide stretches and one (AT)n microsatellite had to be excluded from the matrices (Appendix 3). Indels were coded according to the Simple Indel Coding method using the Indel Coder option of SeqState v. 1.40 (Müller, 2005b). A list of hypothesized microstructural mutations was compiled (Appendix 4) to allow later testing of homology hypotheses (see Borsch & al. 2007, Morrison 2009, Ochoterena 2009). Inversions were placed separately during alignment and reverse-complemented prior to phylogenetic analyzes. Secondary structures of sequence parts with inversions were calculated using RNA structure 5.0 (Mathews & al. 1996+) to check whether these inversions were associated with hairpins. The inversions were coded manually (assumed plesiomorphic state: 0, inverted state: 1) and traced on the phylogenetic trees using the “Trace Character history” option of Mesquite v. 2.72 (Maddison & Maddison, 2009).
3.2.5 Phylogenetic analyses Most parsimonious tree search was carried out using the ratchet as implemented in PRAP (Müller 2004) with the combined dataset and each marker individually. The analysis with the combined dataset was performed including all accessions and also with a reduced dataset with only one accession per OTU. Ratchet settings were 200 iterations with 25% of the positions randomly upweighted (weight = 2) during each replicate and 10 random addition cycles. Tree lengths and homoplasy indices (CI, RI, and RC) were calculated in PAUP* v. 4.0b10 (Swofford 1998). Support for the nodes found by the parsimony ratchet was calculated by jackknifing (JK) with 10.000 replicates, TBR branch swapping, 36.788% of characters being deleted in each replicate and one tree held during each replicate. These settings are based on optimal jackknife parameters described by Müller (2005a). Bayesian Inference (BI) was performed with the combined dataset using MrBayes 3.1 (Huelsenbeck & Ronquist, 2001) with GTR+Γ+I as the best-fitting substitution model as evaluated with jModeltest (Guindon & Gascuel, 2003, Posada, 2008) using the Akaike Information Criterion (AIC). Analyses were performed based on substitutions only and in combination with coded indels, then applying the restriction site (binary) model for the indels partition. Four simultaneous runs of Metropolis-coupled Markov Chain Monte Carlo analyzes, each with four parallel chains, were performed for five million generations, saving one tree every 1000th generation, starting with a random tree. Other MCMC parameters were left with the program’s default settings. The burn-in was determined using Tracer v1.5 (Rambaut & Drummond, 2007) and set at generation 500000, the remaining trees were summarized in a majority rule consensus tree. All trees were imported into the tree editor TreeGraph2 (Stöver & Müller, 2010) for annotation and layout.
3.2.6 Comparison of marker performance / phylogenetic structure R Phylogenetic structure R sensu Müller & al. (2006) was estimated with help of a Perl script as described therein, modified to better account for severely staggered alignments (Krug & al. in prep.). The data partitions were defined as: trnK intron, partial matK – c. 950 nt, as they would be amplified by the primers designed for Caryophyllales by (Cuénoud & al. 2002) and proposed by (Lahaye & al. 2008) for the amplification of matK for barcoding purposes, the entire matK CDS, the rpl16 intron, psbA-trnH and trnQ-rps16. All partitions were compared with each other and analyzes were run with all characters included and only with the informative characters.
3.2.7 Definition of operational taxonomic units (OTUs) A concept using OTUs instead of species names was employed as a basis for any calculations of intraspecific variability or species identification potential of markers. This was done because species limits within the Rhipsalideae have often changed. There are several taxa that have been described as species and later have been downgraded to subspecies or forms or vice versa. Hence, we did not assume that all currently accepted species names reflected “good” species; there might be subspecies that probably merit specific status and vice versa. Then, the phylogenetic hypothesis provided a reliable estimation on OTU delimitation. A list of the defined OTUs is given in Appendix 5. All the OTUs are morphologically recognisable and do correspond to species or subspecies or forms, no OTU was defined just based on sequences.
3.2.8 Testing of OTU identification success The OTU identification success rate for each data partition/marker and any combination of these was computed via a Perl script written by K. Müller (University of Münster) that comprised the following computational steps: First, the individual accessions were assigned to OTUs and this information was read from an OTU definition file. Second, all possible combinations of the data partitions were constructed by reading Nexus files and concatenating sequences accordingly. In doing so, the average number of nucleotides sequenced for each set was computed as a coarse proxy for sequencing effort. The data partitions for testing of OTU identification were defined as above for comparisons of phylogenetic structure, with the only exception that “partial matK” was not included in the successive marker combination analysis as it requires non-overlapping data partitions. All matrices were the same as used for the phylogenetic analyzes, i.e. with mutational hotspots excluded and inversions reversecomplemented. An OTU was considered identifiable if none of the sequences of a given OTU was identical to any of the sequences of another OTU. OTU monophyly was therefore not a requirement for identifiability. In testing equality of two sequences, alignment positions with ‘?’ or ambiguity codes in any of the two sequences were ignored. Uppercase and lowercase letters (the latter reflecting manually edited bases deviating from automated base calls) were treated equally. If one sequence had a gap character at a given position while the other had not, the sequences were treated as different. The percentage of OTUs uniquely identified this way was computed, and this was repeated for all possible combinations of markers.
3.3 RESULTS 3.3.1 Sequence characteristics The final combined matrix comprised 5201 aligned characters, with an average length of 4287 nt per taxon. In total, 15 sequence parts of uncertain homology (mutational hotspots) had to be excluded (Appendix 3). After their exclusion, 4887 aligned characters remained within the matrix with on average 4195 nt per taxon. The full characteristics of the individual regions for the dataset including and excluding hotspots are given in Table 3.1. The psbA-trnH spacer provided the highest percentage of variable and informative characters, followed by the rps3-rpl16 spacer and the rpl16 intron while the trnK intron and the matK gene were least variable. Alignment was straightforward for matK, the trnK intron and rpl16 where mutational hotspots were restricted to poly-A or poly-T stretches but more troublesome for psbA-trnH and especially for trnQ-rps16 where homology of numerous overlapping indels had to be assessed carefully and inversions required further attention.
3.3.2 Microstructural mutations The individual sequence parts marked as mutational hotspots were between 1-3 and 32 nt in length (Table 3.1), the largest hotspots occurred in the rpl16 intron. All hotspots taken together comprised only a small portion of combined dataset, on average 59 nt in length ranging from 42-89 nt. All hotspots were mononucleotide stretches (poly-A or poly-T) or in one case a dimeric (AT)n simple sequence repeat in the rpl16 intron, there were no unalignable sequence parts. Six inversions were observed in all regions except rps3-rpl16 and the rpl16 intron (Table 3.1). All inversions were associated with hairpins and affected the nucleotides forming the terminal loops or stem-loops. The trnK/matK region showed few indels apart from length variable homonucleotide strands. All indels within the matK CDS had a length of multiples of three so the codon structure of the gene is maintained. Highest length variability was observed in the rpl16 intron where six gaps spanned more that 100 nt, the largest being 410 nt. Gaps larger than 100 nt occurred in psbA-trnH and trnQ-rps16 in Rhipsalis and Lepismium.
Table 3.1 Sequence statistics of individual regions in the combined dataset
Dataset including hotspots Position in the alignment Aligned length Length range Mean length (SD) Mutational hotspots Length range of all hotspots Mean length of all hotspots % GC Inversions Dataset excluding hotspots Position in the alignment Aligned length Length range Mean length (SD) % variable characters % informative characters Number of coded indels
3.3.3 Trees from the single loci The single data partitions do not resolve the tree of the Rhipsalideae (Appendices 6 and 7). The trnK intron was least homoplastic (HI 0.158) while the rpl16 intron showed the highest degree of homoplasy in the dataset (HI 0.307). Best resolution from a single partition is obtained from rpl16, albeit with lower support compared to the combined dataset. Besides rpl16 is the only marker to find all major Rhipsalideae clades with high support. Trees from psbA-trnH and trnQ-rps16 result in a large and weakly supported polytomies, with few terminal clades found. Resolution and support from psbA-trnH is weakest, none of the major nodes is found and even the Rhipsalideae are not found as monophyletic, and similar results are obtained from trnQ-rps16 that finds only two clades with support.
3.3.4 Trees from the combined plastid data set Of 4887 total characters in the combined matrix, 546 were parsimonyinformative. The addition of indels provided 113 additional informative characters (of total 165 coded indels). The parsimony analysis including indels resulted in a strict consensus of 144 trees of 1669 steps (CI: 0.712, RI: 0.905, RC: 0.644, HI 0.288), not shown. Figure 3.1 shows the majority-rule consensus tree derived from Bayesian Inference as phylogram. The parsimony tree resulting from the reduced dataset is shown in Figure 3.2. The Rhipsalideae tree was well resolved and supported in both parsimony and Bayesian analyses and species-level resolution could be obtained with high confidence. Rhipsalideae were maximally supported as monophyletic and comprised five well supported clades, which largely agree with the Rhipsalideae genera as currently understood. Rhipsalis and Lepismium are confirmed as monophyletic while the two Hatiora subgenera Hatiora and Rhipsalidopsis p.p. are found as two separate clades and H. epiphylloides is within Schlumbergera. The topologies from both analyzes differ in the position of the genera: the MP topology finds Schlumbergera as sister to the rest of the Rhipsalideae, Hatiora subg. Hatiora to branch off next, followed by Hatiora subg. Rhipsalidopsis, and Lepismium as sister to Rhipsalis, but none of these backbone nodes gets support. The Bayesian analysis finds a weakly supported clade of Lepismium and Hatiora subg. Rhipsalidopsis (0.6 PP) while the positions of the other genera are unresolved. Within the individual genera, the trees from both analyses were almost identical, but the Bayesian analysis provided generally better resolution and higher support values.
3.3.5 Phylogenetic structure R The results are shown in Table 3.2. The rpl16 intron (along with the rps3-rpl16 spacer) showed highest phylogenetic structure R compared to all other markers in this dataset, regardless whether all or only the informative characters were included. The trnK intron had the second-best phylogenetic structure but performed equally well as psbA-trnH when only informative characters were considered. The performances of the other markers differed in the analyses, especially trnQ-rps16 was found to perform better based on the informative characters only. The matK gene, either entire or partial, exhibited lower R then the two introns in the dataset. When compared directly with each other, partial matK showed higher R than the complete gene, the entire matK showed lowest R in both comparisons.
3.3.6 Success of OTU identification The comparison OTU identifications success of each marker is shown in the Appendix 5. The percentage of identified OTUs for each marker combination and in relation to the number of nucleotides sequenced is shown in Fig. 3.3 and Fig. 3.4. The number of successfully identified OTUs increased with more nucleotides and the value of 90% identified OTUs is already reached with slightly more than 1600 nt (psbA-trnH + rpl16 intron + trnQ-rps16). The maximal value of 97% successfully identified OTUs is first reached by 2500 nt (psbA-trnH + rpl16 intron + trnK intron + trnQ-rps16) and even the combination of all markers and 4207 nt does not find more. Hence, of the 61 defined OTUs, 59 could be successfully identified. The only OTUs that could not be found by any marker or combinations were Rhipsalis sulcata and Rhipsalis teres - in each case the R. sulcata sequence was identical with one of the R. teres accessions.
Figure 3. 1 Phylogram from Bayesian Inference based on the combined dataset and coded indels for A) Schlumbergera, Lepismium, Hatiora ad Rhipsalidopsis and B) Rhipsalis. Posterior Probabilities are shown above, JK support values from 10000 replicated below the branches. OTUs with multiple accessions are annotated with square brackets, the Bonn Botanic Garden accession numbers for each sample, the CA-isolate numbers and the countries of origin are given next to the names. Species classification follows Hunt (2006).
Figure 3.1, continued
Figure 3.2 Strict consensus of 11 trees (1556 steps, CI: 0.726, RI: 0.856, RC: 0.621) found by the parsimony ratchet based on the combined dataset and coded indels, annotated with the subgeneric classification of Barthlott & Taylor (1995) and new subgenera as they are proposed here. Jackknife support values from 10.000 replicates are shown above the branches.
3.4 DISCUSSION 3.4.1 Major relationships within Rhipsalideae The dense taxon sampling in our study for many characters unraveled the five major lineages of Rhipsalideae with much improved confidence over previous studies but still could not clarify the relationships between them. The maximum parsimony consensus tree based on plastid data of Calvente et al. (2011) is just inconsistent with weak support, whereas a Bayesian posterior probability of 0.91 alone supports a Hatiora-Lepismium-Schlumbergera-clade. However, a clade hypothesis based on a posterior probability alone, not even reaching 0.95, should be valued with caution (Simmons et al. 2004, Suzuki et al. 2002). There are also only three earlier hypotheses on relationships within the tribe. Berger (1926), Buxbaum (1967) and Barthlott (1987b) had developed their phylogenetic schemes based on an evaluation of characters and an assumed “direction” of evolution. Berger further discussed hypothetical ancestral character states. But most of these earlier assumptions can not be confirmed in view of our data. The basal grade of Schlumbergera, Hatiora and Rhipsalidopsis is unsupported, but all genera share colored flowers and strictly determinate stem-segments, with new segments arising from composite apical areoles in a somewhat oblique position to the preceding one. This indicates that these genera might possess the plesiomorphic states for these characters. In light of the evolution of many other vegetative and floral characters (Chapter 3, this study), the parsimony topology, although the backbone is unsupported, may well reflect the organismic phylogeny. Schlumbergera is found as sister to the rest of the Rhipsalideae. A common earlier view was to regard the morphology of Schlumbergera (or Zygocactus) as most “derived”, because of the zygomorphic flowers (e.g. Barthlott 1987b). Berger (1926) and Buxbaum (1967) further supposed Schlumbergera (and Zygocactus) to have evolved from flat-stemmed taxa with colored actinomorphic flowers as they are found in Rhipsalidopsis. Thus, they assumed close relationships of the two genera but our data do not provide evidence for such a relationship. Our data reveal Hatiora sensu Barthlott and Taylor (1995) as polyphyletic, as also found by Calvente et al. (2011). Barthlott (1987b) classified Hatiora and Rhipsalidopsis both as subgenera of Hatiora, while all preceding authors regarded them as distinct from each other. Our data reveal Hatiora in this expanded circumscription as polyphyletic and find both subgenera as subsequently branching lineages. Alternatively, Hatiora s.str has been regarded as close to Rhipsalis and has even been included in it by Vaupel (1925-1926) and Hunt (1967), mainly because both
Chapter 3 genera produce small flowers. Still, our study does not find any evidence of a close relationship of Hatiora and Rhipsalis. Lepismium and Rhipsalis have been regarded as closely related by Berger (1926) and Buxbaum (1967), although in a different circumscription with only L. cruciforme, while Barthlott (1987a) assumed Lepismium including Pfeiffera, Acanthorhipsalis and Lymanbensonia to be sister to the other genera and the most “ancestral” group of the Rhipsalideae. Rhipsalis and Lepismium are morphologically similar in having small white flowers and terete or flattened stems, but there are no apparent morphological synapomorphies of the two genera. Lepismium as a whole, or parts of it had sometimes been merged in Rhipsalis. Our results provide evidence that both may indeed be sister groups, although the relevant node remains unsupported.
3.4.2 Relationships within main Rhipsalideae lineages, circumscription of genera and subgeneric classification The relationships within the genera of Rhipsalideae could be resolved with high confidence and our results largely confirm the circumscriptions of genera and subgenera as currently understood. Unless stated otherwise, the relationships depicted in our study will be discussed in comparison with the treatments of Barthlott and Taylor (1995) and Hunt (2006). Figure 3.2 shows the earlier classification in comparison with the revised classification as proposed and discussed here.
A clade consisting of the six recognised Schlumbergera species is supported with 100% JK, 1.00 PP, but it additionally includes Hatiora epiphylloides. The Schlumbergera clade as depicted by our data consists of three sublineages: S. opuntioides and S. microsphaerica are sister to the rest of the genus, a position which is also supported by their morphology. They differ in having cylindrical or compressed stem-segments bearing areoles all over the surface of the stems. These two species were originally treated as a separate genus Epiphyllanthus Berger but later interpreted as neotenic forms of Schlumbergera (Barthlott & Rauh, 1975). Hatiora epiphylloides was originally described as Rhipsalis epiphylloides Porto & Werderm. Backeberg (1938) established a monotypic genus Pseudozygocactus Backeb. for it, which was included in Hatiora by Buxbaum (1970b). The current view of this species being part of Hatiora subg. Rhipsalidopsis was proposed by Barthlott (1987a). The placement of Hatiora epiphylloides within Schlumbergera s. str. is unexpected but but was also found by Calvente et al. (2011) and is supported by the plant's stem morphology. The plants are usually smaller in size, but large specimens have been observed in the collection of Countess B. Orssich (W. Barthlott, pers. obs.). The flowers have the structure of a Hatiora flower and yellow color, which is typical for Hatiora but does not occur in any other Schlumbergera species. Actually, the flowers of Hatiora 69
Chapter 3 epiphylloides generally lack all flower synapomorphies of Schlumbergera, such as stamens inserted in two series, a perianth tube and a nectar chamber. The species therefore seems a morphological intermediate. Possible explanations are either morphological homoplasy or convergence or ancient hybridization between a Hatiora s. str. and a Schlumbergera. This hypothesis still needs confirmation from sequences of nuclear markers, in view of the maternal inheritance of the plastid genome. Schlumbergera is known to hybridize freely, the commonly cultivated Christmas Cactus (Schlumbergera × buckley) is a hybrid between S. truncata and S. russelliana and a hybrid between S. truncata and S. opuntioides (Schlumbergera × exotica) is also known (Barthlott & Rauh, 1975). Hybridization may therefore also have played a role during speciation in Schlumbergera. Calvente et al. (2011) did sequence the nuclear ITS region but the ITS tree is basically a large polytomy and their data neither confirm nor reject the possibility of hybridization within Schlumbergera, so other nuclear loci would be needed. The clade consisting of S. russelliana (the type species of Schlumbergera), S. kautskyi, S. orssichiana and S. truncata is well supported (100% JK, 1.00 PP) and can be regarded as Schlumbergera in the strict sense. Schlumbergera kautskyi, which had originally been described as a variety of S. truncata and later raised to species rank, is resolved as distinct and confirmed as a “good” species. Schlumbergera truncata and S. orssichiana are supported as closely related, cannot be separated by the phylogenetic analyzes, but are still found as distinct OTUs. Schlumbergera orssichiana differs considerably from S. truncata by shape and size of its stem-segments, flower morphology and an unusual flowering behaviour, including flowering in summer (Barthlott & McMillan, 1978). Although Schlumbergera consists only of six species and is morphologically well defined, it has had a complex taxonomic history. Some species had been separated as distinct genera (Hunt 1969, McMillan & Horobin, 1995). Our study supports an expanded Schlumbergera to include Hatiora epiphylloides, as it was also suggested by Calvente et al. (2011). But including H. epiphylloides also poses some problems. Schlumbergera is one of the morphologically best defined Rhipsalideae genera, maybe even one of the best defined Cactaceae genera. The features characteristic for it are predominantly zygomorphic flowers with a nectar chamber, a perianth tube, erect, connivent stigmas and stamens inserted in two series. None of these are found in H. epiphylloides. Including it in Schlumbergera would make the genus morphologically heterogeneous. It remains to be tested if nuclear genes result in a deviating phylogeny and if H. epiphylloides perhaps a striking case of reticulate speciation.
Our data reveal Hatiora as polyphyletic. The generic name should only be applied to subgenus Hatiora which includes taxa with cylindrical stems, a terete pericarpel and small yellow-orange or magenta flowers. The corresponding clade of H. salicornioides (type species), H. cylindrica and H. herminiae is highly supported (100% JK/1 PP) and the morphologically different magenta-flowered H. herminiae is resolved as sister to the other two species. Hatiora cylindrica falls into a clade of H. salicornioides specimens. The main characteristics of H. cylindrica are cylindrical stem-segments, a fully expanded perianth and deep red fruits, while H. salicornioides has bottle-shaped stem-segments, flowers which do not open widely and white fruits. Our data indicate that H. cylindrica might either not be a “good” species but a form or variety of H. salicornioides. But it is also possible that what is known as H. salicornioides is more than one species. This is even likely because very distinct races and ecotypes exist in the wild (N. Taylor, pers. obs). Some H. salicornioides forms have been described as separate taxa, but species-limits are hard to define because of intergrading characters and further differences possibly attributable to cultivated plants, so the additional species names are currently treated as synonyms.
The clade consisting of Rhipsalidopsis (= Hatiora) rosea and R. (= Hatiora) gaertneri together with their hybrid R. × graeseri is supported with 100% JK, 1 PP. Rhipsalidopsis was originally established as a genus by Britton & Rose (1923) for R. rosea, which they had separated from Rhipsalis. Rhipsalidopsis gaertneri was at first placed in Schlumbergera but later Moran (1953) combined into Rhipsalidopsis. Barthlott (1987a) had merged Rhipsalidopsis in Hatiora, but as stated above, this expanded Hatiora is polyphyletic. Contrary to the proposal of Calvente et al. (2011), we do not suggest a merger of Rhipsalidopsis with Schlumbergera. First, our data do not find a close relationship of these two. And second, it was already pointed out by several authors that Rhipsalidopsis and Schlumbergera only share vegetative characters but differ considerably in floral characters (e.g. Moran 1953). None of the characters unique for Schlumbergera is found in Rhipsalidopsis. The best taxonomic and nomeclatural conclusion from our results is recognizing Rhipsalidopsis again as a separate genus. It is characterized by flattened stem-segments, an angled pericarpel and large actinomorphic, campanulate pink or red flowers.
The genus is supported as monophyletic with 100% JK, 1.00 PP. Several considerably different generic concepts have been suggested for Lepismium (Table 1). It was either included into Rhipsalis (Schumann 1899, Vaupel 1925-1926) or recognized as monotypic for L. cruciforme (e.g. Britton and Rose, 1923). Backeberg
Chapter 3 (1959) proposed a very different generic concept based mainly on the sunken pericarpel, thus including many species of Rhipsalis, recognizing in total 17 species. Barthlott (1987a) established an altered Lepismium with 14 species and included the former Rhipsalis subgenera Ophiorhipsalis and Houlletia as well as Acanthorhipsalis, Lymanbensonia and Pfeiffera, based on the mesotonic branching as the main diagnostic character. Lepismium in this circumscription was found as polyphyletic and distant from the Rhipsalideae (Nyffeler 2002), and part of it is now treated as Pfeiffera Salm-Dyck and Lymanbensonia Kimnach (Korotkova & al. 2010). In our new circumscription, Lepismium contains 5 species and is characterized by mesotonic branching, indeterminate stem-segments, small, usually white flowers positioned laterally, angled pericarpels and naked fruits.
Rhipsalis is found as monophyletic and contains six lineages basically corresponding to the subgenera sensu Barthlott & Taylor (1995). Erythrorhipsalis is the only subgenus entirely confirmed as monophyletic by our data while subg. Epallagogonium is highly polyphyletic; its species being found in three different lineages. There are four species that do not “fit” in morphologically otherwise well defined clades but are rather morphological intermediates between the clade they are part of and another, more distant clade. These species are Rhipsalis pulchra, R. grandiflora, R. ewaldiana and R. sulcata. Their morphology might either be plesiomorphic, result from homoplasy or convergences, or to be the result of ancient hybridization events. However, no verifiable hybrids between Rhipsalis are currently known, and this hypothesis will have to be investigated using nuclear markers if firm evidence for hybridization in Rhipsalis is to be obtained. Subg. Calamorhipsalis K.Schum. (incl. subg. Epallagogonium K.Schum. p.p.) Subgenus Calamorhipsalis as defined by Barthlott & Taylor (1995) with R. hoelleri, R. neves-armondii and R. puniceodiscus is supported as monophyletic by 98% JK / 1.00 PP. Rhipsalis neves-armondii, which has strictly determinate stem-segments is sister to the pair of R. hoelleri and R. puniceodiscus. Both these species are similar, having indeterminate growth, but R. hoelleri differs in having red flowers. Rhipsalis floccosa, R. trigona and R. dissimilis form a well supported clade (100% JK / 1.00 PP) which is sister to Calamorhipsalis. These three species were referred to as the Rhipsalis floccosa group within subgenus Epallagogonium by Barthlott & Taylor (1995) and are characterized by stem-segments of determinate growth and strictly acrotonic branching, often woolly (floccose) areoles post-anthesis and repeatedly flowering areoles. They furthermore exhibit stem-dimorphism with juvenile segments bearing spines, especially in R. dissimilis, and the seedlings of R. floccosa show developmental phases which pass from ribbed, spiny and cereoid
Chapter 3 through triangular spineless stem-segments before the adult cylindrical segments appear, thereby resembling first R. dissimilis then R. trigona in its ontogenetic stages (N. Taylor, pers. comm.). The subspecies of R. dissimilis and R. floccosa sampled do not form separate clades, but are intermixed. Rhipsalis dissimilis f. epiphyllanthoides was originally described as Lepismium epiphyllanthiodes Backeb., then later regarded as just a form of Rhipsalis dissimilis (Barthlott & Taylor 1995). This form has a small distribution area and is clearly recognizable whereas forma dissimilis is more widespread and varies considerably, depending on its habitat. The two taxa prove to be very distinct in our study, forma dissimilis is part of a clade formed by R. floccosa subsp. pulviningera, subsp. floccosa, subsp. hohenauensis and R. trigona while forma epiphyllanthoides is close to R. floccosa subsp. oreophila and subsp. tucumanensis. Rhipsalis floccosa is widespread and has the second largest distribution area of all Rhipsalis after R. baccifera. It is a variable species with five morphologically different and geographically separated subspecies currently recognized (Hunt, 2006), most of them originally described as distinct species. Our data find R. floccosa as not monophyletic, but apparently forming a complex of closely related morphologically similar species, unless the complex as a whole is not considered as a single species. This alliance also included R. trigona which can not be separated from R. floccosa by DNA sequences although the adult plants are morphologically different. Our data reliably support an expanded subg. Calamorhipsalis, including the R. floccosa group. This circumscription partly corresponds to the original proposal of Schumann (1899), the group “Floccosae” of Vaupel (1925) and almost meets the one proposed by Backeberg (1959), as a subgenus of Lepismium. The subgenus as newly defined is characterized by mainly terete stems (trigonous in R. trigona), a sunken pericarpel, erumpent flower-buds and areoles that are often densely woolly postanthesis. Subg. Erythrorhipsalis Berger This subgenus was originally monotypic and based on R. pilocarpa, then treated as a subgenus of Rhipsalis (Barthlott 1987a), including more species and in a circumscription which is entirely confirmed by our data). Erythrorhipsalis is well defined by a characteristic habit with indeterminate basal extension shoots and subsequent stem-segments decreasing in size toward the branch apex, pendent, slender terete stems, campanulate flowers borne apically on the terminal or penultimate segments (subapically in R. pulchra) and directed downwards. Although relationships between its species could not be fully resolved, all species are found as distinct. Rhipsalis ormindoi, which is currently also included in Erythrorhipsalis, and also has the typical morphology of this subgenus, could not be sampled here.
Chapter 3 Subg. Epallagogonium K.Schum. This subgenus was originally established for R. paradoxa then later expanded to include further species with angular stems and sunken pericarpels. Rhipsalis paradoxa appears isolated within Rhipsalis. It is characterized by stem-segments of determinate growth with three to four discontinuous ribs/angles (i.e. each rib that is actually a podarium is shifted by c. 90° from the preceding one). But excepting its indeterminate stem-segments, R. paradoxa is very similar to R. pacheco-leonis which is almost like R. paradoxa in miniature, so R. paradoxa is morphologically not very distinct. Still, since our data show the subgenus Epallagogonium as polyphyletic, we suggest it should be circumscribed in the sense of Schumann to include only R. paradoxa. Subg. Goniorhipsalis K.Schum. (incl. subg. Epallagogonium p.p., subg. Rhipsalis p.p.) Rhipsalis lindbergiana, R. pentaptera and R. pacheco-leonis form a well supported clade (100% JK /1.00 PP). The two latter species were part of the R. pentaptera group of subg. Epallagogonium and are characterised by angular stems. Rather unexpectedly, Rhipsalis lindbergiana is also part of this grouping. It was believed to be closely related to R. baccifera and R. teres (and is even occasionally mixed up with these). But a closer examination of the plant’s morphology shows R. lindbergiana is indeed similar to R. pentaptera and R. pacheco-leonis and differs mainly by having terete stems. We assign R. lindbergiana, R. pacheco-leonis and R. pentaptera to an additional subgenus Goniorhipsalis, which had not been recognized by Barthlott & Taylor (1995). This subgenus as originally described by Schumann (1899) included R. pentaptera along with R. micrantha and R. trigona, with no type species indicated; R. pentaptera was later chosen as the type by Buxbaum (1970a). We therefore decided to resurrect Schumann’s infrageneric name for our newly found clade of R. pentaptera, R. pachecoleonis and R. lindbergiana. In this new circumscription, the subgenus is characterised by alternating podaria, reduced flowers borne perpendicular to the stem and welldeveloped scale-leaves. However, the differences to R. paradoxa are only ones of relative size of parts and there does not seem to be a single morphological character that absolutely distinguishes R. paradoxa from subg. Goniorhipsalis. Subg. Phyllarthrorhipsalis Buxb. (including subg. Rhipsalis p.p.) Subgenus Phyllarthrorhipsalis is supported as monophyletic (87% JK / 1.00 PP) but has to be expanded to include Rhipsalis grandiflora and R. ewaldiana. The entire subgenus Phyllarthrorhipsalis except for R. grandiflora and R. ewaldiana can be characterised by strictly determinate stem-segments and either angled or flattened stems (R. ewaldiana has additional indeterminate basal extension shoots). The angled
Chapter 3 stems are not restricted to any of its subclades and besides also occur in subgenera Calamorhipsalis, Epallagogonium and Goniorhipsalis (as newly circumscribed above), indicating that this feature is highly homoplastic within Rhipsalis. On the contrary, flattened stems are restricted to subg. Phyllarthrorhipsalis although they occur in several subclades within it, indicating that shifts to flattened stems are quite easy and happened several times. Besides, Phyllarthrorhipsalis differs in its seedling morphology: its species have flattened first stems as seedlings, whereas other Rhipsalis taxa observed have initially terete-ribbed seedlings even if subsequent ones are angled (Taylor & Zappi 2004). The placement of R. grandiflora as sister to the rest of the R. pachyptera-alliance is rather unexpected. It had originally been placed in subg. Rhipsalis based on similarities in stem morphology and flower bud development and is morphologically different from the other Phyllarthrorhipsalis in having terete stems, which do not occur in other taxa of this clade. Since all three specimens of R. grandiflora sampled occur in this position, the placement is unlikely the results of any artefacts. The numerous stamens of R. grandiflora and its ability to produce several flowers per areole tentatively indicate the relationship to Phyllarthrorhipsalis and the deviant morphology could also been explained by R. grandiflora being a hybrid. A clade of R. pachyptera, R. russellii, R. cereoides and R. agudoensis is found with maximal support. All four are morphologically similar and R. agudoensis has even been misinterpreted as an unusual form of R. pachyptera prior to its description. All four species grow semi-erect and have 3-5 ribbed, sometimes also flattened stem segments (R. pachyptera, R. russellii), often produce several flowers per single areole and have fruits that change their colour from white to pink. They are also found growing predominantly as lithophytic, not epiphytic. A clade of R. oblonga, R. crispata, R. cuneata and R. occidentalis, is supported by 97% JK and 1 PP and contains morphologically similar species with thin, flattened and leaf-like stems. This grouping contains geographically distinct species. While R. oblonga and R. crispata are native to Brazil, R. occidentalis and R. cuneata occur in the Andes, mainly in Bolivia and Ecuador. Rhipsalis micrantha and R. elliptica appear in a polytomy, although the two samples of R. elliptica are resolved in a distinct lineage. Rhipsalis elliptica is a flatstemmed species native to SE Brazil while R. micrantha is a widespread and morphologically very variable species that occurs in the Andes of Ecuador and Peru and extends into Central America. Its stem morphology ranges from narrow flattened or angular stems in the typical forma micrantha and especially in forma kirbergii to the more broadly flattened stems of forma rauhiorum. These three forms had originally been described as distinct species closely related to R. micrantha (Barthlott 1974) but were later interpreted as variations in different habitats. The question therefore is
Chapter 3 whether R. micrantha represents a species complex, or if there is a case of incomplete lineage sorting with R. elliptica in fact being derived from ancestral populations of R. micrantha or vice versa. More sequence data and a population-level sampling are needed to get further insights. The placement of R. ewaldiana within subgenus Phyllarthrorhipsalis as sister to R. goebeliana is unexpected. Rhipsalis ewaldiana has been regarded as closely related to R. mesembryanthemoides since both species share dimorphic stem segments with long and short shoots and exhibit partly mesotonic branching. Nevertheless, R. ewaldiana
Phyllarthrorhipsalis. Subgenus Rhipsalis As inferred here, subg. Rhipsalis is not monophyletic as circumscribed by Barthlott & Taylor (1995). One erstwhile subspecies of R. baccifera merit species rank, R. sulcata is additionally included and R. lindbergiana, R. grandiflora and R. ewaldiana have to be excluded (see discussion above). Rhipsalis baccifera and R. teres are the “typical” Rhipsalis with strictly acrotonic branching, terete stems, and a characteristic habit with indeterminate basal extension shoots (as in subg. Erythrorhipsalis) and small whitish flowers with few perianth segments. Both species are widespread, highly variable in morphology and numerous additional names at species and subspecies level have been proposed but are now regarded as synonyms. The R. teres specimens sampled form a clade supported by 60 % JK and 0.95 PP that also includes R. sulcata. The latter can not be recognized as distinct from R. teres based on plastid sequences. It is a poorly known species which had been placed within subg. Epallagogonium and regarded as closely related to R. pentaptera. Although the placement found by our data is unexpected, the plant's morphology does support it. Rhipsalis sulcata has stem-segments with strictly acrotonic branching and shares the habit of R. teres and R. baccifera with indeterminate basal extension shoots. The main differences are the slightly angled stems, which are, however, also sometimes developed in R. teres f. prismatica. A specimen from Costa Rica (C. Horich 4/88, vouchered at BONN) is resolved as sister to the rest of the R. teres-clade. It has to be investigated whether this taxon deserves at least subspecies rank and whether it may represent an alien introduction to the Costa Rican flora (the nearest naturally-occurring populations of R. teres are some 5,000 km distant in SE Brazil). Rhipsalis baccifera is the most widespread of all Cactaceae species and has been described under numerous synonyms. Currently six subspecies are recognized (Barthlott & Taylor 1995). The Rhipsalis baccifera specimens sampled, excluding
Chapter 3 subsp. shaferi, form a moderately supported clade (0.82 PP), only found by BI including indels. Rhipsalis baccifera subsp. shaferi is resolved as sister to the rest of subg. Rhipsalis, indicating it should be treated as a distinct species. It was indeed originally described as R. shaferi and is geographically distinct from subsp. baccifera, ranging through Paraguay, southern Bolivia and northern Argentina to São Paulo state, SE Brazil, and is replaced in northern Bolivia by R. baccifera and in Brazil by R. teres. It also differs morphologically from the rest of the subg. Rhipsalis by having indeterminate stem segments. Rhipsalis baccifera and R. teres may not be exclusive lineages, respectively, and what is known under these names is a complex of very similar taxa. More sequence data and a manifestly larger taxon sampling as well as population-level studies are needed to reliably infer species limits. The Old World Rhipsalis The occurrence of Rhipsalis baccifera in Africa, Madagascar and Sri Lanka has puzzled taxonomists and biogeographers for more than 100 years. These plants have been considered to be Gondwanan relicts (Croizat 1952) or in the other extreme as recently introduced by man (Buxbaum 1970a). The most commonly accepted hypothesis, however, was dispersal to Africa by migratory birds, early in the evolutionary history of Rhipsalis baccifera (Backeberg 1942). There are more examples of taxa of an exclusively New World family occurring in tropical Africa. One species of Bromeliaceae, Pitcairnia feliciana has a small distribution area in West Africa (Porembski & Barthlott 1999). Its dispersal from South America to Africa has recently been estimated to have happened around 10 Mya (Givnish & al. 2007). A similar figure of c. 6 Mya has been estimated for the dispersal of Maschalocephalus dinklagei (Rapateaceae) to Africa (Givnish & al. 2004). Recently, an age of 19.1 – 3.1 Mya has been inferred for the Cactaceae (Ocampo & Columbus 2010). Although no timeframe for the dispersal of Rhipsalis baccifera to Africa was inferred, this age estimate is comparable to the figures as quoted above. The African Rhipsalis baccifera populations differ from their New World relatives in gross morphology, ploidy level, anatomical characters and pollen morphology (Barthlott 1983). The two African specimens sampled here (subsp. erythrocarpa from East Africa and subsp. horrida from Madagascar) are depicted as sisters with high confidence (91%JK, 1.00 PP) within the grouping of South American Rhipsalis baccifera specimens. The divergent sequences of the two specimens sampled here provide another evidence for a long independent evolution of these populations thus arguing against a recent introduction to Africa by man.
3.4.3 The potential of markers for OTU identification within Rhipsalideae The underlying principle of DNA barcoding is that a priori defined species are recognizable by specific DNA sequences (e.g. Hebert & al. 2003). However, there is yet no standardized approach how to distinguish species by DNA sequences and how many sequence characters of a given set of markers will be needed for unambiguous recognition. The accuracy, i.e. the ability of a barcode to identify a species correctly will be highest if it does not only distinguish randomly chosen species or species that occur in single geographical settings (e.g. plots) but provides enough variation to separate closely related species. Nevertheless, it does not seem appropriate to use a generally applicable threshold value for distinguishing sister species due to varying degrees of sequence divergence resulting from rate heterogeneity of markers and lineages. Meyer & Paulay (2005) further argued that thresholds would result either in false positives or false negatives, as there is no discontinuity between intraspecific and interspecific sequence divergence. Therefore, some authors use an approach in which infraspecific p-distances must be smaller compared to interspecific ones (Lahaye & al. 2008, CBOL Plant Working Group 2009). Others simply regard a taxon as unique if it does not share its sequence with any other taxon in the sampling, e.g. Seberg & Petersen (2009). On the other hand, DNA sequences are also useful to evaluate if morphologically similar individuals belong to a species, thereby evaluating alphataxonomy or searching for cryptic or otherwise unrecognized species. Likelihood methods were developed recently that determine the point of transition between population level evolutionary processes and stochastic lineage growth (Pons & al. 2006, Fontaneto & al. 2007, Monaghan & al. 2009). Such methods were also applied in angiosperms to test monophyly of species (Lahaye & al. 2008). On the other hand, extant patterns of angiosperm species diversity, including those of cacti, may involve considerable incomplete lineage sorting (e.g. Jakob & Blattner 2006) or reticulate evolution (e.g., Sang & al. 1997). Complex, multi-faceted approaches are therefore needed to assess and later identify Cactaceae species using molecular markers. In our study, we focus on the molecular evaluation of OTUs that were a priori defined using morphology (Appendix 5). Being well studied and completely available in cultivation, we assume that carefully defined OTUs of Rhipsalideae will already closely match species in most cases. As one of the facets of the above described approach we analyze the species (= OTU) identification potential of a wide spectrum of plastid markers. So far, comprehensive comparative sequence data sets for taxonomically fully sampled lineages of plants are hardly available. In addition, we will discuss situations where OTUs appear not be monophyletic to guide future research on species limits using nuclear sequences and population level sampling.
Figure 3.3 Results from OTU-identification test: percentage of identified OTUs from single markers and all possible combinations.
Chapter 3 Using our approach, maximally 59 out of 61 (97%) of all OTUs could be successfully identified (Figs. 3.3 and 3.4) using a maximum number of sequence characters. The main trend was that all morphologically well recognizable OTUs had distinct sequences as well and appeared as monophyletic in the phylogenetic tree. In contrast, those species which can not be easily separated by morphological features or are morphologically variable, were either not easily resolved by our sequence data or intraspecific sequence variation was observed (1-3 mutations within OTUs). The lineages of Hatiora salicornioides and of Rhipsalis baccifera, R. floccosa, R. teres and probably R. micrantha (individuals of Rh. micrantha lack resolution to Rh. elliptica, although the latter share potential synapomoprhies) are paraphyletic to other morphologically recognizable taxa (Hatiora cylindrica, the African subspecies of R. baccifera, R. dissimilis & R. trigona, R. sulcata). This identification success is higher than observed in other barcoding studies that used a taxonomic setting. Hollingsworth & al. (2009) used seven loci (rpoC1, rpoB, rbcL, matK, psbA-trnH, atpF-atpH, psbK-psbI) but could identify only 69% species of Inga (Fabaceae), and 32% of Araucaria. Seberg and Petersen (2009) found that even six regions (ndhF, matK, psbA-trnH, rps8-pl36, accD, rpoC1, c. 4500 nt per sample) were not sufficient for discriminating more than 92% of Crocus species. In contrast, already c. 2500 nt of four highly performing regions used here (rpl16 intron, trnK intron, psbA-trnH, trnQ-rps16) were sufficient to identify 97% of the OTUs.
% identified OTUs 100 2536 2677 2249
90 1493 1352 1340
1184 1196 1194
440 428 1053 1038
2726 2714 2583
4066 4207 3023 3169 3310
50 299 897
average nucleotides sequenced
Figure 3.4 Percentage of identified OTUs in relation to the number of sequenced nucleotides.
Chapter 3 Among all possible marker combinations this one was the most successful with the least number of sequenced nucleotides; the combination of all markers (4207 nt) yielded the same identification success. Other marker combinations with a comparable number of nucleotides, however, often resulted in lower identification success (Fig. 3.5) – for example, the combination of matK, trnK intron and psbA-trnH (2714 nt) which found only 48 (79%) OTUs. The rpl16 intron was the best single-locus barcode, identifying (38 OTUs, 62%, Appendix 3), followed by matK (36 OTUs, 59%, Appendix 3). The rpl16 intron has not yet been suggested as a barcode but is frequently used in phylogenetic studies at low taxonomic levels. Our results provide evidence that it is not only a powerful phylogenetic marker but should seriously be considered also an effective barcode. The matK gene or a part of it has been repeatedly suggested as plant barcode and its good performance has been corroborated by recent studies (Chase & al. 2007, Little & Stevenson, 2007, Lahaye & al. 2008, Ford & al. 2009). Remarkably, matK alone found even more OTUs than the trnK intron (Appendix 5), although the intron is more variable (Table 2). When looking at single data partitions, each matK and trnK also identified some OTUs uniquely. We have additionally compared the identification success of the entire matK CDS with a part of the gene. This corresponded to the c. 950 bp fragment (partial matK) proposed by Lahaye & al. (2008) as a universal plant barcode. However, partial matK finds only 46% while the entire gene finds 59% of the OTUs, and is therefore more successful. PsbA-trnH has been regarded as one of the most promising angiosperm barcodes (Kress & al. 2005, Cowan & al. 2006, Chase & al. 2007, Kress and Erickson, 2007). In our study it identified only 54% of the OTUs (Fig. 3.4, Appendix 3). The trnQ-rps16 spacer has recently been demonstrated as a good barcode for Paeonia (Zhang & al. 2009) or Psiguria (Steele & al. 2010) but was among the less effective barcode regions with a performance comparable to that of psbA-trnH.
3.4.4 Phylogenetic utility of the regions used None of the single partitions yielded fully or even nearly fully resolved trees. Only the combined dataset of trnK/matK, rps3-rpl16, rpl16 intron, psbA-trnH and trnQ-rps16 provided sufficient resolution and good support. The combined dataset provided not only high resolution, even at species level, but also yielded a highly supported tree with 46 of the 86 nodes gaining JK values higher than 95% in parsimony analyzes. Posterior probabilities from Bayesian analyses were higher; out of 89 supported nodes, 75 have a PP>0.95 and 63 nodes are maximally supported. The two best-performing markers in our dataset were the group II introns in rpl16 and trnK, and this is another peace of evidence for the high phylogenetic performance of GII introns – regardless of taxonomic level – as pointed out by (Borsch &
Chapter 3 Quandt, 2009). Chloroplast spacers were found least informative, in congruency with the results of Löhne & al. (2007) who also found introns to perform better than spacers. Most of the nodes found by an analysis of the combined dataset were also resolved by rpl16, albeit with lower support (Appendix 6). The rpl16 intron also had the highest phylogenetic structure (Table 3.2). Resolution and support from trnK/ matK was comparable to rpl16, although fewer backbone nodes were found and support was lower. The high phylogenetic structure R in rpl16 and trnK/matK as compared to other chloroplast genomic regions has also been observed by Löhne & al. (2007) in Nymphaeales. However, the trnK intron and matK gene differ in their phylogenetic structure and partial matK showed higher phylogenetic structure R that the entire gene when compared directly to each other (Table 3.2). The tree based on the entire gene was better resolved and supported compared to the one from partial matK (Appendix 6 and 7). This could be explained by the different degree of conservation: the 3’ part of the generally fast evolving gene is fairly conserved while the 5’ region is less conserved (Hilu & Liang, 1997) and therefore different parts of the gene may yield different levels of phylogenetic signal or signal directed towards other parts of the tree. The psbA-trnH spacer was the most variable region in our study (Table 3.1). It showed higher phylogenetic structure R compared to matK and trnQ-rps16 but the parsimony tree derived from matK was much better resolved and supported than from psbA-trnH that is very short in Cactaceae (Table 3.1). When analyzed separately, psbA-trnH just yielded a large unsupported polytomy. The inferiority of the phylogenetic performance of psbA-trnH compared to other markers (e.g. matK, trnL-F, ITS) has previously been noted (Sang & al. 1997, Kim & al. 1999) and corresponds to our result here and our recent experience in a study of the genus Pfeiffera (Korotkova & al. 2010). The trnQ-rps16 spacer has hitherto been hardly applied in phylogenetics but was proposed as promising for low taxonomic level studies by Shaw & al. (2007) based on a high percentage of potentially informative characters (PICs). It was successfully applied for Genlisea (Lentibulariaceae) (Fleischmann & al. 2010) and Apiaceae subfamily Saniculoideae (Calviño & Downie, 2007), where it indeed provided a high number of informative characters, but trees based on single markers were not discussed therein. Compared to these lineages, trnQ-rps16 is much shorter in Rhipsalideae (mean length 300 nt vs. 576 and 1370 nt), thus the amount of potentially informative characters is limited.
Chapter 3 T T T G A A A A T G A T
T A C T T T T A C T A
ΔG = -10.1 R. baccifera
A A A A T A G C A T A T A T A T T A G C A T T A
ΔG = -8.1 R. teres
Inversion No. 1 (matK CDS) T T T T T T T
T T G AT
AC G T A A G T A A A
C T T A T T C GC T C A T T C A T T T
ΔG = -5.0 H. herminiae
A T C AA G A A A A A T A A A A GG T A AC G G C T A A T A T G C T A A T A T A T
ΔG = -6.1 H. salicornioides
Inversion No. 2 (trnK intron) C T T
A T AC A A G A A A
A A T T A G T T C T T T
ΔG = -2.8 L. cruciforme
A T T A A T C A A G A A A
G A A T A GT T T C T T T
ΔG = -3.1 L. lorentzianum
Inversion No. 3 (psbA-trnH) AA T A C A T T T A T A A TA A T A C G T A T A T A T A T A T T T A T A T A T A
ΔG = -5.4 R. gaertneri
TA TT T G A A T A T A TA T TA T C G T A T A T A T A T A T A T A T A T A T A
ΔG = -8.0 S. russelliana
Inversion No. 4 (trnQ-rps16)
Figure 3. 5 Inversions found a) in the trnK intron, b) in the matK CDS, c) in psbA-trnH and d) in trnQ-rps16 plotted on the parsimony consensus tree of the Rhipsalideae. Inverted states shown on the right, assumed plesiomorphic states on the left.
Chapter 3 We observed inversions in all markers used with the exception of rps3-rpl16 and the rpl16 intron. The reconstruction of the original and inverted states of these inversions on the parsimony tree showed most of them to be homoplastic (Fig. 3.5). All affect the terminal loops of hairpins. This is the most common pattern for small inversions in the plastid genome (Kelchner and Wendel, 1996; Kelchner and Clark, 1997; Borsch and Quandt, 2009). Such hairpin-associated inversions have already been shown to switch easily, even at population level (Quandt et al., 2003; Quandt and Stech, 2004). Inversions are known to be a problem in phylogenetic analyses. They will influence phylogenetic signal if overlooked in the alignment (Quandt et al., 2003). Perhaps the most severe potential problem, at least in Cactaceae, is the inversion in the matK CDS. It was also observed in other Cactaceae genera (Korotkova et al., 2010) and is highly homoplastic and the two states switch within OTUs. The matK region has to be checked carefully despite of its coding nature since other variable inversions were found for example in Amaranthaceae - Gomphrenoideae (Borsch et al. 2011). Phylogenetic studies in Cactaceae all have shown a comparatively low sequence variation of the markers used and most authors combined at least two regions. The rpl16 intron as a sole marker for the Cacteae resulted in a largely unsupported tree (Butterworth & al. 2002). A combination of the rpl16 intron and psbA-trnH for Mammillaria still did not provide much better resolution (Butterworth and Wallace, 2004). Improved resolution for closely related species was obtained from the rpl16 intron and trnL-F within Peniocereus (Arias & al. 2005). Within Pereskia, only a combination of five regions (psbA-trnH, trnK/matK, rbcL, phyC and cox3) could clarify the relationships (Edwards & al. 2005). A combination of three chloroplast spacers (atpBrbcL, trnL-F and trnK-rps16) for Rebutia and allied genera could identify clades within the genera but did not produce full resolution at species level (Ritz & al. 2007). In our recent study of Pfeiffera, only a combination of trnK/matK, trnS-G, rps3-rpl16, the rpl16 intron, trnQ-rps16 and psbA-trnH provided full resolution between all species (Korotkova & al. 2010). A comparison of our results and former studies within Cactaceae leads to the conclusion that trnK/matK and rpl16 are among the best performing regions within Cactaceae and should be considered as routine markers in future studies, whereas psbA-trnH and trnQ-rps16 cannot be recommended. The usage of psbA-trnH and trnQ-rps16 also has practical limitations: relative to their shortness, both required high sequencing efforts. Obtaining the whole sequence of the spacers with one primer was possible only in an estimated 30% of the taxa; usually two primer reads were necessary because of large homonucleotide stretches. The occurrence of such homonucleotides is also a putative problem for barcoding, as pointed out by Devey & al. (2009).
3.4.5 Comparison of phylogenetic utility and species identification potential of the markers used Similar to phylogenetic utility, identification utility depends on the mutational dynamics of the genomic region at hand and the amount of mutations per sequenced nucleotide. A major difference to phylogenetic utility is that patterns of homoplasy matter a lot when tree reconstruction is the goal. Our data suggest that introns and spacers outperform coding genes in phylogenetic utility, but not in identification utility. The ranking of markers according to their phylogenetic structure (based on identical numbers of characters) is rpl16 intron > trnK intron > psbA-trnH > trnQrps16 > partial matK > complete matK. The best-performing regions are not necessarily those that provide the largest percentage of variable characters; the percentage of variable and informative characters is in fact low in trnK/matK and that of rpl16 is comparable to psbA-trnH and trnQ-rps16 (Table 3.1). Regarding species identification potential, the ranking would look different: rpl16 intron > matK > psbAtrnH > trnQ-rps16 > trnK intron > partial matK. It is interesting that, apart from the different ranking, the best performing phylogenetic marker in our study is also the most successful single-locus species identifier. But apart from this, it seems that levels of variability do not necessarily correlate with phylogenetic signal, since the most variable regions do not provide the highest phylogenetic structure.
3.4.6 An improved classification system for Rhipsalideae Our study has provided a robust framework for a phylogeny-based classification of the Rhipsalideae. Several taxonomic and nomenclatural changes are proposed, as summarized in the following. Since Hatiora was found as polyphyletic, the name should only be applied to the former Hatiora subgenus Hatiora. Subg. Rhipsalidopsis should be recognized again at the genus level, following the “classical” circumscription that includes only R. rosea and R. gaertneri. Furthermore, Hatiora epiphylloides needs to be included into Schlumbergera (necessary new names and combinations are provided below). Within Lepismium, an altered circumscription results from the exclusion of L. incachacanum, which is now part of Lymanbensonia. Subgeneric limits within Lepismium also need to be re-defined. Our data support to recognize subgenus Ophiorhipsalis with its only species L. lumbricoides, but neither confirm subg. Houlletia nor subg. Lepismium as natural groups. We therefore propose uniting L. cruciforme, L. houlletianum, L. warmingianum and L. lorentzianum into subgenus Lepismium and keeping subg. Ophiorhipsalis with L. lumbricoides. Within Rhipsalis, slightly altered subgeneric circumscriptions are proposed for all subgenera but Erythrorhipsalis (see discussion above). Most changes should be made for subgenus Epallagogonium, as its species are found in three Rhipsalis clades. It should to be split and only circumscribed to contain the type species R. paradoxa while the rest is transferred an expanded subg. 85
Chapter 3 Calamorhipsalis, subg. Rhipsalis and the resurrected subgenus Goniorhipsalis. Subgenus
Phyllarthrorhipsalis. This revised classification is also shown in Fig. 3.2. Rhipsalis baccifera subsp. shaferi merits species rank. Its old name Rhipsalis shaferi Britton & Rose can easily be reinstated. A complicated case is the R. floccosa / R. dissimilis alliance where the gross-morphology does not correspond with the molecular phylogeny. A possibility derived from the phylogenetic hypothesis and the OTU recognition analyses would be species ranks for all R. floccosa subspecies and the two R. dissimilis forms, most of the names even already exist. Alternatively all the subspecies/forms could be merged into a much expanded R. floccosa. This would be in line with their ontogenetic stages that resemble each other. But this would likely make taxa of this complex hard to identify because many intergrade in their morphological characters. For the time being we do not propose any nomenclatural changes for this complex. We feel that more detailed studies of this species complex would be needed, sampling more populations or studying the ontogeny in more detail. Altering the formal taxonomy too early might result in taxa that can not be identified easily except with sequence data. There is not even any clear geographical pattern to be observed within the complex and it is not known whether the taxa of this complex interbreed or not.
Chapter 4 Morphology and character evolution of the Rhipsalideae Summary Only few character surveys exist for the Rhipsalideae, apart from compilations of characters in taxonomic treatments. A reconstruction of character evolution in a phylogenetic context is also lacking. Especially hypotheses on characters associated with the epiphytic life-form and the floral traits are missing. Synapomorphies for clades that are formally described as genera or subgenera also still need to be found. The well resolved phylogenetic hypothesis for the Rhipsalideae now enables a detailed study of character evolution. A matrix of 36 characters was compiled and the evolution of these characters was reconstructed on the phylogenetic tree using a Bayesian approach and ACCTRAN and DELTRAN optimization schemes. Epiphytism is reconstructed as crown group synapomorphy of the Rhipsalideae and epilithic and terrestrial growth are found to be reversals or further shifts. The Rhipsalideae are supported by several synapomorphies some of which are adaptations to the epiphytic life-form, such as the thin terete stems and the shrubby, pendent habit. The ancestral flowers of the tribe were reconstructed as actinomorphic, small, with free perianth segments, and not intensely coloured. Innovations in floral characters are zygomorphy, adaptations to bird-pollination, decrease in flower size, reflexion of the perianth and prominent stamen exposure. The degree of homoplasy is high, especially concerning vegetative characters. Reversals are also common. Many characters used to define genera and subgenera in the past are homoplastic. But several characters are homogenous within the respective clades and therefore can be used as diagnostic. So as a result, all the highly supported clades found by the molecular phylogenetic analyses can be defined morphologically.
4.1 INTRODUCTION According to the current knowledge, epiphytism has evolved four times in the Cactaceae (Chapter 1 this study, Hernández-Hernández & al. 2011, Wallace & Gibson 2002). The colonization of the tree canopy as new habitats went along with ample changes in morphology and there are several characters shared by all the epiphytic groups. These morphological shifts include the formation of flattened or terete stems in contrast to multi-ribbed stems in the terrestrial cacti. The stems of the epiphytes are much thinner; spines are mostly absent, inconspicuously developed or bristle-like. Adventitious roots occur in many Cactaceae genera but are especially frequent in creeping terrestrial species and in the epiphytes; some epiphytes can exhibit an exclusively adventitious root system. All epiphytic groups vary in flower morphology: small whitish flowers are found in all groups and at the same time also intensely coloured flowers (red, yellow, pink, magenta), likely bird pollinated. The flowers of some of the epiphytic genera are smaller compared to many terrestrial cacti. Those of Rhipsalis and Pseudorhipsalis are even among the smallest in the whole family. The Rhipsalideae are one of the two largest epiphytic tribes. All molecular phylogenetic studies (Bárcenas & al. 2011, Nyffeler 2002, Hernández-Hernández & al. 2011) resolve the Rhipsalideae as the sister group of a diverse and speciose clade of South American columnar cacti, the tribes Trichocereeae, Browningieae and Cereeae (BCT clade). While the earlier studies yielded only moderate support for the node of the Rhipsalideae+BCT clade (72% BS support in Nyffeler’s study, 61 ML BS support in the Hernández-Hernández & al. study), the most recent phylogenetic study based on trnK/matK provides 0.99 PP for this node (Bárcenas & al. 2011). It appears that the Rhipsalideae are phylogenetically isolated within the Cactaceae, and they are also morphologically very different from their sister group.
4.1.1 Characters applied as diagnostic for taxonomic groups in the Rhipsalideae One of the main characters used to define genera and subgenera was stem morphology – whether the stems are flattened or terete or ribbed was considered significant by all authors. The presence of spines was also considered significant. Floral characters played a key role, mainly the floral symmetry (actinomorphic vs. zygomorphic), the position of the flowers (lateral vs. apical), and the size and coloration of the flowers were considered significant. The taxa with small whitish or white flowers (Rhipsalis, Lepismium) were usually separated from those with larger and coloured flowers (Schlumbergera, Rhipsalidopsis). The presence or absence of a floral tube was sometimes used as a diagnostic character. Another character often considered significant is linked to the development of the areoles. In some Cactaceae species, most commonly in Rhipsalis, the areoles are sunken into the stem tissue. The flowers 88
Chapter 4 develop within the stem and burst through the stem epidermis where the areole would normally be. The ovary is sunken into the stem tissue so that both form a unit termed the pericarpel. The actual ovary is not visible. This sunken pericarpel is easily observable and was therefore also used as a diagnostic character. The interpretation of the morphology of the Rhipsalideae in the past was usually linked with taxonomic treatments of the group. Although the authors were probably implying that the characters of a given group were of common origin, hardly any clear statements about assumed character evolution were made. The first comprehensive Cactaceae monograph was provided by Schumann (1899). The Rhipsalideae subgenera were also first established therein. The characters he considered significant were the sunken vs. superficial pericarpel, the stem, ribbed or flattened and also the presence or absence of spines. Britton & Rose (1923) also emphasized stem morphology, the flower position (lateral or terminal) and the presence or absence of spines. They also considered the flower shape and size significant: their Rhipsalideae (or rather Rhipsalidinae) contained only the small flowered epiphytic species. The species with large, coloured flowers and flattened stems (Schlumbergera, Rhipsalidopsis) were part of a separate subtribe, the Epiphyllanae. The first hypotheses on common ancestry of characters were provided by Berger (1926). He attempted to define groupings within the Rhipsalideae based on assumptions of common origins of characters. Berger regarded the stem morphology, the position of the flowers and the floral symmetry as the most important characters. He assumed the putative ancestor of the Rhipsalideae had thin, terete stems that were retained in some Rhipsalis, in Hatiora and in Erythrorhipsalis. In contrast, he assumed flattened stems to have evolved twice in Rhipsalis and Rhipsalidopsis + Zygocactus / Schlumbergera and Lepismium + Pseudorhipsalis + Acanthorhipsalis. Remarkably, Berger did not mention taxa with angled stems, although most of them had already been described. Berger for the first time examined the funiculi and pointed out their potential diagnostic value. He noted that these of Pfeiffera were long-stalked and branched, while those of the Rhipsalideae are short-stalked and unbranched and this could be a character separating Pfeiffera from the Rhipsalideae (see Chapter 1). Some further assumptions on character evolution are found in Backeberg’s works, even though his approach was generally phenetic. Backeberg (1959) also emphasized the position and morphology of flowers. He considered the “Rhipsalides” with their small flowers with a reduced hypanthium and short funiculi to be the most “ancient” group. The tendency to smaller flowers, a reduced hypanthium and simple funiculi he recognised in the majority of the Rhipsalidineae. In contrast, the comparatively large zygomorphic flowers with a perianth tube (e.g. Schlumbergera, Zygocactus) were considered as more derived (Backeberg 1959). 89
Chapter 4 Although all authors used basically the same characters to define genera and subgenera, they often came to different conclusions or points of view as to which the most significant characters were. The sunken pericarpel was used by Schumann (1899) and later by Barthlott & Taylor (1995) to define Rhipsalis subgenera Calamorhipsalis and Epallagogonium (that included Schumann’s Calamorhipsalis). Backeberg (1959) based Lepismium on the sunken pericarpel while Barthlott (1987) based Lepismium on the mesotonic branching. The prominent bristle-spines of Rhipsalis pilocarpa were interpreted as a character that separates this species from other Rhipsalis and it was placed in a monotypic genus Erythrorhipsalis by Berger and later combined into Rhipsalis subgenus Erythrorhipsalis that was based on apical flowers (Barthlott 1987). Schlumbergera had been based on the actinomorphic flowers while Zygocactus was based on zygomorphic flowers. Epiphyllanthus was treated as a separate genus because of the well developed spines. The fact that it has the same flowers as Zygocactus was not considered (McMillan & Horobin 1995). Hunt (1968) pointed out that all share stamens arranged in two series and erect, connivent stigmas, which in combination with the zygomorphic flowers became the new diagnostic characters of an expanded Schlumbergera. A reconstruction of character evolution in a phylogenetic context is still lacking. Thus synapomorphies for clades that are also formally described as genera or subgenera still need to be found.
State of knowledge on morphological characters and earlier character surveys Apart from compilations of characters in taxonomic treatments of the
Rhipsalideae and the genera, few character surveys exist. The gross morphology, including the vegetative characters, the flower and fruit characters are well covered in many of the taxonomic treatments. Among the most detailed literature sources are the studies of Buxbaum. He undertook a detailed examination of some areole characters, especially of the composite apical areoles (Buxbaum 1942). He also provided many very detailed listings and drawings of vegetative, floral, fruit and seed characters at the generic level, including hypotheses on the homology of these characters in the Rhipsalideae genera studied (e.g. Buxbaum 1970a, b, c). Further characters were discussed in the “Morphologie der Kakteen” (Buxbaum 1957-1960). A survey of seed characters is available for the Cactoideae (Barthlott & Hunt 2000) but therein, seeds of only 9 Rhipsalideae species were analysed. Pollen characters of the Rhipsalideae were first studied by Leuenberger (1976) in the context of a survey of pollen morphology of the Cactaceae. Barthlott & Rauh (1977) have studied the pollen of Schlumbergera and pollen morphology of all Rhipsalideae was analyzed in a diploma thesis (Binski 2002, unpublished), carried out at the Nees Institute in the working group of T. Borsch. 90
Chapter 4 There are several studies of Rhipsalideae stem anatomy, starting with a first survey of Vöchting (1873). However, all these studies included only very few species. Dettke & Milaneze-Gutierre (2008) characterised the stems of seven Cactaceae epiphytes: Epiphyllum phyllanthus, 3 Lepismium species, 3 Rhipsalis species, and Hatiora salicornioides. The authors suggested that the anatomical characters had taxonomic value and would be useful for separating species. Calvente & al. (2008) also provided a survey of anatomical, especially epidermis characters. They examined six Rhipsalis species aiming at the evaluation of the taxonomic relevance of these characters and concluded epidermis characters were useful to differentiate Rhipsalis species. A very detailed anatomical and crystallographic study of the Rhipsalideae has been made as part of a dissertation (Hartl 2000) carried out at the Nees Institute in the working group of W. Barthlott. The results are largely unpublished besides the survey of the generation of calcium oxalate crystals in all Rhipsalideae species (Hartl & al. 2003). They found a unique crystal type in Rhipsalis, which forms exclusively monoclinic calcium monohydrate crystals and besides found the crystal types useful for differentiation of genera. The first molecular phylogenetic analysis of the Rhipsalideae, focussing on Hatiora and Schlumbergera was published recently (Calvente & al. 2011). This study also included ancestral state reconstructions for six vegetative and floral characters which all had been or still are used in classification systems. Ancestral states were reconstructed for the flower symmetry, the presence or absence of a flower tube, the branching pattern, the stem growth (indeterminate or determinate), the stem shape and the flower colour. The study of Calvente & al. (2011) which appeared during the final phase of this dissertation yielded some first insights into the character evolution within the Rhipsalideae. Only six characters were included, so there are still numerous characters to be analysed. Also, some of the relevant nodes, including most of the nodes in Rhipsalis were unresolved. The well resolved phylogenetic tree based on a complete taxon sampling presented in Chapter 3 now provides the framework for the detailed study of character evolution in the Rhipsalideae. The aims of the survey presented in this chapter were first, to compile a detailed dataset of morphological characters for the Rhipsalideae and to infer whether the clades (genera, subgenera) found by the molecular phylogenetic analyses can also be characterised morphologically and which characters are synapomorphic for these clades. The second aim was to reconstruct the evolution of these characters, with emphasis on the characters associated with the epiphytic life-form and the floral traits.
MATERIAL AND METHODS
Taxon sampling Morphological characters were scored for all Rhipsalideae species that were also
represented in the molecular analysis, thus covering all but one species of the tribe. The phylogenetic hypothesis was used as a guideline to which infraspecific taxa should also be included in the morphological matrix. Some taxa currently ranked as forms or subspecies were found as distinct by the molecular analyses (e.g. the forms of R. dissimilis or Rhipsalis baccifera subsp. shaferi). These taxa were also included in the morphological matrix. Other infraspecific taxa were included because they differ in morphological character states, e.g. the forms of Rhipsalis micrantha. Browningia hertlingiana, Calymmanthium substerile, and Echinopsis aurea were included as outgroup taxa.
Morphological data A matrix comprising 36 characters listed in detail below was compiled. The
complete matrix is shown in the Appendix 8. The morphological data were obtained from own observations of the living plants in the Botanical Gardens Bonn, from literature data and from the original diagnoses. The main literature sources were McMillan & Horobin (1995) and Barthlott & Rauh (1975) for Schlumbergera, the studies of Buxbaum (1942, 1970a, b, c), the Rhipsalideae checklist of Barthlott & Taylor (1995) and the treatment of the eastern Brazilian cacti of Taylor & Zappi (2004). The terminology for characters and their states was adopted from the last two sources. Data on ploidy levels were available from the chromosome counts of Barthlott (1976). Some pollen characters were scored from Binski (2002). A detailed survey on the Rhipsalideae pollen will be the task of future studies.
Analysis of character evolution Characters were coded as categorical data with multiple states, and with
polymorphisms, if polymorphisms have been observed. The Bayesian majority-rule consensus tree was considered as the best approximation of the organismal phylogeny for the character reconstructions. Character state transformations were mapped using WinClada v. 0.9.9 (Nixon 2002), examining unambiguous transformations as well as accelerated (ACCTRAN) and delayed (DELTRAN) optimization schemes. Homoplasy was mapped by character states, i.e. only discontinuous states were mapped as homoplastic. Under parsimony, ancestral states were reconstructed using the parsimony model with unordered states and the “trace character history” option of Mesquite v. 2.74 (Maddison & Maddison 2010). Posterior probabilities for ancestral states were reconstructed using BayesTraits (Pagel & al. 2004). As polymorphisms are
Chapter 4 not allowed for calculations in BayesTraits, a modified and reduced version of the matrix with 20 characters was constructed (Appendix 8). Here, either only the predominant character states were scored or the coding was modified so that each of the states was coded separately. The modifications are described below in more detail. The trees used for the Bayesian character state reconstruction were obtained using BEAST 1.0 (Drummond & Rambaut 2003). The trees obtained before with MrBayes contained polytomies which produced error messages in BayesTraits so that using these trees for the character reconstruction was not possible. A sample of 500 trees from the BEAST run was extracted using a perl script (K. Müller, unpubl.) which generates a BayesTraits input file from BEAST or MrBayes output files. The nodes for which ancestral states were wanted, were added to the ancestral state reconstruction using the “AddNode” command. The chain was run for 5050000 generations and rate coefficients and ancestral states were sampled every 100 generations. The mean values of all the posterior probabilities found were afterwards calculated with Excel and illustrated as pie chart diagrams using TreeGraph 2 (Stöver & Müller 2010).
Modifications of the matrix for BayesTraits analyses The matrix was reduced to the most significant vegetative and floral characters
and included 20 characters (Appendix 8). Some of the characters were modified to remove polymorphisms, which are not allowed for BayesTraits analyses. These modifications are described in the following; all the other characters in the matrix were the same as for the ancestral states reconstruction under parsimony. The life forms were scored as separate characters and the different states were coded as follows: 1) Epiphytic growth: (0): not epiphytic, (1): epiphytic; 2) Epilithic growth: (0): not epilithic, (1): epilithic; 3) Terrestrial growth: (0) not terrestrial, (1): terrestrial. Only the predominant states of the habits were scored. The flower colours were reduced just to two states: (0): flower white or whitish, not conspicuously coloured, (1): flowers intensely coloured (bright yellow, orange, red, magenta). The following characters were removed from the matrix: Adventitious roots, stem diameter, hair, flower buds position, flower size, stamen colour, anther colour, pollen colour, stamen insertion, style colour, stigma shape, pericarpel, fruit colour, and chromosome numbers.
4.3 LIST OF MORPHOLOGICAL CHARACTERS AND THEIR STATES Life-form and main vegetative characters Growth form (0): tree-like, with a conspicuous woody trunk, (1): large columnar, (2): medium-sized to small columnar, (3): shrubby, (4): globular / barrel Life-form (0): terrestrial, (1): epiphytic, (2): epilithic Habit (0): erect, (1): sub-erect or semi-erect, (2): pendent, (3): creeping, (4): spreading, (5): arching The states apply to adult plants; many species are erect in their juvenile stage, then pendent. In these cases, only the state “pendent” was scored. Furthermore, only the predominant states were scored as transitions between the states are often observed. Branching pattern (0): mesotonic, (1): acrotonic (incl. subacrotonic), (2): basitonic Adventitious roots (0): absent/rarely developed, (1): present Stem-segments growth habit (0): indeterminate, (1): (strictly) determinate, i.e. after the primary stem segment reaches a certain, probably predetermined size/length, the growth stops and a new segment or a new order of segments begins to develop. (2): “Firework habit”: a special pattern of indeterminate basal extension shoots, and other segments decreasing in size towards the distal part of the plant, e.g. in subg. Rhipsalis, (3): “mixed”: primary axes indeterminate, lateral axes determinate (e.g. Rhipsalis mesembryanthemoides). Shedding of old segments (0): old segments not deciduous, (1): old segments deciduous = shed by well developed abscission zones at the joints
Stem form (0): ribbed (angled); with 3-5 ribs, (1): flattened (only 2 ribs), (2): terete, (3): 5-more ribs, (4): cladodes, i.e. flattened stems but resulting not from reduction of ribs but resembling the stem segments of Opuntia (only Schlumbergera opuntioides). Opuntia is the only Cactaceae genus besides the epiphytes with flattened stems joints, but they are of different origin compared to the flattened stems of the epiphytes. The cladodes of Opuntia result from flattened cylindrical stems, while the flattened portion of a flattened stem of an epiphyte is produced in the same way as a rib (Gibson & Nobel 1986). Therefore, this character is scored separately, not homologous to state 1. Podaria (0): absent, (1): present The podarium is a structure unique to the Cactaceae. It is a product of the fusion of the leaf base and the stem. The result is either a tubercle or, if all podaria are arranged longitudinally, the cactus ribs (Buxbaum 1937). Stem diameter The stem diameter is used here as an approximation of the degree of succulence. There are several ways for its measurement, as demonstrated for example in a recent study of Crassula (Jones & al. 2011). The degree of succulence is commonly defined as the “water content per unit area of surface” (Delf 1912). This first attempt measured succulence as the amount of water in grams per square decimetre (dm2) of a leaf. Alternatively succulence can be measured in grams of water per gram of plant tissue (von Willert & al. 1992). Categories for thickness of stems were defined based on the average diameters of terete or angled stems and the size and thickness of the flattened segments. (0): lowest: filiform terete stems ≤ 0,5 cm, or very thin cladodes, (1): low: filiform terete stems 0,6 – 1 cm, or thin flattened stems, (2): medium: thick angled or terete stems 1-5 cm or thick flattened stems, (3): succulent: thick angled stems +5 cm
Areoles, spines and hair Position and development of the areoles (0): all areoles superficial, never sunken, growing throughout the life-cycle of the plant, (1): all areoles sunken, also the apical areole, (2): areoles sunken, except the apical areole,
primordium/primordial scale Those areoles that are truly sunken into the cortical tissue are regarded and termed as sunken. The flower buds and new stem segments developing at those areoles burst through the stem-epidermis where the areole would normally be (termed erumpent). Areoles of Lepismium, however, appear sunken but develop in a different way 95
Chapter 4 (Buxbaum 1970). They are almost superficial at the beginning of their development and deepened later and therefore are not treated as homologous to the sunken areoles of Rhipsalis but instead termed “depressed”. Apical composite areoles (0): absent, (1): present (but sometimes hidden) Spines (0): absent or inconspicuous (1): present, well developed, stiff, (2): present, bristly In their juvenile stages, many Rhipsalideae bear spines which are reduced later. Here, only the presence or absence and appearance of spines on adult stems were considered. Sometimes there are also spines on some basal extension shoots of mature plants that are otherwise spineless. In this case spines were scored as absent. Trichomes (0): absent or not significantly developed, (1): dense wool
Flower characters Position of the flowers (0): lateral to apical, (1): only lateral, (2): only apical, at composite apical areoles (“terminal”). Although the flowers on apical composite areoles appear to be terminal, and are sometimes termed as such, they do not terminate the stem – stem growth continues from the composite areole. Therefore the term “apical” is preferred. Orientation of the flowers (relative to the surface of the ground) (0): random, not conspicuously oriented, (1): pendent or directed downwards (e.g. Rhipsalis subg. Erythrorhipsalis) Position of flower buds (0): oblique, (1): perpendicular, (2): aligned with stem-axis (e.g. Rhipsalis subg. Erythrorhipsalis, Schlumbergera) Number of flowers at a solitary lateral areole contemporaneously (0): one, (1): two or more flowers. This character only applies to the production of several flowers at a solitary lateral areole. Composite areoles often produce more than one flower.
Repeated flowering at one areole (0): areoles flower only once, (1): areoles flower repeatedly This character does not apply to repeated flowering at a collective areole but only at a single lateral areole. Floral symmetry (0): actinomorphic, (1): zygomorphic Perianth segments fusion (0): free, (1): fused, forming a tube Perianth segments curvature (0): not reflexed, i.e. partially expanded to patent, (1): reflexed Flower size (diameter or length if the flower is tubular) (0): very small (smaller than 1 cm), (1): small (1-3 cm), (2): medium-sized (4 – 6 cm), (3): large (+7 cm) Flower colour (0): white / whitish, (1): yellowish, (2): bright yellow, (3): pink / magenta, (4): red, (5): orange, (6): pale pink
Androecium and gynocecium Nectaries (0): unspecific, (1): disc, (2): nectar chamber Stamen / filament colour (0): white / whitish or cream = not conspicuously coloured, (1): coloured Stamens insertion (0): stamens inserted in one series, (1): stamens inserted in two series Style colour (0): white/whitish or cream = not conspicuously coloured, (1): coloured Stigma shape (0): stigma lobes spreading, (1): stigma lobes erect, connivent
Pericarpels and fruits Pericarpels (0): smooth, not angled, (1): angled, (2): ridged, (3): slightly or inconspicuously angled Fruit shape (0): longer than broad, (1): globose, (2): subglobose Fruit colour (0): white / whitish, (1): red, (2): pink, (3): yellow, (4): greenish, (5): dark red to almost black, (6): orange In some species, fruits are white at first then changing their colour to pink so that both colours can be observed at the same time. In this case, the colour of the ripe fruits was coded. Some other species, e.g. R. puniceodiscus have forms with differing fruit colours and consequently both states were coded in such cases. Those white fruits that have a reddish ring around the perianth scar were scored only as whitish.
Pollen characters Pollen colour (0): white / whitish or cream = not conspicuously coloured, (1): coloured (mostly yellow or red) Pollen size (average diameter) (0): small (< 40 μm), (1): medium-size (41–50 μm), (2): large (51–100 μm) Aperture numbers (3): 3 apertures, (6): 6 apertures, (9): 9 apertures, (1): 12 apertures Rhipsalideae pollen is uniformly colpate. Aperture number variation within species or sometimes individuals is common in the Rhipsalideae. Therefore all the observed states within a species were coded. It is not always possible to determine the number of apertures from SEM images, and especially 6-colpate and 9-colpate pollen cannot always be distinguished. In such cases, it was decided to score 6 colpi since Leuenberger (1976) reports 9-colpate pollen to be rare within Cactaceae. Chromosome number (2): diploid 2n=2x=22, (4): tetraploid 4n=4x=44, (6): hexaploid 6n=6x=66, (8): octoploid 8n=8x=88. All chromosome count for the Cactaceae so far yield a basic chromosome number of 11 and multiples of 11 in the polyploidy taxa (Arakaki & al. 2007, Cota-Sanchez & Wallace 1995, Das & al. 1999, Negron-Ortiz 2007, Pinkava & McLeod 1971, Pinkava & al. 1998, Ross 1981). So far no dysploid changes were observed. 98
4.4 RESULTS AND DISCUSSION CHARACTER EVOLUTION IN THE RHIPSALIDEAE 4.4.1
Synapomorphies of the Rhipsalideae Considering unambiguous character changes only, the epiphytic life-form, the
pendent habit, the thin stems, the absence of spines and trichomes are found as synapomorphic for the Rhipsalideae. The ACCTRAN optimization finds additionally the shrubby habit, the acrotonic branching, the terete stems, the small flowers and small pollen (< 40 μm diameter). The DELTRAN optimization finds the same characters except the acrotonic branching and the small pollen, but suggests the determinate
Apomorphic versus highly homoplastic characters The trees summarizing character states transformations are shown in Figs. 4.1 –
4.3 and the results of the Bayesian ancestral state reconstructions are shown in Figs 4.4 – 4.9. There is a strikingly high degree of homoplasy and also numerous reversals in character states. There are only 3 unambiguous apomorphic state transformations that characterise larger clades (Fig. 4.1). The non-“deciduous” stem-segments are observed only in Lepismium and are consequently found as a synapomorphy of this genus by all optimization methods (Fig. 4.1-4.3). The fruit colour changed to white or whitish in Rhipsalis (but other fruit colours are also found within Rhipsalis. More than one
Phyllarthrorhipsalis. Apart from these characters which changed only once, there are several characters which characterise a given clade but are also convergently found in one or two species outside it (compare also Table 4.1). The branching pattern was considered an informative character in the Rhipsalideae already by Barthlott (1987), especially to separate the mesotonically-branched Lepismium from Rhipsalis. Acrotonic branching is reconstructed as plesiomorphic within the Rhipsalideae (PP 0.78). Only those Rhipsalis that no longer develop the apical composite areoles (e.g. R. puniceodiscus, R. hoelleri) exhibit subacrotonic branching. Within the Rhipsalideae, mesotonic branching is
mesembryanthemoides and R. ewaldiana (but they are not sister species). The exceptional mesotonic branching in Lepismium results from the loss of the apical composite areoles. This is another feature characteristic for Lepismium, but also found in four Rhipsalis species.
Chapter 4 The flower morphology of Schlumbergera is exceptional within the Rhipsalideae. The flowers have two series of perianth segments with the inner segments fused and forming a perianth tube, synapomorphic for the genus. Tubular flowers are common in the Cactaceae, but the tube is commonly formed by the pericarpel, not by the perianth. Apart from Schlumbergera, a perianth tube is found only in Disocactus and Pseudorhipsalis
predominantly zygomorphic flowers (except S. russelliana); the rest of the tribe has exclusively actinomorphic flowers. The stigma lobes are erect and connivent, this is also exceptional. All these characters can be regarded as synapomorphies of Schlumbergera (e.g. Hunt, 1969). However, Hatiora epiphylloides that falls in Schlumbergera based on the sequence data, lacks all these synapomorphies (discussed in more detail below), thus causing difficulties for the character reconstruction because reversals for almost all the character states have to be assumed in this taxon. The fruits of Schlumbergera, Hatiora and Rhipsalidopsis are mostly longer than broad and obconic in shape; pericarpels are mostly angled. In contrast, the fruits of Rhipsalis are predominantly globose (spherical) or subglobose, barrel-shaped. Especially subgenus Rhipsalis is characterised by such fruits. Pericarpels are never angled in Rhipsalis and in Hatiora. Coloured fruits, usually pink, are found in many Rhipsalis species and are especially characteristic for the R. cereoides-clade and some Erythrorhipsalis. Lepismium is exceptional by having very dark red, almost black fruits.
Figure 4.1 Tree showing the unambiguous character changes within the Rhipsalideae. The apomorphic character state shifts are shown as black boxes, state shifts that occurred more than once are shown as white boxes. Character and state numbers correspond to the matrix in Appendix 4.1.
Figure 4.2 Character state changes within the Rhipsalideae as inferred from the accelerated optimization (ACCTRAN, using “fast optimization” option in winClada 0.9.9). The apomorphic character state shifts are shown as black boxes, state shifts that occurred more than once are shown as white boxes. Character and state numbers correspond to the matrix in Appendix 4.1.
Figure 4.3 Character state changes within the Rhipsalideae as inferred from the delayed optimization (DELTRAN, using “slow optimization” option in winClada 0.9.9). The apomorphic character state shifts are shown as black boxes, state shifts that occurred more than once are shown as white boxes. Character and state numbers correspond to the matrix in Appendix 4.1.
Figure 4.4 BayesTraits ancestral states reconstruction for the life-forms and the main vegetative characters. Posterior Probabilities are shown as pie chart sectors.
Figure 4.5 BayesTraits ancestral states reconstruction for the areole characters. Posterior Probabilities are shown as pie chart sectors.
Figure 4.6 BayesTraits ancestral states reconstruction for the flower characters (part I). Posterior Probabilities are shown as pie chart sectors.
Figure 4.7 BayesTraits ancestral states reconstruction for the flower characters (part II). Posterior Probabilities are shown as pie chart sectors.
Chapter 4 Besides the characters discussed above, most characters, including also most of those used to define genera and subgenera, evolved at least twice and should therefore be considered highly homoplastic. Determinate stem-segments are reconstructed as the predominant plesiomorphic state in the Rhipsalideae (PP 0.53). Shifts to indeterminate stem-segments happened 4 times: in Lepismium, in Rhipsalis subg. Goniorhipsalis, in the pair of R. hoelleri and R. puniceodiscus and also in R. pulchra (Fig. 4.3, DELTRAN optimization). The independent shifts to indeterminate segments are probably connected with the loss of composite apical areoles because both states occur predominantly together, subg. Goniorhipsalis being the only exception. The apical composite areoles were lost four times: in Rhipsalis puniceodiscus and R. hoelleri, in R. lindbergiana, in R. puchra (this one can produce apical composite areoles on rare occasions), and in Lepismium. The “firework habit”, a specialized pattern of indeterminate basal extension shoots and segments decreasing in size towards the distal part of the plant, evolved independently in Rhipsalis subg. Rhipsalis (except R. shaferi) and in subg. Erythrorhipsalis. All the stem forms likely evolved several times independently. Therefore, the same character state in different genera is often not homologous; for example the flattened stems of Schlumbergera and Rhipsalidopsis or of Schlumbergera and Rhipsalis. The highest posterior probability for any of the states plesiomorphic for the Rhipsalideae is 0.36 for terete stems (Fig. 4.4), and they are also found as synapomorphic for the tribe (Figs. 4.1-4.3). Terete stems also found as plesiomorphic in Rhipsalis (PP 0.89) and they predominate in the subgenera Calamorhipsalis, Erythrorhipsalis
Rhipsalidopsis, in part of Lepismium, Schlumbergera, except S. opuntioides and S. microsphaerica, and in Rhipsalis subg. Phyllarthrorhipsalis. While areoles are superficial in most Cactaceae and also in most Rhipsalideae, areoles that are sunken into the stem tissue are characteristic for Rhipsalis, although they are found in only some of its species (Fig. 4.2, ACCTRAN optimization). They seem to have evolved three times: in the Rhipsalis floccosa-group, R. paradoxa, and Rhipsalis subg. Goniorhipsalis (Fig. 4.3, DELTRAN optimization). The deepened/ depressed areoles evolved only in Lepismium, most likely in the subgenus Lepismium (Fig. 4.2, ACCTRAN, and PP 0.72), then lost in L. lorentzianum. The different positions and development of areoles, leading to sunken pericarpels has been regarded as informative in the past: Backeberg (1959) based his Lepismium almost solely on this character while Barthlott & Taylor (1995) regarded it significant for Rhipsalis subg. Epallagogonium. As noted above, the sunken pericarpels of Lepismium and Rhipsalis are not homologous because they develop in a different way (Buxbaum 1970) and the deepened areoles of Lepismium are unique and characteristic for part of it. The sunken
Chapter 4 areoles in Rhipsalis could either be considered homoplastic or as a synapomorphy of the genus, as suggested by the ACCTRAN optimization. While trichomes are either absent or inconspicuous in all the other Rhipsalideae, densely woolly areoles evolved independently the R. floccosa group and R. paradoxa (Figs. 4.1-4.3). The flower position is one of the characters often considered significant for the delimitation of genera. Indeed, especially the apical flowers define some clades. The clade of Schlumbergera, Hatiora and Rhipsalidopsis is characterised by flowers only at apical composite apical areoles (Fig. 4.1-4.3, PP 0.86). Within Rhipsalis, apical flowers are observed only in subg. Erythrorhipsalis (except R. pulchra) and are likely the results of a shift from lateral or lateral to apical flowers, which are found as plesiomorphic for Rhipsalis (PP 0.66 for lateral to apical, Fig. 4.6). This result confirms one of Barthlott’s earlier assumption (1987). He was the first to define Rhipsalis subg. Erythrorhipsalis by the apical, often campanulate flowers and they are not synapomorphic but still characteristic for this subgenus. In contrast, lateral or lateral to apical flowers occur in all other Rhipsalis subgenera. Different kinds of flower orientation characterise some Rhipsalis clades but evolved independently in each of them. Flowers oriented downwards occur in Rhipsalidopsis and Lepismium, Rhipsalis subg. Erythrorhipsalis, R. paradoxa, R. hoelleri and R. puniceodiscus. The flower orientation is not necessarily linked with the flower position - apical flowers are not necessarily directed downwards. Oblique flower buds are characteristic for Lepismium. This is due to the fact that the depressed areoles are themselves aligned obliquely to the stem axis. Oblique flower buds are also found in Rhipsalis subg. Calamorhipsalis p.p. (not conspicuous in the R. floccosa-group), also for Rhipsalis paradoxa and R. pacheco-leonis and R. pulchra, but the oblique or perpendicular flower buds are not linked with flower position or orientation. In contrast, all species with exclusively apical flowers also have flower-buds aligned with the stem axis. Areoles that grow throughout the plant’s life cycles and flower repeatedly are found throughout the Rhipsalideae and are likely plesiomorphic (PP 0.72). In contrast, one-time flowering is only found in the clade of R. puniceodiscus, R. hoelleri and R. neves-armondii, also in Rhipsalis subg. Erythrorhipsalis, subg. Rhipsalis and in R. lindbergiana. The ability for repeated flowering at one areole seems to be lost independently in these species. The largest pollen grains occur in Schlumbergera and Rhipsalidopsis and the reduction of the pollen grain size is a trend throughout the Rhipsalideae. But at the same time, there seem to be also several independent secondary increases in pollen size. There are different scenarios for the evolution of the pollen size. The ACCTRAN optimization suggests small pollen in all Rhipsalideae and then independent increases in Rhipsalidopsis and Schlumbergera, also in Rhipsalis subg. Goniorhipsalis. The DELTRAN optimization finds three shifts from medium-sized to small pollen in Lepismium, in Hatiora and in Rhipsalis, with secondary increases in Rhipsalis subg. 109
Chapter 4 Goniorhipsalis. A varying number of apertures is found in all Rhipsalideae genera except Schlumbergera, which is uniformly 6-colpate. Aperture number therefore is not informative within Rhipsalideae. Pollen with 3 colpi is still found within the first branching Rhipsalis clades (subg. Calamorhipsalis and Erythrorhipsalis) while the rest has higher aperture numbers, most commonly 6, which is found as a synapomorphy of the Rhipsalideae by the DELTRAN optimization. A further increase in aperture number is characteristic for subg. Rhipsalis that has more or less uniformly 6 and 12 colpi, and usually both states are observed within a taxon; and within subg. Phyllarthrorhipsalis some species also have 12-colpate pollen. There are no reversals from 6- or 12-colpate to 3-colpate pollen throughout the Rhipsalideae. The general pattern seems to be what is termed successiformy i.e. the increase of aperture numbers by doubling. This appears to happen frequently and independently.
Evolution of characters associated with the epiphytic life-form It is difficult to formulate hypotheses on the evolution of epiphytism in the
Rhipsalideae from a comparison with their closest relatives. The sister group of the Rhipsalideae, the BCT-clade is morphologically very different. Notably, there seems to be a slight but recurrent tendency for epiphytism in the BCT-clade: two Cleistocactus, one Samaipaticereus and two Echinopsis species are commonly found as epiphytes in Bolivia (Ibisch & al. 2000); Echinopsis arboricola is even an obligate epiphyte (Kimnach 1990). Using a Bayesian approach for ancestral states reconstruction, and sampling genera from all major Cactaceae clades, Hernández-Hernández & al. (2011) reconstructed the common ancestor of the Rhipsalideae and the BCT clade as an erect and ribbed, less probably barrel-like cactus (PP ribbed 0.99, erect 0.71, barrel-like 0.56). They suggest that the steps during the evolution of epiphytism therefore would have involved first a shift to shrubby habit in the ancestor of the Rhipsalideae and in the next step, the evolution of the pendent habit (PP shrubby 0.47, epiphytic 0.99, nonerect 0.99). The results of this study confirm this; the shrubby and pendent habit is reconstructed as synapomorphic for the Rhipsalideae, with even higher Posterior Probabilities (PP shrubby 0.99, pendent 0.78). It is therefore likely that the ancestor on the Rhipsalideae was a terrestrial plant that had the ability to grow epiphytic and finally shifted to fully epiphytic. Within the Rhipsalideae, most of the species are obligate epiphytes, epilithic growing species and also terrestrials are observed, but no hemiepiphytic species. The epilithic growth predominates in some clades or taxa: in Schlumbergera, Rhipsalis teres and especially in the R. cereoides-group but there are only two obligate lithophytes which are the two forms of R. dissimilis. All other species growing as lithophytes are also found as epiphytes. It appears there was no “transition” from terrestrials to lithophytes to epiphytes but rather a direct shift from terrestrials to epiphytes. However, the possibility of extinction at the branch leading to the Rhipsalideae must also be considered.
Chapter 4 Epiphytism is reconstructed as a crown group synapomorphy in the Rhipsalideae (PP 1.0), Figs. 4.2-4.4. The epilithic and terrestrial growth in the Rhipsalideae appear to be reversals or further shifts. There are different possible scenarios for the epilithic growth. It may represent an ancestral condition within the tribe (PP 0.44) which has been retained and became predominant in some clades. Within Rhipsalis, the probability for epilithic growth found for the backbone nodes is small and increases only in the ancestor of the R. cereoides-clade and the R. teres - R. baccifera clade. It seems therefore, that these groups have independently shifted from epiphytic to predominantly or facultative epilithic habit. The terrestrial growth in Hatiora is also found as a reversal (Fig. 4.4), not as an ancestral condition. The pendent habit appears connected with the epiphytic life-form. The semi-erect or spreading habit appears derived and often connected to epilithic life-form. Adventitious roots are commonly found in epiphytic cacti and are believed to be connected with the epiphytic life-form to allow the plants to attach themselves to the tree bark or rock and to absorb water and minerals (Gibson & Nobel, 2002). It is therefore not surprising that adventitious roots are also frequent in the Rhipsalideae. They are developed in Schlumbergera, Lepismium, Rhipsalis subg. Erythrorhipsalis and subg. Rhipsalis and in 9 other Rhipsalis species. The epiphytes often have a different stem morphology compared to terrestrial cacti. Especially the formation of flattened stems is characteristic and found in all the epiphytic genera. In Rhipsalideae, thin terete stems are also very common and they are found as one synapomorphy of the tribe (Figs 4.1-4.4). Reconstructing the evolution of the different stem-forms is not straightforward. There are numerous shifts between the different stem forms, involving convergent evolution and reversals. The two principal states – the flattened and terete stems evolved both from multi-ribbed stems but it seems that multi-ribbed stems can be either transformed in terete or flattened stems. The flattened stems are derived from ribbed stems as result of the formation of only two ribs (Gibson & Nobel 1986, Wallace & Gibson 2002). They occur consequently only in those clades where also ribbed stems are found and evolved independently in Schlumbergera and in Rhipsalis subg. Phyllarthrorhipsalis. It appears that there are several reversals from flattened to 3-ribbed stems in subg. Phyllarthrorhipsalis. These shifts therefore must be rather easy, and some species (e.g. Rhipsalis pachyptera) produce segments with 3-4 ribs before producing a flattened segment, often also both types of segments are observed on one plant. The thin terete stems which are so typical for the Rhipsalideae could have evolved from cylindrical stems with prominent podaria, as they can be observed for example in Schlumbergera microsphaerica. The podaria were subsequently reduced until they were finally not visible any more, resulting in perfectly terete stems. Such a case can be observed in the R. floccosa-group: most of the taxa do have podaria but they are less prominent in some of the species (e.g. in R. floccosa subsp. oreophila). The 111
Chapter 4 sister group, the R. puniceodiscus-clade have perfectly cylindrical stems, indicating a reduction of the podaria. The Bayesian reconstruction find a PP of 0.7 for the absence of podaria at the node leading to the Rhipsalideae (Fig. 4.4) but the presence or absence of podaria in Rhipsalis appear equally likely (PPs 0.46 absent, 0.54 present). Low or moderate posterior probabilities to form podaria are also found for most of the Rhipsalis clades (Fig. 4.4.), especially for those clades with terete stems, e.g. Rhipsalis subg. Erythrorhipsalis and Phyllarthrorhipsalis. The stem diameter was used here as an approximate indicator for the degree of succulence, assuming that thin stems do not store large amounts of water (Gibson & Nobel 1986). Even if this is only an approximation, there seems to be a tendency for the reduction of succulence in the whole tribe. All optimization schemes find the thin stems with an assumed low degree of succulence as one of the synapomorphies of the Rhipsalideae (Figs. 4.1-4.3) and shifts to the smallest stem diameters in Hatiora, Schlumbergera and in the Rhipsalis subgenera Phyllarthrorhipsalis, Erythrorhipsalis and Rhipsalis. The more succulent stems in the Rhipsalis cereoides-clade would then result from an increase. This increase in succulence is possibly connected with the predominant epilithic habit of this clade; a higher degree of succulence is also found in the likewise epilithic Rhipsalis dissimilis forms. Spines are either absent or reduced to bristles in the majority of the Rhipsalideae and are found as one of the synapomorphies (Figs. 4.1-4.3). Prominent spines are only developed in Schlumbergera opuntioides and S. microsphaerica, in Lepismium lumbricoides forma aculeatum and in Rhipsalis baccifera subsp. horrida. The probability for spines is low (PP 0.23) in the node leading to the Rhipsalideae. This would mean that the prominent spines result from reversals. Barthlott (1983) and Barthlott & Rauh (1975) considered spines on the adult stems of the Rhipsalideae neotenic. Most of the species do have spines in their juvenile stage and often also on primary stem-segments but later reduce the spines. Neoteny is an evolutionary mechanism that allows retaining juvenile traits in the adult stage and thus to re-gain traits that have been reduced or lost. Neoteny is especially common and well known in animals but probably also relevant for angiosperms (Takhtajan 1972).
Figure 4.8 Flowers of the Rhipsalideae. A-C: Rhipsalis. A: R. floccosa: actinomorphic, otherwise unspecialised flowers characteristic for subg. Calamorhipsalis, B: R. elliptica showing the “stamen brush” syndrome with reflexed perianth segments, and exserted stamens as the main visible attractant. This flower type is typical for Rhipsalis. C: R. clavata: apical, campanulate flowers directed downwards, as characteristic for subg. Erythrorhipsalis. D: Lepismium lumbricoides has the same floral syndrome as Rhipsalis subg. Erythrorhipsalis but the flowers are lateral, not apical. E: Rhipsalidopsis rosea: intensely coloured, apical, campanulate flowers. F-G: Schlumbergera. F: S. russelliana: apical actinomorphic flowers with perianth segments fused and reflexed, G: Schlumbergera truncata: zygomorphic flowers with perianth segments fused and reflexed. H: Hatiora salicornioides (left) and H. herminiae (right): apical, campanulate coloured flowers.Photos C-H: W. Barthlott.
4.4.4 Evolution of floral traits and assumed pollination syndromes A first attempt to reconstruct the flower morphology of the Rhipsalideae in a phylogenetic context was recently done by Calvente & al. (2011). They analysed three flower characters: the floral symmetry (actinomorphic vs. zygomorphic), the flower tube (conspicuous vs. inconspicuous) and the flower colour (strong vs. translucent) but functional aspects of the flowers and possible pollination syndromes were not discussed. The ancestral state reconstruction of floral traits in this study indicates that the ancestral flowers of the tribe were actinomorphic (PP 0.6), small (1-3 cm in diameter), with free perianth segments (PP 0.7), and not intensely coloured (PP 0.6). Schlumbergera has many floral innovations, which are the fusion of the perianth segments and thus the formation of a floral tube (PP 0.98), a nectar chamber and zygomorphic flowers. All character optimization schemes suggest that the zygomorphic flower evolved even twice within Schlumbergera, in S. opuntioides and S. microsphaerica and independently in the S. truncata-clade. The Bayesian approach, however, suggests a common origin of zygomorphic flowers already at the node leading to Schlumbergera (PP 0.97). The actinomorphic flowers of S. russelliana would then be the result of a reversal. The same scenario was also found by Calvente et al. (2001). Hatiora epiphylloides that is resolved as belonging in Schlumbergera however has actinomorphic, campanulate flowers and lacks the perianth tube as well as all the other floral synapomorphies of Schlumbergera, such as stamens in two series and connivent stigma lobes. It has the vegetative morphology of Schlumbergera but flowers of Hatiora or Rhipsalidopsis. The different flower morphology of Hatiora epiphylloides is thus not straightforward to interpret and many reversals have to be assumed. But as already discussed in Chapter 3, H. epiphylloides could be a hybrid of a true Hatiora and a Schlumbergera and this might be an explanation of its intermediate morphology. The reduction of flower size seems to be a tendency throughout the Rhipsalideae. Schlumbergera and Rhipsalidopsis have the largest flowers within the tribe while flower size is highly reduced in Hatiora. Within Rhipsalis, small flowers predominate and flower size is even further reduced to a diameter less than 1 cm in subg. Rhipsalis and in some species of subg. Phyllarthrorhipsalis. The character reconstruction suggests small to medium-sized flowers as plesiomorphic, medium-sized to small flowers in the ancestor of the SHLR-clade and a reduction of flower size independently in Lepismium, in Hatiora and in Rhipsalis. Flowers
Schlumbergera, in Hatiora herminiae, and in Rhipsalis. In Rhipsalis, this character is synapomorphic for the clade formed by the subgenera Epallagogonium, Goniorhipsalis, Rhipsalis and Phyllarthrorhipsalis. It is also found convergently in Rhipsalis pilocarpa, R. trigona and R. floccosa subsp. tucumanensis. A nectar disc is also
Chapter 4 developed. The reflexed perianth in Rhipsalis causes prominent exposure of the stamens leading to a unique flower type termed “stamen-brush” flowers in the following. (Fig. 4.8 B). Coloured flowers are found with high probability as absent in Rhipsalis (PP 0.99). One of the exceptions is R. hoelleri, which has intensely red coloured flowers, indicating it attracts different pollinators than the rest of Rhipsalis. Some Rhipsalis have yellowish or yellow flowers (e.g. R. elliptica Fig. 4.8 B). The presence of coloured flowers appears to be the result of reversals, so it seems shifts from coloured to noncoloured flowers or vice versa are rather easy. The pollination of epiphytic cacti is difficult to study in the field and consequently, there is only very limited information on their pollination biology from field observations. In general, very few cactus genera have been studied in the field for their pollination biology (Pimienta-Barrios & del Castillo 2002). Nevertheless, Cactaceae flowers can be classified in different pollination syndromes, based on flower shape and colour. The major pollinator groups are bats, moths, birds and bees (Pimienta-Barrios & del Castillo 2002, Porsch 1938, 1939). For the Rhipsalideae, it is possible to classify the flowers in two main groups: as bird- or insect pollinated. The character reconstruction allows four main flower “types” found in the Rhipsalideae, based on combinations of different characters. The most distinct flowers are the comparatively large, tubular, flowers of Schlumbergera which have an intensely magenta coloured perianth and also coloured styles, filaments and pollen. They can be considered bird-pollinated and hummingbirds have indeed been reported visiting Schlumbergera flowers (McMillan & Horobin 1995). Rose & Barthlott (1994) also suggest that red coloured pollen is also part of the bird-pollination syndrome, as a mimetic adaptation. The pollen colour is similar to the colour of the bird’s beak and is thus less irritation for the bird than a contrasting pollen colour would be. Rhipsalidopsis has actinomorphic, campanulate flowers, also intensely coloured and probably pollinated by birds as well. The flowers of Hatiora are more difficult to classify, they are actinomorphic, small, either intensely yellow or magenta and may be visited by birds or/and by insects. Within Rhipsalis, there are three different flower “types” which possibly attract different pollinators. Most common are the white or whitish stamen-brush flowers with a reflexed perianth (Fig. 4.8 B). Other Rhipsalis have no reflexed perianth and the flowers are either funnel-shaped or more or less campanulate but with no apparent species characteristics and usually not coloured (Fig. 4.8 A). Finally, Lepismium and Rhipsalis subg. Erythrorhipsalis share the campanulate, pendent flowers directed downwards with a white perianth (pink only in some forms of L. cruciforme). It is therefore likely that Rhipsalis subg. Erythrorhipsalis attracts other pollinators compared to the rest of Rhipsalis but the same pollinators as Lepismium.
Morphological intermediates in the Rhipsalideae There are five Rhipsalis species which seem misplaced because they do not “fit”
in morphologically well defined clades. In fact, their morphology appears intermediate between the clade they are part of and another, more distant clade. Rhipsalis pulchra lacks the apical flowers characteristic for subg. Erythrorhipsalis and resembles R. puniceodiscus. Rhipsalis grandiflora and R. pittieri have neither angled nor flattened stems characteristic for subg. Phyllarthrorhipsalis but instead have terete stems. On the other hand, R. grandiflora shares most characters with Phyllarthrorhipsalis, such as multiple flowering at one areole and repeated flowering. The placement of R. pittieri in Phyllarthrorhipsalis is more difficult to explain, as it shares hardly any characters with the rest of this subgenus but is instead very similar to the other R. floccosa-clade and has been included in it, as a subspecies of R. floccosa, which it very closely resembles. Rhipsalis ewaldiana has a characteristic vegetative habit with long shoots and
mesembryanthemoides. Based on this, these two species have been regarded as sister species. But Barthlott & Taylor (1995) in their first description of R. ewaldiana also stated that it might even be a hybrid of R. mesembryanthemoides and a species with winged or angled stems. But R. ewaldiana fits well into subgenus Phyllarthrorhipsalis by having angled stems. Considering that the sister relationship of R. ewaldiana and R. goebeliana is supported with 100%, R. goebeliana might have been the second hybrid parent with flattened stems. Finally, Rhipsalis sulcata falls in the R. teres alliance and cannot even be separated from the R. teres accessions sampled. This species has angular stems and Barthlott & Taylor (1995) therefore considered it related to the other species with angular stems, e.g. R. paradoxa or R. pentaptera. But at the same time, Rhipsalis sulcata has the habit of R. teres and R. baccifera: pendent stems, indeterminate basal extension shoots and strict acrotonic branching. Even if the angled stems appear exceptional in subg. Rhipsalis, they are sometimes developed also in R. teres (f. prismatica). One possible explanation for the placement of these taxa is that they are hybrids and found next or close to their hybrid mother-parent. Hybrids are common in Cactaceae and known from Schlumbergera and Rhipsalidopsis. No definite hybrids in Rhipsalis are known from cultivation so far; although Taylor (1999) reports a plant from cultivation which appears to be an intermediate between Rhipsalis puniceodiscus and R. neves-armondii and he assumes to be of hybrid origin. Therefore, these morphological intermediates can be a first hint towards hybridization in Rhipsalis, but this has to be confirmed using nuclear markers.
Morphological characterization of the clades inferred by sequence data corresponding to genera The aim of a classification based on a phylogenetic hypothesis is to classify
(=name) monophyletic entities. The formally recognized taxa such as genera or subgenera should desirably be also recognizable by their morphology. However, Cactaceae pose problems for finding morphological characters defining taxonomic entities, as already discussed in the Chapters 1 and 2. The result from the character reconstruction here is that most vegetative characters are homoplastic, including many of the characters used to define genera in the past. Emphasis of single characters may therefore be misleading when they are used to define taxonomic units. Therefore, not all characters are equally useful for delimitation of all the genera and subgenera. While stem morphology defines some clades, flower morphology defines others and mostly only combinations of characters allow an unambiguous diagnosis. But although most of the characters appear more than once, they are homogenous within the respective clades and therefore can be used as diagnostic characters. So as a result, all the highly supported clades found by the molecular phylogenetic analyses can be defined morphologically. The most relevant characters are the branching pattern, the determinate growth of stem-segments, the shedding of old segments, the stem form, the position of the areoles (superficial compared to sunken areoles), woolly areoles postanthesis, the flower position, flower orientation, flower number per lateral areole. The diagnostic characters for the genera of Rhipsalideae as found by the character optimization schemes are summarised in Tables 4.1 and the characters defining the subgenera of Rhipsalis are listed in Table 4.2.
Rhipsalis Branching Stem segments
Stem form Old segments Apical composite areoles Flower position
Flower size Flower colour
Pericarpel Stamen insertion Fruits Fruit colour
strictly acrotonic or subacrotonic commonly determinate, in some subgenera indeterminate terete, flattened or 3ribbed, sometimes with prominent podaria deciduous present (rarely absent) predominantly lateral or lateral to apical or only apical at composite areoles actinomorphic, perianth often reflexed, stamen-brush, OR campanulate, pendent, oriented downwards, stamen-brush not developed small to very small mostly white/ whitish, or pale yellow. Rarely intense yellow or pink smooth, never angled, terete and naked (bristly in R. pilocarpa) one series globose or subglobose (barrel-shaped) white or coloured
zygomorphic, rarely actinomorphic with a well developed perianth tube
small/ very small
white/whitish, pink only in L. cruciforme
yellow or pink
red or pink
angled or ridged
angled (sometimes only slightly angled)
smooth, not angled
coloured, mostly dark red to almost black
greenish or pink
red or yellow
Table 4.1 Morphological characteristic for the Rhipsalideae genera. Apomorphic characters are highlighted in bold.
Table 4.2 Diagnostic features of the Rhipsalis subgenera. Apomorphic characters are highlighted in bold
determination of stem segments stem form areoles development woolly areoles post-anthesis flower buds position flower position flower orientation flowers per lateral areole flower morphology
Calamorhipsalis Erythrorhipsalis Goniorhipsalis
sunken, except apical areole (not in R. lindbergiana)
sunken, except apical ar
aligned with stem axis
lateral to apical
lateral to apical
lateral to apical
lateral to apical
several (rarely one)
actinomorphic, perianth not reflexed
actinomorphic, campanulate, pendent
perianth reflexed, stamen brush developed
perianth reflexed, stamen brush developed
perianth reflexed, stamen brush developed
perianth reflexed, stamen brush developed
terete + offset podaria sunken, sometimes also the apical areoles
Chapter 5 Towards understanding the historical phylogeography of Rhipsalis baccifera, the most widespread cactus Summary Rhipsalis baccifera is the most widespread cactus and the only cactus that is native to tropical Africa. The distribution patterns of Rhipsalis baccifera are addressed in this chapter using tree building methods and haplotype network algorithms. The taxon sampling included 42 Rhipsalis baccifera specimens covering most of the area. A haplotype network based on the rps3-rpl16 spacer and the rpl16 intron was constructed using the statistical parsimony as implemented in TCS, and Maximum Likelihood (ML) methods. The TCS algorithm found 10 haplotypes whereas a network derived from ML analysis found 17 haplotypes. Two main groups of plastid haplotypes were found using both methods: a northern South American haplotype that included specimens from the Caribbean and Mesoamerica and a haplotype shared by the African specimens. Besides, unique haplotypes were found in several South American and African specimens. These results suggest a single dispersal of Rhipsalis baccifera to Africa and reveal high genetic diversity within its populations.
5.1 INTRODUCTION Most Cactaceae have distribution areas of about 10.000 km2 but Rhipsalis baccifera occupies an area which is estimated to be 2000 times larger (Barthlott et al., unpublished data). Its range, as illustrated in Fig. 5.1, covers large parts of northern tropical South America to the Caribbean and Mexico. Rhipsalis baccifera is thus the most widespread cactus and besides, it is the only cactus with a natural occurrence in the old World where it ranges through large parts of tropical Africa to Madagascar and to Sri Lanka (Barthlott, 1983). There are currently 5 accepted subspecies of Rhipsalis baccifera (Barthlott & Taylor 1995, Hunt 2006) which also have distinct geographical distributions – their areas do not overlap. The subsp. baccifera is found throughout northern South America, in the Caribbean, in southern Florida and in Mexico. The subsp. hileiabaiana is endemic to the state Bahia in south eastern Brazil, in the state of Bahia. The erstwhile subsp. shaferi (= Rhipsalis shaferi, see Chaper 2) replaces subsp. baccifera in Paraguay, Bolivia and northern Argentina. In Africa, subsp. mauritiana is found throughout tropical Africa, subsp. erythrocarpa occurs in the mountains of tropical east Africa and subsp. horrida is endemic on Madagascar. The occurrence of Rhipsalis baccifera in the Old World has long been known and has puzzled taxonomists and biogeographers for more than 100 years. The fist formally proposed name (Rhipsalis aethiopica Welw.) was published in 1859 but the plants were known at least 50 years before (references in Barthlott 1973). Most authors considered them to be identical with or closely related to the South American Rhipsalis baccifera. However, the exact origin of the African populations is unknown. The only suggestion is that of Backeberg (1942) who assumed dispersal to Africa from north-eastern South America. The most commonly accepted hypothesis how Rhipsalis baccifera may have reached Africa was and still is dispersal by migratory birds (Backeberg 1942, Barthlott, 1983). But there are no migratory birds known that cross the Atlantic Ocean and may have brought Rhipsalis baccifera seeds to Africa. Consequently, these plants have also been considered to be Gondwana relicts (e.g. Croizat, 1952) or in the other extreme as introduced by man in the last 200 years (e.g. Buxbaum, 1970a). But these two theories were purely speculative, with no supporting data. The first detailed study of the palaeotropic Rhipsalis baccifera was conducted by Barthlott (1973, 1984). He examined the morphology, pollen and ploidy level of the palaotropic Rhipsalis. He found the African populations to show more variability and also unique characters, not found in their South American relatives. All the African
Chapter 5 populations studied so far are polyploid (4n=44, 6n=66 and 8n=88). The subsp. erythrocarpa has red fruits, otherwise not found in subgenus Rhipsalis. The populations from Madagascar are often terrestrial and spinyand their pollen has a unique reticulate tectum (Barthlott 1973, 1976, 1983). These results argued against a recent introduction of Rhipsalis baccifera into Africa but rather suggested a long independent evolution of these populations. The theory that Rhipsalis was a Gondwana relict was also rejected because it appeared too derived to be an old Gondwana taxon (Barthlott 1983). The population of Rhipsalis baccifera have not yet been analysed using molecular data. So far, sequences of plastid markers have already revealed some genetic diversity within the specimens sampled (Chapter 2, this study) and therefore a more detailed study seemed promising and was conducted here. The main questions on the biogeography of Rhipsalis baccifera are: How different are the African populations in comparison to their South American relatives? A high genetic distance should support the hypothesis of a long independent evolution thus arguing against a recent introduction. The next immediate question is therefore: When was the dispersal to Africa? Was there a single dispersal event or were there even independent dispersals? From where in South America or Mesoamerica did the dispersal take place? Do the morphologically different Malagasy populations result from further independent evolution on Madagascar? The analyses presented in this chapter are a first step towards understanding the evolutionary history and distribution patterns of Rhipsalis baccifera. Traditional treebuilding methods and haplotype network construction algorithms are applied. Both rely on sequence data from the rps3-rpl16 spacer and the rpl16 intron. This region is very variable, as found in the datasets of Chapters 1 and 2. It also shows variation at population level and was therefore chosen for the analyses here.
Figure 5.1 Estimated distribution area of Rhipsalis baccifera. Distribution data from Barthlott 1983, Taylor & Zappi 2004,
5.2 MATERIAL AND METHODS 5.2.1 Plant material and taxon sampling The plant material was obtained from the living collections of the Botanic Gardens Bonn and the Botanical Garden Berlin-Dahlem. The sampling strategy was to include as many accessions from different origins as possible thus trying to cover most of the area. Totaling 42 accessions of the subspecies baccifera, mauritiana, erythrocarpa and horrida were sampled, with 20 specimens from South America, and 22 specimens from the Old World distribution area. The complete source information is provided in the Appendix 1.
5.2.2 Isolation of genomic DNA, amplification and sequencing The plant material freshly collected then cut in small pieces and dried on silicagel in a drying chamber at 35°C for app. 24 hrs. Genomic DNA was then isolated using a CTAB method as described in Chapter 1. DNA concentration and purity (A260/A280 ratio) were measured using a NanoDrop ND-1000 (peqLab, Erlangen, Germany). A working dilution of 10ng/μl was made to be used for PCR. The rpl16 intron and the rps3-rpl16 spacer were co-amplified as described in Chapters 1 and 2, the primer sequences are listed in the Appendix 2. All PCR products were stained with 100x SybrGreen nucleic acid stain and electrophoresed on a 2% agarose gel, excised and purified using the Gel/PCR DNA Fragment Extraction Kit (Avegene) according to manufacturer’s instructions and sequenced via Macrogen Inc. (Seoul, South Korea). Manual editing of pherograms, assembly of sequences and manual sequence alignment was done using PhyDE v.0995 (Müller & al. 2005+).
5.2.3 Phylogenetic analyses and haplotype network construction All the sequences of Rhipsalis baccifera accessions sampled were added to the rpl16 dataset of Chapter 2 and analysed using Bayesian Inference as described therein. The analysis was run for 5000000 generations. The first 2000 trees were discarded and the remaining trees were summarised into a majority-rule-consensus tree. A haplotype network was constructed using TCS (Clement & al. 2000) which implements the Statistical parsimony (Templeton & al. 1992). Standard phylogenetic reconstruction methods were additionally applied as they have recently been shown to perform well for haplotypic data (Salzburger & al. 2011). Trees were build using Bayesian Inference with MrBayes 3.1 (Huelsenbeck & Ronquist 2001), also using the heuristic search in PAUP* (Swofford 1998) under Maximum Parsimony (MP) and Maximum Likelihood (ML) and using Neighbour-Joining (NJ). The ML search was based on the best-fitting nucleotide substitution model (F81) as evaluated with jModeltest (Posada 2008) and the AIC information criterion. The resulting trees were imported into Haplotype Viewer (G. Ewing, available at www.cibiv.at/~greg/ 124
Chapter 5 haploviewer). This software converts trees build from traditional phylogenetic methods to haplotype genealogies.
5.3 RESULTS 5.3.1 Sequence characteristics, phylogenetic analyses and haplotype network construction The amplified fragment of the rps3-rpl16 spacer
CA176 Rh baccifera CO CA179 Rh baccifera VE CA162 Rh baccifera horrida MG CA189 Rh baccifera CO CA159 Rh baccifera EC CA181 Rh baccifera PE CA145 Rh baccifera RE CA146 Rh baccifera GA CA147 Rh baccifera TO CA150 Rh baccifera GF CA167 Rh baccifera SC CA166 Rh baccifera ZW CA185 Rh baccifera ZW CA155 Rh baccifera mauritiana SC 0.59 CA015 Rh baccifera erythrocarpa RW CA174 Rh baccifera mauritiana KE CA180 Rh baccifera mauritiana CD CA157 Rh baccifera mauritiana IC CA014 Rh baccifera subsp horrida MG CA160 Rh baccifera mauritiana MG CA164 Rh baccifera mauritiana MG CA165 Rh baccifera horrida MG CA175 Rh baccifera mauritiana MG CA178 Rh baccifera mauritiana MG CA158 Rh baccifera mauritiana MG CA186 Rh baccifera horrida MG CA187 Rh baccifera horrida MG CA188 Rh baccifera horrida MG CA117 Rh baccifera CO 1 CA153 Rh baccifera mauritiana ZA CA152 Rh baccifera subsp hileiabaiana 1 CA099 Rh baccifera CR 0.99 CA184 Rh baccifera MX CA118 Rh teres f heteroclada CA002 Rh baccifera BR CA013 Rh baccifera VE CA161 Rh baccifera VE CA168 Rh baccifera VE CA169 Rh baccifera GT CA170 Rh baccifera CO CA173 Rh baccifera PY 0.95 CA177 Rh baccifera BR CA183 Rh baccifera JM CA135 Rh baccifera CO CA138 Rh baccifera CU CA062 Rh sulcata 0.89 CA097 Rh teres f capilliformis CA100 Rh teres f heteroclada CA011 Rh teres f heteroclada CA101 Rh mesembryanthemoides 0.88 CA102 Rh teres CA003 Rh shaferi Figure 5.2 50% majority-rule consensus tree from 1 Phyllarthrorhipsalis Inference including all the Rhipsalis baccifera accessions 1 Goniorhipsalis The other Rhipsalis subgenera have been reduced to single Epallagogonium Erythrorhipsalis for better readability. Numbers above branches are Calamorhipsalis Probabilities. Outgroup 0.73 1
and the rpl16 intron was 1183-1269 nt in length with
a mean length of 1200 nt. and 1,2 % variable characters. Most variability was found in length variable mononucleotide stretches within the rpl16
intron. Larger indels were also found, a 19 nt gap occurs in four of the sampled accessions.
The relationships found within subg. Rhipsalis,
including all the 42 Rhipsalis baccifera accessions
sampled, are shown in Fig. 5.2. The TCS haplotype network of Rhipsalis baccifera is shown in Fig. 5.3 and the results from the haplotype network derived from the ML analysis is shown in Fig. 5.4. These different methods find a different number of haplotypes: The TCS algorithm found 9 haplotypes whereas the network
haplotypes. There are two haplotypes characterising the majority of the samples. The first haplotype is found in specimens from northern South America (Brazil,
Caribbean (Cuba and Jamaica).
Bayesian sampled. branches Posterior
Chapter 5 The second haplotype comprises all the specimens sampled from tropical Africa, Madagascar, the Seychelles and Réunion. It is also found in 3 samples from Northern South America (northern Colombia, Ecuador, Peru, French Guiana). Besides, TCS and ML find 7 and 15 unique haplotypes, respectively. These are derived from either the northern South American or the African haplotype and found in only 1 or 2 specimens.
Northern South America, Caribbean BR PY VE CO GT CR CU JM (12)
CA002 BR CA153 ZA
CA176 CO CA162 MG
Northern S-America: CO EC PE (3) tropical Africa
MG RW RE GA TO GF SC ZW KE CD (22)
Figure 5. 3 TCS Haplotype network of Rhipsalis baccifera based on sequence data of rps3-rpl16 and the rpl16 intron.
Figure 5.4 Haplotype network generated from a Maximum Likelihood analysis of Rhipsalis baccifera based on sequence data of rps3-rpl16 and the rpl16 intron.
5.4 DISCUSSION 5.4.1 First insight into the biogeography of Rhipsalis baccifera As outlined in the introduction, the commonly accepted hypothesis for the origin of the African Rhipsalis baccifera is that they are derived from South American populations. This is supported by all the data on these populations available so far (Barthlott 1983). The haplotype network analysis here finds one haplotype characterising all the African populations and derived from the northern South American haplotype. This supports the assumption of a single dispersal to Africa from South America. There are also unique haplotypes found in specimens from Africa which are secondly derived from the African haplotype. Of the 10 specimens sampled from Madagascar, 2 unique haplotypes are found by the TCS algorithm and 4 by the ML analysis. The ML analysis additionally finds a unique haplotype in the specimen from Gabun (CA146). These unique haplotypes suggest a long independent evolution and argue against a recent introduction of Rhipsalis baccifera to Africa by man. These results suggest further diversification and independent evolution of the African and especially Malagasy populations. This is in line with the hypothesis of Barthlott (1984) who assumed Madagascar to be an evolutionary centre of Rhipsalis baccifera. The haplotypes found also reveal genetic diversity within the South American and Mesoamerican populations. The specimens from Mexico, Costa Rica, Guatemala and from Brazil have each unique haplotypes. This suggests several dispersals from northern South America further southwards and northwards. There are some haplotypes that differ from the two frequent haplotypes, especially in South America. The intermediate haplotypes are missing (marked by black dots in the network). Several mutational steps have to be assumed in order to explain these haplotypes. The missing intermediate haplotypes could either be explained by incomplete taxon sampling or by the loss of these haplotypes. Most difficult to explain is the fact that some South American populations have the African haplotype (Figs. 5.3 and 5.4). The TCS network finds reticulations involving the African haplotype and the samples CA179 and CA176. The ML analysis also haplotypes in specimens from Colombia (CA176) and Venezuela (CA179) to be derived from the African haplotype, with missing haplotypes in between (Fig. 5.3). There is also a unique haplotype in the South African specimen (CA153). It is derived from the northern South American, not from the African haplotype. A possible explanation for this would be multiple dispersals, including dispersal from Africa back to South America and maybe even a second dispersal to Africa. But
Chapter 5 multiple dispersals of Rhipsalis baccifera have not yet been assumed and therefore need further investigation with a more thorough taxon sampling.
5.5 CONCLUSIONS AND OUTLOOK The study presented in this chapter is just a first step towards a more detailed study of Rhipsalis baccifera. Unfortunately, the data do not provide any structure within the northern South American and African specimens, respectively. Therefore no conclusions about ancestral distribution and possible migration routes are possible at this point. Future work will be based on a larger taxon sampling, including desirably more specimens from individual populations. Also more markers, plastid and nuclear, will be added for the construction of the haplotype network and microsatellite markers will be used as well. Chromosome numbers for all the samples would need to be collected provide insights whether there was one polyploidization event connected to the dispersal to Africa and whether or if there are polyploids, maybe independently, already in South America or Mesoamerica. The timeframe for the dispersal to Africa also still has to be inferred using molecular clock dating.
Chapter 6 Development of microsatellite loci for Rhipsalis baccifera using 454 sequencing
6.1 INTRODUCTION Rhipsalis baccifera is the most widespread Cactaceae species and the only cactus that ranges through a large part of tropical Africa. However, its biogeographic patters have not been studied in detail and especially the origin of the African populations is an open question. At the same time, it seems that the morphological differentiation in Rhipsalis baccifera is connected with its distribution patterns. Some populations in northern South America and especially the African populations differ morphologically. But it is hardly possible to assess this variation with “classical” methods studying the morphology. It is well possible that there are still unrecognised cryptic species under the name Rhipsalis baccifera. A detailed study of the biogeography of Rhipsalis baccifera should therefore provide insights into its evolutionary history and possibly also into the mode of speciation. A better understanding on the genetic variation within the populations could also be of value for conservation assessments if genetically unique populations will be found. So far, sequences of plastid markers have already revealed some genetic diversity within the Rhipsalis baccifera specimens sampled (Chapter 2 and 5, this study). But the plastid markers do not provide enough resolution between the individual populations and therefore more variable markers are needed. Microsatellite loci appear especially promising for this purpose. Microsatellites are highly variable DNA stretches with tandem repeats of few nucleotides, most commonly one, two, three and four nucleotides. Microsatellites offer some advantages in comparison to other population-level markers such as AFLPs (Vos & al. 1995). Using microsatellites, partial datasets can be generated that can be later expanded once suitable primers are designed. One of the disadvantages of using microsatellite loci was in the past that their development was laborious and expensive and required extensive cloning. The cloning steps unnecessary when next generation sequencing methods are applied. Using the next generation 454 sequencing ten thousands of reads can be generated with just one run. The reads obtained can then be screened for repeat-containing motifs and several algorithms and software is available for that purpose. The 129
Chapter 6 development of microsatellite loci using 454 sequencing was initially tested by Santana et al. 2009. Since then, an increasing number of studies develop microsatellite markers using 454 sequencing (Abdelkrim & al. 2009, Allentoft & al. 2009, Castoe & al. 2010, Csencsics & al. 2010, Lee & al. 2009, Tangphatsornruang & al. 2009) In this chapter, microsatellite markers for Rhipsalis baccifera have been developed using 454 sequencing and based on a genomic library enriched for repeat motifs. The first followed the AFLP protocol of Vos & al. (1995) and subsequent steps largely followed the protocol for isolation of microsatellite loci provided by Glenn & Shable (2005). The actual testing of primers and the application of the selected loci are beyond the scope of this study and will be the object of further work.
6.2 MATERIAL AND METHODS 6.2.1 Plant material and taxon sampling The Rhipsalis baccifera accession no. 166048323 (Leuenberger 3088, Brazil, Bahía, Camarca, isolate number CA148, vouchered at B) was chosen from the Cactaceae living collection of the Botanical Garden Berlin-Dahlem.
6.2.2 Chromosome count The number of chromosomes was determined since the plant had to be desirably diploid for the following genomic library construction and polyploidy is occasionally observed in Rhipsalis baccifera (Barthlott 1976), although so far only in specimens from the Caribbean and Florida and the African populations. Growing root tips of the aerial roots were collected at c. 8:40 h in the morning and pre-treated in 0,002 M solution of 8-hydroxychenoline for c. 4 h in a refrigerator at 5-8°C. They were then fixed with a mixture of 3:1 ethanol 96%-acetic acid for app. 24 h in a refrigerator at 58°C. The root tips were then hydrolyzed in 1 N HCL for 10 min. at 60°C then transferred into dest. water. A piece of c. 2 mm of the root tip was carefully squashed with a needle, then stained with aceto-orcein and carefully squashed under a cover glass. The root tips were examined using a light microscope (Zeiss standard 14) and documented using a digital camera (Zeiss AxioCam MRc).
6.2.3 Isolation of genomic DNA The plant material freshly collected then cut in small pieces and dried on silicagel in a drying chamber at 35°C for app. 24 hrs. Genomic DNA was then isolated using a CTAB method as described in Chapter 2. DNA concentration and purity (A260/A280 ratio) were measured using a NanoDrop ND-1000 (peqLab, Erlangen, Germany).
6.2.4 Genomic library construction 18.104.22.168
The genomic DNA was digested using the restriction enzymes EcoRI and MseI (New England Biolabs) using a reaction mixture of 2.50 μL NEB 10x Ligase Buffer (pre-heated to 50°C to get all components in solution), 0,25 μL 100x Bovine Serum Albumine (New England Biolabs, supplied with the enzymes), 0.25μL 5M NaCl (50 mM final), 1 μL EcoRI, 1 μL MseI, 20 μL genomic DNA (@concentration 120 ng/μL). The restriction digest set-up was incubated in a thermal cycler at 37°C for 2 hours. The success of the digestion was verified by running 4 μl of the digested DNA on a 1% agarose gel (30 min, 100 V). To ensure that the DNA fragments to be used for the following steps were between 400-800 nt in length, the whole volume of the digested DNA from the previous step was run on a 2% agarose gel for c. 1 h. DNA of the desired size was excised from the gel and purified using the QIAquick Gel Extraction Kit (Quiagen) according to the manufacturer’s instructions. 22.214.171.124
Ligation of EcoRI and MseI adapters to the restriction fragments
To form double stranded adapters, equal volumes of equal molar amounts of EcoRI and MseI adapters were mixed (6,5 μL EcoRI-linker, 6,5 μl MseI-linker, (10 μm each). The mixture was heated to 95°C and cooled down in a water bath to room temperature and then incubated at 16°C overnight. Adaptors used: EcoRI adaptor: 5'-CTCGTAGACTGCGTACC CATCTGACGCATGGTTAA-5'
For the adapter ligation, 6,5 μl of EcoRI and 6,5 μl of MseI adapters were mixed with 4 μl of 10x T4 DNA ligase buffer and 3 μl T4 DNA ligase (New Englad Biolabs, Cat. No. Mo202S) (400 U/μl). This mixture was added to the DNA from the previous step. To test the success of the ligation, a test PCR was run. A 50μl reaction containing 4 μl of the linker ligated DNA, 5 μl peqlab Taq polymerase buffer, 10 μl peqlab PCR enhancer solution, 5 μl BSA @ 350 μg/ml, 2, 6 μl EcoRI primer EcoRI primer: 5'CTCGTAGACTGCGTACCAATTC, 2,6 μl MseI primer 5'-GACGATGAGTCCTGAGTAA, 3 μl dNTPs,
0,4 μl peqlab Taq DNA polymerase and 17,4 μl ultrapure H2O. The
amplification conditions were: 95°C for 2 min. followed by 20 cycles of 95°C for 20 sec., 60°C for 20 sec., 72°C for 1.5 min., then the reaction was held at 15°C. Success of the PCR was checked by running 4μl of the product on a 1.5% agarose gel.
Chapter 6 126.96.36.199
Enrichment of the genomic library for repeat motifs
The PCR products from the previous step were used. Two enrichments of repeatcontaining sequence motifs were carried out using two different 3-biotinylated oligomixes. Oligo mix 1: (AG)12, (TG)12, (AAG)8, (AAT)12, (ACT)12 Oligo mix 2: (AAAC)6, (AAAG)6, (AATC)6, (AATG)6, (ACAG)6, (ACCT)6, (ACTC)6, (ACTG)6 The linker ligated DNA was hybridized with the two oligo mixes following the protocol of Glenn & Shable (2005). The hybridized DNA was afterwards added to 150 μl of Dynabeads washed as described by (Glenn & Schable 2005). The following steps also followed the procedure described therein. The enriched DNA was recovered using a PCR with 2 μl of the DNA and the same reaction set-up as described above for the testing of adaptor ligation success. The cycling conditions were 95°C for 2 min. followed by 25 cycles of 95°C for 20 sec., 60°C for 20 sec., 72°C for 1.5 min.; then 72°C for 10 min.; then hold at 15°C. Success of the PCR was checked by running 4μl of the product on a 1.5% agarose gel. The product from this PCR was used for the second enrichment.
6.2.5 454 sequencing The genomic library was first purified using the Amplicon Library Preparation Protocol (Roche) according to the manufacturer’s instructions. Then the library was sequenced of a GS FLX System using the LibL Kits and the GS FLX Titanium Sequencing Kit XLR70 (both Roche) according to the manufacturer’s instructions.
6.2.6 Screening for repetitive motifs and primer design The search for repetitive sequence motifs was done using QDD (Meglécz & al. 2010). This software uses a series of perl scripts in combination with BLAST (ftp://ftp.ncbi.nih.gov/blast/executables/), Clustal X (Larkin & al. 2007) and primer3 (Rozen & Skaletsky 2000). In the first step, contigs from reads containing the same sequence motifs containing repetitive motifs are assembled. In the following step primers are automatically designed for the loci selected by the software.
6.3 RESULTS 6.3.1 Chromosome counts The selected plant is diploid and has a chromosome number of 2n=22.
6.3.2 454 sequencing and microsatellite loci found The 454 run produced 86125 reads with a length between 130 and 400 bp. The screening for repetitive elements using QDD produced 716 loci found for which primers could be designed of which 103 were marked as best by the software. The output containing the repeat motif, their length and the designed primers and their properties is shown in Table 6.1. Dinucleotide repeats are most common, tri and –tetranucleotide repeats were also found and the largest repeat motif comprises 6 nucleotides. Most microsatellites are between 10 and 24 nt in length, few are between 30-36 nt, only three longer microsatellites (of 42 nt) were detected.
6.4 DISCUSSION The approach using the enriched genomic library and 454 sequencing was very effective successful. The genomic library construction took about one week but can be even done faster. The whole procedure from isolation of genomic DNA to the 454 sequencing can be done less that two weeks each steps are done immediately after the previous step. The 454 reads are rather short, the run here produced reads with often only 100-130 bp. But usually read lengths of c. 400 nt are expected. During the genomic library construction, attention was paid to obtain restriction fragments of about 400 nt in leght to ensure reads which are long enough for primer binding sites. Nevertheless, the reads were still suitable for primer design. A high number of reads containing microsatellites was found. This is a very good prerequisite for selecting the loci for initial testing. The loci to be testes will include those with di, tri, tetra and hexanucleotide repeats, preferably with longer repeat stretches. Those fragment which have primer combinations with a similar Tm will be preferred so that the same annealing temperature can be used for all primer combinations. Following the approach of Csencsics et al. (2010), the following criteria will be applied: amplification products larger than 100 bp, primer melting temperature of 60.0 °C, primer GC content of 50% and low levels of self- or paircomplementarity of the primers. Successfully amplified loci will then be amplified for c. 20 samples to test whether they are polymorphic and heterozygous.
Chapter 6 Table 6.1 Primers for the microsatellite loci found. No.
F R F R F R F R F R F R F R F R F R F R F R F R F R F R F R F R F R F R F R F R F R F R F
References Abdelkrim, J., Robertson, B. C., Stanton, J. A. L. & Gemmell, N. J. 2009: Fast, costeffective development of species-specific microsatellite markers by genomic sequencing. - Biotechniques 46: 185-191. Allentoft, M. E., Schuster, S. C., Holdaway, R. N., Hale, M. L., McLay, E., Oskam, C., Gilbert, M. T. P., Spencer, P., Willerslev, E. & Bunce, M. 2009: Identification of microsatellites from an extinct moa species using high-throughput (454) sequence data. - Biotechniques 46: 195-+. Anderson, E. F. 2001: The Cactus Family. - Portland, Oregon. Applequist, W. L. & Wallace, R. S. 2002: Deletions in the plastid trnT-trnL intergenic spacer define clades within Cactaceae subfamily Cactoideae. - Plant Syst. Evol. 231: 153-162. Arakaki, M., Soltis, D. E. & Speranza, P. 2007: New chromosome counts and evidence of polyploidy in Haageocereus and related genera in tribe Trichocereeae and other tribes of Cactaceae. - Brittonia 59: 290-297. Arias, S., Terrazas, T. & Cameron, K. 2003: Phylogenetic analysis of Pachycereus (Cactaceae, Pachycereeae) based on chloroplast and nuclear DNA sequences. Syst. Bot. 28: 547-557. Arias, S., Terrazas, T., Arreola-Nava, H. J., Vazquez-Sanchez, M. & Cameron, K. M. 2005: Phylogenetic relationships in Peniocereus (Cactaceae) inferred from plastid DNA sequence data. - J. Plant Res. 118: 317-328. Backeberg, C. 1938: Zur systematischen Übersicht / Diagnosen neuer Gattungen und Untergattungen. - Blätter für Kakteenforschung 6. Backeberg, C. 1942: Zur Geschichte der Kakteen im Verlauf der Entwicklung des amerikanischen Kontinentes. - Neudamm. Backeberg, C. 1958-1962: Die Cactaceae: Handbuch der Kakteenkunde. - Jena. Backeberg, C. 1959: Die Cactaceae. Handbuch der Kakteenkunde. Band II Cereoideae (Hylocereae - Cereae). - Jena. Backeberg, C. 1966: Das Kakteen Lexikon. - Jena. Bárcenas, R. T., Yesson, C. & Hawkins, J. A. 2011: Molecular systematics of the Cactaceae. - Cladistics 27: 1-20. Barfuss, M. H. J., Samuel, R., Till, W. & Stuessy, T. F. 2005: Phylogenetic relationships in subfamily Tillandsioideae (Bromeliaceae) based on DNA sequence data from seven plastid regions. - Am. J. Bot. 92: 337-351. Barthlott, W. & Hunt, D. 1993: Cactaceae. - Pp. 161-197 in: Kubitzki, K. (ed.), The families and genera of vascular plants, Vol. II. Berlin Heidelberg New York Tokyo. Barthlott, W. & Hunt, D. 2000: Seed-diversity in Cactaceae subfam. Cactoideae. Milborne Port. 139
References Barthlott, W. & McMillan, A. J. S. 1978: A new species of Schlumbergera (Cactaceae). Cact. Succ. J. (Los Angeles). L: 31-35. Barthlott, W. & Rauh, W. 1975: Notes on the morphology, palynology, and evolution of the Genus Schlumbergera Lemaire (Cactaceae). - Cact. Succ. J. (Los Angeles). supplemental volume: 5-21. Barthlott, W. & Rauh, W. 1977: Die Wildarten und Hybriden der Weihnachtskakteen (Gattung Schlumbergera). - Kakteen And. Sukk. 28: 273-278. Barthlott, W. & Rauh, W. 1987: Pfeiffera miyagawae, a new orange flowered species from Bolivia. - Cact. Succ. J. (Los Angeles). 59: 63-65. Barthlott, W. & Taylor, N. P. 1995: Notes towards a monograph of Rhipsalideae (Cactaceae). - Bradleya 13: 43-79. Barthlott, W. 1973: Beitrag zur Taxonomie und Biogeographie der palaeotropischen Rhipsalis (Cactaceae) und zur allgemeinen Mikromorphologie der Cactaceen. – PhD thesis, Institut für systematische Botanik und Pflanzengeographie, Universität Heidelberg. Barthlott, W. 1974: Der Rhipsalis micrantha (HBK) DC-Komplex in Ecuador und Peru. - Tropische und Subtropische Pflanzenwelt 10. – Wiesbaden. Barthlott, W. 1976: Chromosome number report on Cactaceae. - IOPB Chromosome Number Reports LIV. Taxon 25: 644-645. Barthlott, W. 1979: Cacti. - Cheltenham. Barthlott, W. 1983: Biogeography and evolution in neo- and palaeotropical Rhipsalinae (Cactaceae). - Pp. 241-248 in: Kubitzki, K. (ed.), Proc. Int. Symp. Dispersal and Distribution. Hamburg. Barthlott, W. 1987a: New names in Rhipsalidinae (Cactaceae). - Bradleya 5: 97-100. Barthlott, W. 1987b: Tribe Rhipsalideae DC. - Pp. 115 in: Epiphytes (McMillan, A. J. S., ed.) Barthlott, W. 1988: Über die systematische Gliederung der Cactaceae. - Beitr. Biol. Pflanzen 63: 17-40. Bauer, R. 2003: A synopsis of the tribe Hylocereeae F.Buxb. - Cactaceae Syst. Init. 17: 3-63. Bauer, R. 2005: More notes on Pfeiffera. - Cactaceae Syst. Init. 20: 6-10. Berger, A. 1926: Die Entwicklungslinien der Kakteen. - Jena. Binski, H. 2002: Pollenmorphologie epiphytischer Kakteen (Rhipsalideae/Cactaceae). – Diploma thesis. Botanisches Institut, Bonn. Borsch, T. & Quandt, D. 2009: Mutational dynamics and phylogenetic utility of noncoding chloroplast DNA. - Plant Syst. Evol. 282: 169-199. Borsch, T., Hilu, K. W., Quandt, D., Wilde, V., Neinhuis, C. & Barthlott, W. 2003: Noncoding plastid trnT-trnF sequences reveal a well resolved phylogeny of basal angiosperms. - J. Evol. Biol. 16: 558-576. 140
References Borsch, T., Hilu, K. W., Wiersema, J. H., Lohne, C., Barthlott, W. & Wilde, V. 2007: Phylogeny of Nymphaea (Nymphaeaceae): Evidence from substitutions and microstructural changes in the chloroplast trnT-trnF region. - Int. J. Plant Sci. 168: 639-671. Borsch, T., Korotkova, N., Raus, T., Lobin, W. & Löhne, C. 2009: The petD group II intron as a genus and species level marker: Utility for tree inference and species identification in the diverse genus Campanula (Campanulaceae) -Willdenowia 39: 7-33. Britton, N. L. & Rose, J. N. 1919-1923: The Cactaceae. Descriptions and illustrations of plants of the cactus family IV. - Washington. Britton, N. L. & Rose, J. N. 1923: Subtribe 8. Rhipsalidanae. - Pp. 208-247 in: The Cactaceae. Descriptions and illustrations of plants of the cactus family IV. Washington. Butterworth, C. A. & Wallace, R. S. 2004: Phylogenetic studies of Mammillaria (Cactaceae) - insights from chloroplast sequence variation and hypothesis testing using the parametric bootstrap. - Am. J. Bot. 91: 1086-1098. Butterworth, C. A. & Wallace, R. S. 2005: Molecular phylogenetics of the leafy cactus genus Pereskia (Cactaceae). - Syst. Bot. 30: 800-808. Butterworth, C. A. 2006: Resolving "Nyffeler's puzzle" - The intriguing taxonomic position of Blossfeldia. – Haseltonia. 12: 3-10. Butterworth, C. A., Cota-Sanchez, J. H. & Wallace, R. S. 2002: Molecular systematics of tribe Cacteae (Cactaceae: Cactoideae): A phylogeny based on rpl16 intron sequence variation. - Syst. Bot. 27: 257-270. Buxbaum, F. 1942: Rhipsaliden-Studien. - Cactaceae 1942: 4-13. Buxbaum, F. 1957-1960: Morphologie der Kakteen. - Pp. Morphologie [1-110] in: Krainz, H. (ed.), Die Kakteen. Stuttgart. Buxbaum, F. 1962: Das phylogenetische System den Cactaceae. - Unpaged in: Krainz, H. (ed.), Die Kakteen VIII. Stuttgart. Buxbaum, F. 1967: Der gegenwärtige Stand der stammesgeschichtlichen Erforschung der Kakteen. - Kakteen And. Sukk. 18: 6-9. Buxbaum, F. 1970a: Gattung Lepismium. - Pp.CIIe in: Krainz, H. (ed.), Die Kakteen X. Stuttgart. Buxbaum, F. 1970b: Gattung Hatiora. - Pp. C IIc in: Krainz, H. (ed.), Die Kakteen VI. Stuttgart. Buxbaum, F. 1970c: Gattung Rhipsalis. - Pp. CIVe in: Krainz, H. (ed.), Die Kakteen VI. Stuttgart. Buxbaum, F. 1971: Gattung Pfeiffera. - Pp. C IIc in: Krainz, H. (ed.), Die Kakteen VI. Stuttgart. Calvente, A. M., Andreata, R. H. P. & Vieira, R. C. 2008: Stem anatomy of Rhipsalis (Cactaceae) and its relevance for taxonomy. - Plant Syst. Evol. 276: 1-7. 141
References Calvente, A., Zappi, D. C., Forest, F. & Lohmann, L. G. 2011: Molecular phylogeny of tribe Rhipsalideae (Cactaceae) and taxonomic implications for Schlumbergera and Hatiora. - Mol. Phylogenet. Evol. 58: 456-468. Calviño, C. I. & Downie, S. R. 2007: Circumscription and phylogeny of Apiaceae subfamily Saniculoideae based on chloroplast DNA sequences. - Mol. Phylogenet. Evol. 44: 175-191. Castoe, T. A., Poole, A. W., Gu, W. J., de Koning, A. P. J., Daza, J. M., Smith, E. N. & Pollock, D. D. 2010: Rapid identification of thousands of copperhead snake (Agkistrodon contortrix) microsatellite loci from modest amounts of 454 shotgun genome sequence. – Mol. Ecol. Res. 10: 341-347. CBOL Plant Working Group. 2009. A DNA barcode for land plants. - Proc. Natl. Acad Sci. USA 106: 12794-12797. Chase, M. W., Cowan, R. S., Hollingsworth, P. M., van den Berg, C., Madriñán, S., Petersen, G., Seberg, O., Jørgsensen, T., Cameron, K. M., Carine, M., Pedersen, N., Hedderson, T. A. J., Conrad, F., Salazar, G. A., Richardson, J. E., Hollingsworth, M. L., Barraclough, T. G., Kelly, L. & Wilkinson, M. 2007: A proposal for a standardised protocol to barcode all land plants. - Taxon 56: 295-299. Clement, M., Posada, D. & Crandall, K. A. 2000: TCS: a computer program to estimate gene genealogies. – Mol. Ecol. 9: 1657-1659. Cota-Sanchez, J. H. & Philbrick, C. T. 1994: Chromosome-number variation and polyploidy in the Genus Echinocereus (Cactaceae). - Am. J. Bot. 81: 1054-1062. Cowan, R. S., Chase, M. W., Kress, W. J. & Savolainen, V. 2006: 300,000 species to identify: problems, progress, and prospects in DNA barcoding of land plants. Taxon 55: 611-616. Croizat, L. 1952: Manual of phytogeography or an account of plant-dispersal throughout the world. - The Hague. Csencsics, D., Brodbeck, S. & Holderegger, R. 2010: Cost-effective, species-specific microsatellite development for the endangered dwarf bulrush (Typha minima) using next-generation sequencing technology. - Journal of Heredity 101: 789-793. Cuénoud, P., Savolainen, V., Chatrou, L. W., Powell, M., Grayer, R. J. & Chase, M. W. 2002: Molecular phylogenetics of Caryophyllales based on nuclear 18S rDNA and plastid rbcL, atpB, and matK DNA sequences. - Am. J. Bot. 89: 132-144. Das, A. B., Mohanty, S. & Das, P. 1999: Karyotype diversity in thirteen species of Mammillaria in the family Cactaceae. - Cytobios 97: 117-125. de Candolle, A. P. 1828: Prodromus systematis naturalis regni vegetabilis 3. - Paris. de Fatima Freitas, M., de Moraes Calvente, A. & Braga, J. M. A. 2009: A new species of Rhipsalis (Cactaceae) from Brazil. - Syst. Bot. 34: 505-509. Delf, E. M. 1912: Transpiration in succulent plants. - Ann. Bot. 26: 409-442. Dettke, G. A. & Milaneze-Gutierre, M. A. 2008: Anatomia caulinar de espécies epífitas de Cactaceae, subfamília Cactoideae. - Hoehnea 35: 583-595.
References Devey, D. S., Chase, M. W. & Clarkson, J. J. 2009: A stuttering start to plant DNA barcoding: microsatellites present a previously overlooked problem in non-coding plastid regions. - Taxon 58: 7-15. Drummond, A. J. & Rambaut, http://beast.bio.ed.ac.uk/.
Edwards, D., Horn, A., Taylor, D., Savolainen, V. & Hawkins, J. A. 2008: DNA barcoding of a large genus, Aspalathus L. (Fabaceae). - Taxon 57: 1317-1327. Edwards, E. J., Nyffeler, R. & Donoghue, M. J. 2005: Basal cactus phylogeny: Implications of Pereskia (Cactaceae) paraphyly for the transition to the cactus life form. - Am. J. Bot. 92: 1177-1188. Eggli U. & Zappi D. C. 2003: Repert. Pl. Succ. 53. Erixon, P. & Oxelman, B. 2008: Reticulate or tree-like chloroplast DNA evolution in Sileneae (Caryophyllaceae)? - Mol. Phylogenet. Evol. 48: 313-325. Fazekas, A. J., Burgess, K. S., Kesanakurti, P. R., Graham, S. W., Newmaster, S. G., Husband, B. C., Percy, D. M., Hajibabaei, M. & Barrett, S. C. H. 2008: Multiple multilocus DNA barcodes from the plastid genome discriminate plant species equally well. - PLoS ONE 3: e2802. Fazekas, A. J., Kesanakurti, P. R., Burgess, K. S., Percy, D. M., Graham, S. W., Barrett, S. C. H., Newmaster, S. G., Hajibabaei, M. & Husband, B. C. 2009: Are plant species inherently harder to discriminate than animal species using DNA barcoding markers? - Molecular Ecology Resources 9: 130-139. Fishbein, M., Kephart, S. R., Wilder, M., Halpin, K. M. & Datwyler, S. L. 2010: Phylogeny of Camassia (Agavaceae) inferred from plastid rpl16 intron and trnDtrnY-trnE-trnT intergenic spacer DNA sequences: Implications for species delimitation. - Syst. Bot. 35: 77-85. Fleischmann, A., Schaferhoff, B., Heubl, G., Rivadavia, F., Barthlott, W. & Muller, K. F. 2010: Phylogenetics and character evolution in the carnivorous plant genus Genlisea A. St.-Hil. (Lentibulariaceae). - Mol. Phylogenet. Evol. 56: 768-783. Fontaneto, D., Hernious, E. A., Boschetti, C., Caprioli, M., Melone, G., Ricci, C. & Barraclough T. G. 2007. Independently evolving species in asexual bdelloid rotifers. PLoS Biology 5: 914-921. Ford, C. S., Ayres, K. L., Toomey, N., Haider, N., Stahl, J. v. A., Kelly, L. J., Wikström, N., Hollingsworth, P. M., Duff, R. J., Hoot, S. B., Cowan, R. S., Chase, M. W. & Wilkinson, M. J. 2009: Selection of candidate coding DNA barcoding regions for use on land plants. - Bot. J. Linn. Soc. 159: 1-11. Ford, V. S., Lee, J., Baldwin, B. G. & Gottlieb, L. D. 2006: Species divergence and relationships in Stephanomeria (Compositae): PgiC phylogeny compared to prior biosystematic studies. - Am. J. Bot. 93: 480-490. Funk, D. J. & Omland, K. E. 2003: Species-level paraphyly and polyphyly: Frequency, causes, and consequences, with insights from animal mitochondrial DNA. – Annu. Rev. Ecol. Evol. Syst. 34: 397-423.
References Ganong, F. W. 1898: Contributions to a knowledge of the morphology and ecology of the Cactaceae: II. The comparative morphology of the embryos and seedlings. Ann. Bot. 12: 423-474. Gibson, A. & Nobel, P. S. 1986: The cactus primer. - Cambridge. Givnish, T. J., Millam, K. C., Berry, P. E. & Sytsma, K. J. 2007: Phylogeny, adaptive radiation, and historical biogeography of Bromeliaceae inferred from ndhF sequence data. - Aliso 23: 3-26. Givnish, T. J., Millam, K. C., Evans, T. M., Hall, J. C., Pires, J. C., Berry, P. E. & Sytsma, K. J. 2004: Ancient vicariance or recent long-distance dispersal? Inferences about phylogeny and South American-African disjunctions in Rapateaceae and Bromeliaceae based on ndhF sequence data. - Int. J. Plant Sci. 165: S35-S54. Glenn, T. C. & Schable, N. A. 2005: Isolating microsatellite DNA loci. – Method. Enzymol. 395: 202-222. Gonzalez, M. A., Baraloto, C., Engel, J., Mori, S. A. & Pétronelli, P. 2009: Identification of amazonian trees with DNA barcodes. - PLoS ONE 4: e7483. Gorelick, R. 2004: Resolving the phylogenetic placement of Blossfeldia liliputana (Gactaceae): reticulate evolution, chloroplast inheritance, and graft-chimeras. Bradleya 22: 9-14. Guindon, S. & Gascuel, O. 2003: A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. - Syst. Biol. 52: 696-704. Hartl, W. P. 2000: Anatomische, kristallographische und palynologische Untersuchungen an Rhipsalideae (Cactaceae) - Ein Beitrag zur Systematik der Tribus. – PhD thesis, Botanisches Institut, Mathematisch-Naturwissenschaftliche Fakultät, Universität Bonn. Hartl, W. P., Barbier, B., Klapper, H., Müller, P. & Barthlott, W. 2003: Dimorphism of calcium oxalate crystals in stem tissues of Rhipsalideae (Cactaceae) - a contribution to the systematics and taxonomy to the tribe. - Bot. Jahrb. Syst. 124: 287-302. Hebert, P. D. N., Cywinska, A., Ball, S. L. & DeWaard, J. R. 2003: Biological identifications through DNA barcodes. - Proc. R. Soc. Lond. Ser. B-Biol. Sci. 270: 313-321. Hernández-Hernández, T., Hernández, H. M., De-Nova, J. A., Puente, R., Eguiarte, L. E. & Magallón, S. 2011: Phylogenetic relationships and evolution of growth form in Cactaceae (Caryophyllales, Eudicotyledoneae). - Am. J. Bot. 98: 44-61. Hernández-Hernández, T., Hernández, H. M., De-Nova, J. A., Puente, R., Eguiarte, L. E. & Magallón, S. 2011: Phylogenetic relationships and evolution of growth form in Cactaceae (Caryophyllales, Eudicotyledoneae). - Am. J. Bot. 98: 44-61. Hilu, K. W. & Liang, H. P. 1997: The matK gene: Sequence variation and application in plant systematics. - Am. J. Bot. 84: 830-839. Hollingsworth, M. L., Andra Clark, A., Forrest, L. L., Richardson, J., Pennington, R. T., Long, D. G., Cowan, R., Chase, M. W., Gaudeul, M. & Hollingsworth, P. M. 144
References 2009: Selecting barcoding loci for plants: evaluation of seven candidate loci with species-level sampling in three divergent groups of land plants. - Molecular Ecology Resources 9: 439-457. Huelsenbeck, J. P. & Ronquist, F. 2001: MrBayes: Bayesian inference of phylogenetic trees. - Bioinformatics 17: 754-755. Hunt D. & Taylor N. P. (ed.) 2002: Notulae systematicae Lexicon Cactacearum spectantes II. – Cactaceae. Syst. Init. 14: 7-19. Hunt D. 1998: CCI Workshop at Milborne Port, 5.-8. April 1998. – Cactaceae Cons. Init. 5: 1 – 4. Hunt D. 2003: Reductio ad absurdum. – Cactaceae. Syst. Init. 15: 2 –3. Hunt, D. & Taylor, N. P. 1986: The genera of the Cactaceae: towards a new consensus. - Bradleya 4: 65-78. Hunt, D. & Taylor, N. P. 1990: The genera of the Cactaceae: progress towards consensus. - Bradleya 8: 85-107. Hunt, D. 1967: Cactaceae. - Pp. in: Hutchinson, J. (ed.), The genera of flowering plants. Dicotyledones Vol. II. Oxford. Hunt, D. 1969: Contributions to the Flora of Tropical America: LXXVII: A Synopsis of Schlumbergera Lem. (Cactaceae). - Kew Bull. 23: 255-263. Hunt, D. 1999: CITES Cactaceae Checklist. - London. Hunt, D. 2006: The New Cactus Lexicon. - Milborne Port. Ibisch, P. L., Kessler, M., Nowicki, S. & Barthlott, W. 2000: Ecology, biogeography and diversity of the Bolivian epiphytic cacti – with the description of two new taxa. Bradleya 18: 2-30. Ito, Y., Ohi-Toma, T., Murata, J. & Tanaka, N. 2010: Hybridization and polyploidy of an aquatic plant, Ruppia (Ruppiaceae), inferred from plastid and nuclear DNA phylogenies. - Am. J. Bot. 97: 1156-1167. Jakob, S. S. & Blattner, F. R. 2006: A chloroplast genealogy of Hordeum (Poaceae): Long-Term persisting haplotypes, incomplete lineage sorting, regional extinction, and the consequences for phylogenetic inference. - Mol. Biol. Evol. 23: 1602-1612. Johnson, L. A. & Soltis, D. E. 1995: Phylogenetic inference in Saxifragaceae s.str. and Gilia (Polemoniaceae) using trnK sequences. - Ann. Missouri Bot. Gard. 82: 149175. Jones, L. A. 2011: Anatomical adaptations of four Crassula species to water availability. - Bioscience Horizons 4: 13-22. Kelchner, S. A. & Clark, L. G. 1997: Molecular evolution and phylogenetic utility of the chloroplast rpl16 intron in Chusquea and the Bambusoideae (Poaceae). - Mol. Phylogenet. Evol. 8: 385-397. Kelchner, S. A. & Wendel, J. F. 1996: Hairpins create minute inversions in non-coding regions of chloroplast DNA. - Curr. Genet. 30: 259-262.
References Kelchner, S. A. 2002: Group II introns as phylogenetic tools: Structure, function, and evolutionary constraints. - Am. J. Bot. 89: 1651-1669. Kelchner, S.A. 2000. The evolution of non-coding chloroplast DNA and its application in plant systematics. – Ann. Missouri Bot. Gard. 87: 482-498. Kim, S.-C., Crawford, D. J., Jansen, R. K. & Santos-Guerra, A. 1999: The use of a noncoding region of chloroplast DNA in phylogenetic studies of the subtribe of the subtribe Sonchinae (Asteraceae: Lactuceae) -Plant Syst. Evol. 215: 85-99. Kimnach, M. 1983: A revision of Acanthorhipsalis. - Cact. Succ. J. (Los Angeles). 55: 177-182. Kimnach, M. 1984: Rhipsalis brevispina. - Cact. Succ. J. (Los Angeles). 56: 122-124. Kimnach, M. 1990: A new epiphytic cactus: Trichocereus arboricola. - Cact. Succ. J. (Los Angeles). 62: 3-5. Korotkova, N., Zabel, L., Quandt, D. & Barthlott, W. 2010: A phylogenetic analysis of Pfeiffera and the reinstatement of Lymanbensonia as an independently evolved lineage of epiphytic Cactaceae within a new tribe Lymanbensonieae. - Willdenowia 40: 151-172. Kress, W. J. & Erickson, D. L. 2007: A two-locus global DNA barcode for land plants: The coding rbcL gene complements the non-coding trnH-psbA spacer region. PLoS ONE 2: e508. Kress, W. J. & Erickson, D. L. 2007: A two-locus global DNA barcode for land plants: The coding rbcL gene complements the non-coding trnH-psbA spacer region. PLoS ONE 2: e508. Kress, W. J. 1989: The systematic distribution of vascular epiphytes. - Pp. 234-261 in: Lüttge, U. (ed.), Vascular plants as epiphytes. Berlin. Kress, W. J., Wurdack, K. J., Zimmer, E. A., Weigt, L. A. & Janzen, D. H. 2005: Use of DNA barcodes to identify flowering plants. - Proc. Natl. Acad Sci. USA 102: 83698374. Lahaye, R., van der Bank, M., Bogarin, D., Warner, J., Pupulin, F., Gigot, G., Maurin, O., Duthoit, S., Barraclough, T. G. & Savolainen, V. 2008: DNA barcoding the floras of biodiversity hotspots. - Proc. Natl. Acad Sci. USA 105: 2923-2928. Larkin, M. A., Blackshields, G., Chenna, R., McGettigan, P. A., McWilliam, H., Valentin, F., Wallace, I. M., Wilm, A., Lopez, R., Thompson, J. D., Gibson, T. J. & Higgins, D. G. 2007: Clustal W and Clustal X version 2.0. - Bioinformatics 23: 2947-2948. Lee, R. M., Thimmapuram, J., Thinglum, K. A., Gong, G., Hernandez, A. G., Wright, C. L., Kim, R. W., Mikel, M. A. & Tranel, P. J. 2009: Sampling the Waterhemp (Amaranthus tuberculatus) genome using pyrosequencing technology. - Weed Science 57: 463-469. Leuenberger, B. E. 1976: Die Pollenmorphologie der Cactaceae und ihre Bedeutung für die Systematik. - Dissertationes Botanicae 31. - Vaduz.
References Little, D. P. & Stevenson, D. W. 2007: A comparison of algorithms for the identification of specimens using DNA barcodes: examples from gymnosperms. Cladistics 23: 1-21. Löhne, C. & Borsch, T. 2005: Molecular evolution and phylogenetic utility of the petD group II intron: A case study in basal angiosperms. - Mol. Biol. Evol. 22: 317-332. Löhne, C., Borsch, T. & Wiersema, J. H. 2007: Phylogenetic analysis of Nymphaeales using fast-evolving and noncoding chloroplast markers. - Bot. J. Linn. Soc. 154: 141-163. Maddison, W. P. & Maddison, D. R. 2010: Mesquite: a modular system for evolutionary analysis. -. http://mesquiteproject.org. Mathews, D. H., Zuker, M. & Turner, D. H. 1996+: RNAstructure 5.0 http://rna.urmc.rochester.edu/RNAstructure.html. Mayr, E. 1942: Systematics and the origin of species. - New York. McMillan, A. J. S. & Horobin, J. F. 1995: Christmas Cacti. The genus Schlumbergera and its hybrids. - Milborne Port. McNeill J., Barrie F. R ., Burdet H. M ., Demoulin V., Hawksworth D. L., Marhold K., Nicolson D. H ., Prado J., Silva P. C., Skog J. E ., Wiersema J. H . & Turland N. J . (ed.) 2006: International Code of Botanical Nomenclature (Vienna Code) adopted by the Seventeenth International Botanical Congress Vienna, Austria, July 2005. – Regnum Veg. 146. Meglécz, E., Costedoat, C., Dubut, V., Gilles, A., Malausa, T., Pech, N. & Martin, J.-F. 2010: QDD: a user-friendly program to select microsatellite markers and design primers from large sequencing projects. - Bioinformatics 26: 403-404. Metzing, D. & Kiesling, R. 2008: The study of cactus evolution: The pre-DNA era. – Haseltonia 14: 6-25. Meyer, C. P. & Paulay, G. 2005: DNA barcoding: Error rates based on comprehensive sampling. - PLoS Biology 3: e422. Minder, A. M. & Widmer, A. 2008: A population genomic analysis of species boundaries: neutral processes, adaptive divergence and introgression between two hybridizing plant species. – Mol. Ecol. 17: 1552-1563. Monaghan, M. T., Wild, R., Elliot, M., Fujisawa, T., Balke, M., Inward, D. J. G., Lees, D. C., Ranaivosolo, R., Eggleton, P., Barraclough, T. G. & Vogler, A. P. 2009: Accelerated species inventory on Madagascar using coalescent-based models of species delineation. - Syst. Biol. 58: 298-311. Moran, R. 1953: Taxonomic Studies in the Cactaceae. II-Notes on Schlumbergera, Rhipsalidopsis and allied genera. - Gentes Herbarium VIII Fasc. IV: 328-345. Morrison, D. 2009: A framework for phylogenetic sequence alignment. - Plant Syst. Evol. 282: 127-149. Müller, J., Müller, K. F., Neinhuis, C. & Quandt, D. 2005+: PhyDE: Phylogenetic Data Editor. – www.phyde.de. -.
References Müller, K. F. & Borsch, T. 2005: Phylogenetics of Amaranthaceae based on trnK/matK sequence data - Evidence from Parsimony, likelihood, and Bayesian analyses. Ann. Missouri Bot. Gard. 92: 66-102. Müller, K. F. 2004: PRAP - computation of Bremer support for large data sets. - Mol. Phylogenet. Evol. 31: 780-782. Müller, K. F. 2005a: The efficiency of different search strategies in estimating parsimony jackknife, bootstrap, and Bremer support. - BMC Evol. Biol. 5: 58. Müller, K. F. 2005b: SeqState: Primer design and sequence statistics for phylogenetic DNA datasets. - Appl. Bioinformatics 4: 65-69. Müller, K. F., Borsch, T. & Hilu, K. W. 2006: Phylogenetic utility of rapidly evolving DNA at high taxonomical levels: Contrasting matK, trnT-F, and rbcL in basal angiosperms. - Mol. Phylogenet. Evol. 41: 99-117. Negron-Ortiz, V. 2007: Chromosome numbers, nuclear DNA content, and polyploidy in Consolea (Cactaceae), an endemic cactus of the Caribbean islands. - Am. J. Bot. 94: 1360-1370. Newmaster, S. G., Fazekas, A. J. & Ragupathy, S. 2006: DNA barcoding in land plants: evaluation of rbcL in a multigene tiered approach. – Can. J. Bot. 84: 335341. Newmaster, S. G., Fazekas, A. J., Steeves, R. A. D. & Janovec, J. 2008: Testing candidate plant barcode regions in the Myristicaceae. – Mol. Ecol. Res. 8: 480-490. Nixon, K. C. 2002: Winclada (BETA) ver. 0.9.9. - published by the author, Ithaca, NY. Nyffeler, R. & Eggli, U. 2010a: Disintegrating Portulacaceae: A new familial classification of the suborder Portulacineae (Caryophyllales) based on molecular and morphological data. - Taxon 59: 227-240. Nyffeler, R. & Eggli, U. 2010b: A farewell to dated ideas and concepts – molecular phylogenetics and a revised suprageneric classification of the family Cactaceae. Schumannia 6: 109-151. Nyffeler, R. 2000: Should Pfeiffera be resurrected? - Cactaceae Syst. Init. 10: 10-11. Nyffeler, R. 2002: Phylogenetic relationships in the cactus family (Cactaceae) based on evidence from trnK/matK and trnL-trnF sequences. - Am. J. Bot. 89: 312-326. Nyffeler, R. 2007: The closest relatives of Cacti: insights from phylogenetic analyses of chloroplast and mitochondrial sequences with special emphasis on relationships in the tribe Anacampseroteae. - Am. J. Bot. 94: 89-101. Ocampo, G. & Columbus, J. T. 2010: Molecular phylogenetics of suborder Cactineae (Caryophyllales), including insights into photosynthetic diversification and historical biogeography. - Am. J. Bot. 97: 1827-1847. Ochoterena, H. 2009: Homology in coding and non-coding DNA sequences: a parsimony perspective. - Plant Syst. Evol. 282: 151-168. Pagel, M., Meade, A. & Barker, D. 2004: Bayesian estimation of ancestral character states on phylogenies. - Syst. Biol. 53: 673 - 684. 148
References Pimienta-Barrios, E. & del Castillo, R. F. 2002: Reproductive biology. - Pp. 75-90 in: Nobel, P. S. (ed.), Cacti. Biology and uses. Berkeley, California. Pinkava, D. & McLeod, M. 1971: Chromosome numbers in some cacti of western North America. - Brittonia 23: 171-176. Pinkava, D. J., Rebman, J. P. & Baker, M. A. 1998: Chromosome numbers in some cacti of western North America-VII. - Haseltonia: 32-41. Pons, J., Barraclough, T. G., Gomez-Zurita, J., Cardoso, A., Duran, D. P., Hazell, S., Kamoun, S., Sumlin, W. D. & Vogler, A. P. 2006: Sequence-based species delimitation for the DNA taxonomy of undescribed insects. - Syst. Biol. 55: 595609. Porembski, S. & Barthlott, W. 1999: Pitcairnia feliciana: the only indigenous African Bromeliaceae. – Harvard Papers in Botany 4: 175-184. Porsch, O. 1938: Das Bestäubungsleben der Kakteenblüte. - Jahrbuch der Deutschen Kakteengesellschaft: 1 -18. Porsch, O. 1939: Das Bestäubungsleben der Kakteenblüte. - Jahrbuch der Deutschen Kakteengesellschaft: 81-142. Posada, D. 2008: jModelTest: Phylogenetic Model Averaging. - Mol. Biol. Evol. 25: 1253-1256. Quandt, D. & Stech, M. 2004: Molecular evolution and phylogenetic utility of the chloroplast trnT-trnF region in bryophytes. - Plant Biol. 6: 545-554. Quandt, D., Müller, K. & Huttunen, S. 2003: Characterisation of the chloroplast DNA psbT-H region and the influence of dyad symmetrical elements on phylogenetic reconstructions. - Plant Biol. 5: 400-410. Rambaut, A. & Drummond, A. J. 2007: Tracer v1.5. -. http://beast.bio.ed.ac.uk/Tracer Rieseberg, L. H. & Brouillet, L. 1994: Are many plant species paraphyletic? - Taxon 43: 21-32. Rieseberg, L. H. & Willis, J. H. 2007: Plant Speciation. - Science 317: 910-914. Ritz, C. M., Martinss, L., Mecklenburg, R., Goremykin, V. & Hellwig, F. H. 2007: The molecular phylogeny of Rebutia (Cactaceae) and its allies demonstrates the influence of paleogeography on the evolution of South American mountain cacti. Am. J. Bot. 94: 1321–1332. Rose, M.-J. & Barthlott, W. 1994: Coloured pollen in Cactaceae - a mimetic adaptation to hummingbird-pollination? – Bot. Acta 107: 402-406. Ross, R. 1981: Chromosome counts, cytology, and reproduction in the Cactaceae. - Am. J. Bot. 68: 463-470. Rozen, S. & Skaletsky, H. 2000: Primer3 on the WWW for general users and for biologist programmers. - Methods Mol. Biol. 132: 365-386. Salm-Dyck, J. z. 1845: Cacteae in Horto Dyckensi Cultae. Anno 1845. - Paris. Salm-Dyck, J. z. 1850: Cacteae in Horto Dyckensi Cultae Anno 1849. - Bonn. 149
References Salzburger, W., Ewing, G. B. & Von Haeseler, A. 2011: The performance of phylogenetic algorithms in estimating haplotype genealogies with migration. – Mol. Ecol. 20: 1952-1963. Sánchez del-Pino, I., Borsch, T. & Motley, T. J. 2009: trnL-F and rpl16 sequence data and dense taxon sampling reveal monophyly of unilocular anthered Gomphrenoideae (Amaranthaceae) and an improved picture of their internal relationships. - Syst. Bot. 34: 57-67. Sang, T., Crawford, D. J. & Stuessy, T. F. 1997: Chloroplast DNA phylogeny, reticulate evolution, and biogeography of Paeonia (Paeoniaceae). - Am. J. Bot. 84: 1120-1136. Schäferhoff, B., Müller, K. F. & Borsch, T. 2009: Caryophyllales phylogenetics: disentangling Phytolaccaceae and Molluginaceae and description of Microteaceae as a new isolated family. - Willdenowia 39: 209-228. Schumann, K. M. 1899: Gesamtbeschreibung Cactacearum). - Neudamm.
Seberg, O. & Petersen, G. 2009: How many loci does it take to DNA barcode a Crocus? - PLoS ONE 4: e4598. Shaw, J., Lickey, E. B., Beck, J. T., Farmer, S. B., Liu, W., Miller, J., Siripun, K. C., Winder, C. T., Schilling, E. E. & Small, R. L. 2005: The tortoise and the hare II: relative utility of 21 noncoding chloroplast DNA sequences for phylogenetic analysis. - Am. J. Bot. 92: 142-166. Shaw, J., Lickey, E. B., Schilling, E. E. & Small, R. L. 2007: Comparison of whole chloroplast genome sequences to choose noncoding regions for phylogenetic studies in angiosperms: the tortoise and the hare III. - Am. J. Bot. 94: 275-288. Simmons, M. P. & Ochoterena, H. 2000: Gaps as characters in sequence-based phylogenetic analyses. - Syst. Biol. 49: 369–381. Steele, P. R., Friar, L. M., Gilbert, L. E. & Jansen, R. K. 2010: Molecular systematics of the neotropical genus Psiguria (Cucurbitaceae): implications for phylogeny and species identification. - Am. J. Bot. 97: 156-173. Stöver, B. & Müller, K. 2010: TreeGraph 2: Combining and visualizing evidence from different phylogenetic analyses. - BMC Bioinformatics 11: 7. Suzuki, Y., Glazko, G. V. & Nei, M. 2002: Overcredibility of molecular phylogenies obtained by Bayesian phylogenetics. - Proc. Natl. Acad Sci. USA 99: 16138-16143. Swofford, D. L. 1998: PAUP*. Phylogenetic Analysis Using Parsimony (*and other Methods). - Sunderland, Massachussets. Syring, J., Farrell, K., Businský, R., Cronn, R. & Liston, A. 2007: Widespread genealogical nonmonophyly in species of Pinus subgenus Strobus. - Syst. Biol. 56: 163-181. Taberlet, P., Coissac, E., Pompanon, F., Gielly, L., Miquel, C., Valentini, A., Vermat, T., Corthier, G., Brochmann, C. & Willerslev, E. 2007: Power and limitations of the chloroplast trnL (UAA) intron for plant DNA barcoding. - Nucleic Acids Research 35: e14. 150
References Takhtajan, A. 1972: Pattern of ontogenetic alterations in the evolution of higher plants. - Phytomorphology XXII: 164-171. Tangphatsornruang, S., Somta, P., Uthaipaisanwong, P., Chanprasert, J., Sangsrakru, D., Seehalak, W., Sommanas, W., Tragoonrung, S. & Srinives, P. 2009: Characterization of microsatellites and gene contents from genome shotgun sequences of mungbean (Vigna radiata (L.) Wilczek). - BMC Plant Biol. 9: 137. Taylor, N. P. & Zappi, D. C. 2004: Cacti of Eastern Brazil. - London. Templeton, A. R., Crandall, K. A. & Sing, C. F. 1992: A cladistic analysis of phenotypic associations with haplotypes inferred from restriction endonuclease mapping and DNA sequence data. III. Cladogram estimation. - Genetics 132: 619-633. Tesfaye, K., Borsch, T., Govers, K. & Bekele, E. 2007: Characterization of Coffea chloroplast microsatellites and evidence for the recent divergence of C. arabica and C. eugenioides chloroplast genomes. - Genome 50: 1112-1129. Vaupel, F. 1925-1926: Die Kakteen. Monographie der Cactaceae. - Berlin. Vöchting, H. 1873: Beiträge zur Morphologie und Anatomie der Rhipsalideen. – PhD thesis. Philosophische Honoren-Fakultät, Göttingen. von Willert, D. J., Eller, B. M., Werger, M. J. A., Brinckmann, E. & Ehlenfeldt, H. D. 1992: Life strategies of succulents in deserts: With special reference to the Namib desert. - Cambridge. Wallace, R. S. & Gibson, A. C. 2002: Evolution and Systematics. - Pp. 1-21 in: Nobel, P. S. (ed.), Cacti. Biology and uses. - Berkeley. Wallace, R. S. 1995: Molecular systematic study of the Cactaceae: Using chloroplast DNA variation to elucidate Cactus phylogeny. - Bradleya 13: 1-12. Whitlock, B. A., Hale, A. M. & Groff, P. A. 2010: Intraspecific inversions pose a challenge for the trnH-psbA plant DNA barcode. - PLoS ONE 5: e11533. Wicke, S. & Quandt, D. 2009: Universal primers for the amplification of the plastid trnK/matK region in land plants. - Anal. Jardin Bot. Madrid. 66: 285-288. Wood, T. E. & Nakazato, T. 2009: Investigating species boundaries in the Giliopsis group of Ipomopsis (Polemoniaceae): Strong discordance among molecular and morphological markers. - Am. J. Bot. 96: 853-861. Worberg, A. 2009: Non-coding and fast-evolving chloroplast genomic regions and their utility for reconstructing evolutionary relationships among eudicots: towards resolving the radiation of the rosids. PhD Thesis, Nees Institute for Biodiversity of Plants, Universität Bonn. Zhang, J., Wang, J., Xia, T. & Zhou, S. 2009: DNA barcoding: species delimitation in tree peonies. - Science in China Series C: Life Sciences 52: 568-578.
Summary Korotkova, Nadja. 2011. Phylogeny and evolution of the epiphytic Rhipsalideae (Cactaceae). PhD thesis, Mathematisch-Naturwissenschaftliche Fakultät, Rheinische Friedlich-Wilhelms-Universität Bonn, Germany. Cactaceae are one of the major floristic components of the New World’s arid as well as seasonally moist tropical regions and at the same time one of the most popular plant families in horticulture. The taxonomic units (tribes, genera) and species limits in the Cactaceae have been difficult to define due to intergrading vegetative characters, phenotypic plasticity and the largely uniform flower morphology. Molecular phylogenetic studies so far yielded largely unresolved or poorly supported trees so relationships within Cactaceae remained insufficiently understood. Besides, Cactaceae taxonomy is still often unreliable. But a high proportion of cacti is CITESlisted and accurate species delimitation and identification are therefore desirable for conservation Red List assessments. This study focuses on the Rhipsalideae, a predominantly epiphytic tribe of Cactaceae from the tropical rainforests of South and Central America. The Rhipsalideae have hitherto not been subject of a detailed phylogenetic study so far but are well-suited for this purpose: they are a comparatively small group and are well known morphologically. All but one species were available for this study so it is one of the most comprehensive species-level studies carried out within the Cactaceae so far. The major aims of this study were to resolve species-level relationships in the Rhipsalideae, also to get better insights into species limits and to find morphological characters synapomorphic or at least characteristic for the genera and subgenera. In order to resolve relationships between so closely related species, rapidly evolving plastid markers with high phylogenetic structure were selected. The phylogenetic relationships were analysed using sequence data from intergenic spacers (psbA-trnH, rps3-rpl16, trnS-trnG, trnQ-rps16), group II introns (trnK, rpl16, trnG) and the coding region matK. Trees were inferred with Maximum Parsimony and Bayesian Inference. Haplotype network construction was carried out for examining patterns within Rhipsalis baccifera and allies. First, the position and circumscription of the genus Pfeiffera was addressed. It had formerly been included in the Rhipsalideae but earlier studies showed it to be distantly related. A dataset of seven regions was generated with c. 7000 nucleotides sequenced per sample. All but one Pfeiffera species with multiple accessions were sampled. Detailed phylogenetic analyses of this study revealed Pfeiffera polyphyletic, comprising two unrelated clades, both well resolved and highly supported. One clade includes the type species, P. ianthothele; the second contains two Pfeiffera and one 153
Summary erstwhile Lepismium species. These results and a re-evaluation of the morphological characters justify a generic status for this newly found clade. It includes the type species of the earlier-proposed monotypic genus Lymanbensonia and, therefore, its reinstatement is proposed in an amplified circumscription. A further taxonomic and nomenclatural consequence is the establishment of a separate tribe Lymanbensonieae, formally proposed here, to contain the genera Lymanbensonia and Calymmanthium. The results further underscore that epiphytism evolved more frequently in Cactaceae than hitherto assumed. To resolve phylogenetic relationships in the Rhipsalideae, a dataset of six regions was generated with c. 4200 nucleotides sequenced per sample for 120 accessions. The regions used were evaluated for their phylogenetic performance and species discrimination power for DNA based species recognition (DNA barcoding) based on beforehand defined operational taxonomic units (OTUs). The Rhipsalideae were found as monophyletic and contain five major clades that correspond to the genera Rhipsalis, Lepismium, Schlumbergera, Hatiora subg. Hatiora and Hatiora subg. Rhipsalidopsis. The relationships between the major clades corresponding to genera could not be clarified. But the species-level relationships were well resolved and supported. Based on the results, a reinstatement of Rhipsalidopsis at generic level and a revised subgeneric classification for Rhipsalis are proposed. Already c. 2500 nt of four regions (rpl16 intron, trnK intron, psbA-trnH, trnQrps16) were sufficient to identify 97% of the OTUs in the Rhipsalideae. Among all possible marker combinations this one was the most successful with the least number of sequenced nucleotides. The combination of all markers (4207 nt) yielded the same number of identified OTUs. The rpl16 intron was the best single-locus barcode, finding 60% of the OTUs. The two markers providing the best phylogenetic signal for the Rhipsalideae were the group II introns in rpl16 and trnK. The phylogenetic performance of the markers was found to be not determined by the level of sequence variability. Comparisons of the OTU identification potential of the markers with their phylogenetic performance revealed that these two qualities are not necessarily correlated. The reliable phylogenetic hypothesis for the Rhipsalideae provided a framework for a detailed study of character evolution. A matrix of 36 characters was compiled and ancestral states were reconstructed using a Bayesian approach. A focus was put on the characters associated with the epiphytic life form and the floral traits. The degree of homoplasy was found to be high but many characters were homogenous within the clades and all the highly supported clades (genera, subgenera) found by the molecular phylogenetic analyses could also be defined morphologically. Rhipsalis baccifera is the most widespread cactus and the only cactus native to Africa. To get more insights into the relationships between the South American and 154
Summary the African populations, the distribution patterns of Rhipsalis baccifera were analysed. Tree building methods and haplotype network algorithms were applied to sequences of the rps3-rpl16 spacer and the rpl16 intron. Two main groups of plastid haplotypes were found: a northern South American / Caribbean / Central American haplotype and an African haplotype. These results suggest a single dispersal of Rhipsalis baccifera to Africa and reveal high genetic diversity within its populations on both continents. To obtain further resolution among the populations, microsatellite markers for Rhipsalis baccifera have been developed using 454 sequencing. The analyses resulted in almost completely resolved and well supported species level trees which were hitherto hardly achieved in the Cactaceae. This study could therefore serve as a case study for resolution of species-level relationships between closely related and recently diverged species, in other Cactaceae groups or in other plant families that pose similar problems. The results also lead to the conclusion that morphology-based taxonomic units can be misleading to guide taxon sampling and the best solution is to sample the study group as completely as possible for a reliable phylogeny inference. This study is also the first DNA barcoding study for the Cactaceae. The identification success here is higher than observed in other studies that also used a taxonomic setting and can serve as an example for future studies. The results furthermore emphasize that the outcome of a phylogenetic study and a barcoding study will largely depend on the markers chosen. So far, plastid markers have provided a solid phylogenetic hypothesis for the Rhipsalideae which is also in line with morphological characters. But hybridization is common in Cactaceae (although supposed to be rare in the Rhipsalideae). Future work should aim at including nuclear markers which are so far hardly applied in Cactaceae.