﻿Generic concepts and species diversity within the Gynoxyoid clade (Senecioneae, Compositae)

﻿Abstract The Gynoxyoid clade of the Senecioneae (Asteraceae) until now included the five genera Aequatorium, Gynoxys, Nordenstamia, Paracalia and Paragynoxys as diagnosed using selected morphological characters. In their pre-phylogenetic circumscription, the genera Aequatorium and Paragynoxys were considered to inhabit the northern Andes in contrast to Nordenstamia and Paracalia that occur in the central Andes. The most species-rich genus, Gynoxys, was believed to be distributed throughout the Andes. We use a recently established plastid phylogenomic framework that rendered Gynoxys paraphyletic to further evaluate the delimitation of genera in the Gynoxyoid clade. We examine the morphological variation of all members of the Gynoxyoid to identify characters potentially informative at genus level. This results in a matrix of eleven, mostly multistate characters, including those originally used to diagnose these genera. The ancestral character state inference displays a high level of homoplasy, but nevertheless supports the recognition of four genera. Aequatorium is characterised by white radiate capitula. Paracalia and Paragynoxys share white flowers and floral characteristics, such as flower opening and length of disc flowers lobes, as plesiomorphic states, but differ in habit (scandent shrubs vs. trees). Paracalia also retained white flowers, but its two species are characterised by the absence of outer phyllaries. The genera Gynoxys and Nordenstamia comprise species with yellow capitula which appear to be a derived feature in the Gynoxyoids. The genus Nordenstamia, with eight species, is synonymised under Gynoxys since molecular evidence shows its species nested within various parts of the Gynoxys subclade and the morphological variation of Nordenstamia falls well within that of Gynoxys. With the goal to assign all species to four genera (Aequatorium, Gynoxys, Paracalia and Paragynoxys), we assess the states for the eleven characters for all members of the Gynoxyoids and generate new ETS and ITS sequences for 171 specimens belonging to 49 species to further support their generic placement. We provide a taxonomic treatment for the four genera recognised here including amended diagnoses and morphological descriptions. Furthermore, a species-level taxonomic backbone is elaborated for all genera using electronic tools that list 158 currently accepted names and synonyms (209 names in total) with the respective protologue and type information, as well as notes on the current understanding of species limits. Eleven names are newly synonymised, two are lectotypified and eight are newly transferred to other genera.


Introduction
The Gynoxyoid group is a New World clade of the subtribe Tussilagininae (Senecioneae, Asteraceae) that was estimated to comprise around 150 species in five genera (Nordenstam et al. 2009).The clade includes shrubs, trees and, more rarely, lianas, growing at the higher elevations of the Andes, in humid mountain forests, subalpine forests and in the paramo.Originally, Jeffrey (1992) suggested the existence of this group of putatively related genera, based on cylindrical anther-collars, polar endothecial thickening and high chromosome numbers, based on x = 10.He included Capelio B.Nord.(as Alciope DC.) from South Africa (Nordenstam 2002), the Andean genera Paracalia Cuatrec., Paragynoxys (Cuatrec.)Cuatrec., Gynoxys Cass.and Aequatorium B.Nord. and the Caribbean genus Herodotia Urb.& Eckm.Subsequently, Robinson et al. (1997) restricted the group to the South American genera and pointed out that it is characterised by a chromosome number of 2n = 80.The Roldana clade, sister to the Gynoxys clade (Pelser et al. 2007(Pelser et al. , 2010)), in contrast, has a chromosome number of 2n = 60 (Jeffrey 1992).These high chromosome numbers can be explained by ancient polyploidisation in the Tussilagiinae.The genus Nordenstamia Lundin was later erected to accommodate certain species previously placed in Aequatorium and Gynoxys (Lundin 2006).
The first phylogenetic data for the Gynoxyoid group were provided by Pelser et al. (2007) in the context of inferring relationships within the Senecioneae, based on sequences of the nrITS region.The authors resolved a clade with the genera Aequatorium, Gynoxys, Nordenstamia and Paragynoxys and found Nordenstamia (2 species) nested within Gynoxys (4 species).Pelser et al. (2010) extended the taxon sampling with a representative of Paracalia and increased the number of molecular markers (nrITS and nrETS and plastid ndhF,rbcL,5′ and 3′ trnK, and essentially confirmed their earlier results.Recently, Escobari et al. (2021) provided a comprehensive plastid phylogenomic framework, based on 17 complete plastid genomes representing all five genera and close American relatives within the Tussilagiinae.Their results corroborated the Gynoxyoid group as monophyletic with high support.The three representatives of the genus Nordenstamia were found nested within a broadly paraphyletic genus Gynoxys.Additionally, the plastid genome sequence of Paracalia jungioides appeared as sister to G. baccaroides and G. violacea within Gynoxys, whereas P. pentamera was retrieved as sister to all other members of the Gynoxyoids.The second diverging clade was comprised of the monophyletic Paragynoxys and the only representative of the genus Aequatorium.
The Gynoxyoid group represents one of the speciose Andean plant lineages and, thus, contributes significantly to the high species diversity and endemism in the Andes as one of the global biodiversity hotspots (Myers et al. 2000;Padilla-Gonzalez et al. 2021).The uplift of the Andes led to shifts in ecosystem barriers (Luebert and Weigend 2014;Bacon et al. 2022) and enabled the creation of new habitats (Colwell et al. 2008;Moreira-Munoz et al. 2020;Perez-Escobar et al. 2022) which seem to have triggered rapid speciation of Andean plants (e.g.Madriñán et al. 2013;Zhang et al. 2021;Perez-Escobar et al. 2022).Amongst the studies focusing on the evolution of Andean plant groups (see Hughes and Atchison (2015)), several dealt with genera of the sunflower family, such as Diplostephium Kunth (Vargas et al. 2017), Espeletia Mutis ex Bonpl.(Pouchon et al. 2018) and Loricaria Wedd.(Kandziora et al. 2022).In all three cases, the authors reported low genetic distances, complicating the study of species relationships and species limits.Moreover, frequent events of reticulate evolution and incomplete lineage sorting have been reported from rapidly evolving Andean plant groups (Garcia et al. 2014;Vargas et al. 2017;Schley et al. 2021;Kandziora et al. 2022).Low genetic distances were also observed amongst plastid genomes in the Gynoxyoid clade in our previous study (Escobari et al. 2021).Consequently, we demonstrated that complete plastid genome sequences, including the more variable intron and spacer partitions, were needed to achieve resolution at species and even genus level.The results of Escobari et al. (2021) underscored that Gynoxys is not monophyletic as currently circumscribed and that an evaluation of morphological characters hitherto used to diagnose the genera of the Gynoxyoid clade in an evolutionary context is warranted.Cassini (1827) described the genus Gynoxys as having a tree-like habit, opposite leaves, the presence of an indumentum on the lower leaf surface, corymbiform capitula and the apex of style branches vested by papillose hairs as diagnostic characters.Weddell (1855) subdivided Gynoxys into two sections: one with radiate and the other with discoid capitula, which has lately been adopted by Correa (2003).The first taxonomic treatment including a larger number of species was made by Herrera (1980) who dealt with the 30 species distributed in Peru.That author redefined the genus by having usually opposite leaves, an indumentum on the lower leaf face, discoid or radiate capitula with up to 32 yellow disc flowers, an inconspicuously sagittate anther base and a conical, hispid and caudate style-branch apex.According to published regional checklists, Gynoxys is distributed from Bolivia to Venezuela at altitudes between 1600 and 4700 m above sea level and estimated to comprise about 180 species (Brako and Zarucchi 1993;Jorgensen and Leon-Yanez 1999;Beck and Ibañez 2014;Bernal et al. 2019).
Paragynoxys was first described by Cuatrecasas (1951) as Senecio sect.Paragynoxys, but raised to generic rank shortly thereafter (Cuatrecasas 1955).It is characterised by a tree-or shrub-like habit, subcoriaceous petiolate alternate leaves, a corymbose-paniculate terminal synflorescence, few-flowered discoid capitula, white corollae with the limb divided to its base, conical style-branches and a distribution in Colombia and Venezuela.The only taxonomic revision by Correa (2003) recognised 12 species and extended its diagnosis by having radiate capitula with five or more inner phyllaries and up to 12 flowers.
Aequatorium was published by Nordenstam (1978) to accommodate two shrubby species with alternate leaves, a rusty tomentum of stellate hairs, white corollae, sagittate or auriculate anther bases and blunt style-branches apices.Subsequently, several new species were added (i.e.Díaz-Piedrahita and Cuatrecasas 1990;Jeffrey 1992;Díaz-Piedrahita and Cuatrecasas 1994;Nordenstam 1997), resulting in an ongoing discussion on morphological features suitable for circumscribing the genus (see Nordenstam 1997).Based on the presence of stellate hairs and the differently-shaped involucre, Jeffrey (1992) transferred Gynoxys section Praegynoxys to Aequatorium.Nordenstam (1997) concurred with this hypothesis and divided Aequatorium in two subgenera.Aequatorium subg.Aequatorium included species with (generally) alternate leaves, peltate trichomes forming two layers, white flowers, apically obtuse style branches; distributed in Ecuador and Colombia.Aequatorium subg.Praegynoxys included species with opposite or alternate leaves, irregular branching trichomes, absence of the overlying brownish tomentum, yellow flowers and apically pointed style branches and distributed in Argentina, Bolivia, Peru and southern Ecuador.Interestingly, he even suspected that the latter subgenus may be closer to Gynoxys than to Aequatorium.These concerns were taken up by Lundin (2006), who raised Aequatorium subg.Praegynoxys to a genus of its own, Nordenstamia, including 14 species.
Since the establishment of Gynoxys, the first genus in the clade, almost 200 years ago, new species continue to be described in this conspicuous Andean plant group (Cuatrecasas 1950(Cuatrecasas , 1951(Cuatrecasas , 1954(Cuatrecasas , 1955;;Robinson and Cuatrecasas 1992;Beltrán and Baldeón 2009;Beltrán and Calvo 2020).However, monographic work aiming at a synthesis of taxonomic data was largely limited to Gynoxys and Paragynoxys (Herrera 1980;Robinson and Cuatrecasas 1984;Nordenstam 1997;Correa 2003) or to geographically-confined areas (Dillon et al. 1993;Nordenstam and Lundin 1999;Badillo et al. 2008;Beck and Ibáñez 2014;Avila et al. 2016).The considerable species number, the shallow morphological differentiation within the clade and the absence of a robust phylogenetic hypothesis added considerable uncertainty and instability to the circumscription of the genera of the Gynoxyoids, which has found its expression in frequent transfers of species between genera.A consistent taxonomic synthesis is, therefore, needed for the whole Gynoxyoid clade.
The availability of electronic sources for names and protologue citations (IPNI, www.ipni.org;TROPICOS, www.tropicos.org),as well as online access to digitised type specimens (JSTOR Global Plants, https://plants.jstor.org/)and electronic tools to support the taxonomic workflow (EDIT Platform; Berendsohn (2010)) has facilitated the way taxonomic treatments are undertaken.More recently, a comprehensive name source is available through the World Flora Online Plant list which is regularly updated (worldfloraonline.org).Therefore, names can be imported into an electronic taxonomic working tool so that the actual taxonomic research can focus on checking validity of names and testing taxon concepts at species level.At the same time, the taxonomic workflows are revolutionised by structured data (Kilian et al. 2015) and evolutionary approaches to investigate species limits (Stuessy and Lack 2011;Marhold et al. 2013).
For the Gynoxyoid clade, we have taken on the task to check all names and to present a consistent classification at species level as a baseline hypothesis for the whole clade using the available data.While our approach is still largely based on morpho-species, it utilises some phylogenetic data that could be generated for specimens representing part of the species.Our goal was to elaborate an expert-revised taxonomic backbone for a plant group throughout its range of distribution in the sense of the workflow of the World Flora Online (WFO; see Borsch et al. (2020)), ideally including all validly-published names assigned to a status as accepted name or synonym.Such a taxonomic backbone also provides the best possible taxonomic knowledge in time as this is needed for conservation status assessments, biodiversity monitoring etc.
Considering this situation, the aims of this investigation are: [1] to revise the generic classification of the Gynoxyoids making use of molecular (plastome and nrDNA) and morphological data and [2] to provide a revised species inventory of the Gynoxyoids for the entire range of distribution.

Plant material and sources for specimen data
The study was based on plants observed, collected and photo-documented in the field during three collecting trips in Bolivia and Peru, as well as physical specimens loaned to B from AAU, F, G, K, LPB, MA, MO, NY and P (Thiers, continuously updated).Specimens that were physically examined are listed in Suppl.material 1.In addition, high resolution digital images of herbarium specimens, in particular types, were consulted online either accessed through JSTOR Global Plants (https://plants.jstor.org/),GBIF (https://www.gbif.org/)or directly through online databases of the individual herbaria.

Sources of names and compilation into a checklist of the species of the Gynoxys clade
The species inventory of the Gynoxyoids was built in a database using the EDIT Platform for Cybertaxonomy (Berendsohn 2010), based on imports of names and associated data (authors, protologue citations) from the International Plant Names Index (IPNI) (https://www.ipni.org/)supplemented by TROPICOS (https://tropicos.org/home), the Global Asteraceae Database (https://www.Asteraceae.org/aphia.php?p=stats) and the World Flora online (http://www.worldfloraonline.org/)

Definition and assessment of morphological characters and states
The first round of assessing the morphological variation in the Gynoxyoid group included all species of the genera Aequatorium, Nordenstamia, Paracalia and Paragynoxys and a representative selection in terms of morphological diversity of Gynoxys species, altogether 65 species.We examined the diagnostic characters stated in the protologues and in other studies of the five genera, but also compared specimens to detect morphological variance to develop a list of characters and their states.For this investigation, a character state was considered taxonomically relevant and selected for further processing if its expression marked morphological discontinuities at supra-specific level.For each such character, unordered categorical states were defined following the terminology by Roque et al. (2009) and Beentje (2010).In cases where a more detailed homology statement was needed due to conflicting or unclear use of character definitions or terms, a description and illustration were included.For later reconstruction of character evolution, a specimen-based matrix of characters and states suitable for reliable delimitation and characterisation of supra-specific entities was constructed using the specimens included in the plastid phylogenomic analysis of Escobari et al. (2021).For certain characters, for example, the plant habit, the respective states were recorded from literature if not given on the specimen label.

Ancestral character state reconstruction
Only Bayesian trees obtained from complete plastome sequences with indels coded and alignments manually corrected as provided by Escobari et al. (2021) were used as the hypothesis of the phylogenetic relationships in the Gynoxyoids, because lack of resolution rendered the use of nrDNA marker trees impossible.The reconstruction of character states at ancestral nodes was performed with a Bayesian approach using BayesTraits version 2.0 (Pagel and Meade 2006), which uses a selection of post-burn-in trees obtained from the t.files of the Bayesian analysis.This random selection of 800 of the total of 1600 post-burn-in trees taken from Escobari et al. (2021) was obtained through Mesquite version 3.7 (Maddison and Maddison 2021).The file stating the relevant nodes of the tree to be addressed by the analyses of BayesTraits was generated with TreeGraph v.2.14beta (Stöver and Müller 2010).The inference of the ancestral character state reconstruction was performed using the reverse jump MCMC approach with 5,050,000 iterations, with a burn-in of 50000, a sample frequency of 1000 and, following the recommendation by Pagel and Meade (2006), a hyper-prior where the mean of the exponential is drawn from a uniform 0-100 distribution.TreeGraph v.2.14beta (Stöver and Müller 2010) was used to plot the results from the BayesTraits output log file with the function Import BayesTraits data on the Bayesian major consensus tree.We excluded the other genera of the Tussilagineae that were present in the plastid phylogenomic investigation, considering that the outgroup sampling in their dataset is incomplete with respect to the morphological diversity.

Extraction, amplification and phylogenetic tree inference of nuclear ribosomal DNA
To achieve a better overview on species-level phylogenetic relationships within the Gynoxyoid clade and to test if groups of samples identified with the same species name appeared in terminal subclades, 171 samples belonging to 50 species (Suppl.material 1) were included into an extended molecular dataset.These samples were selected to cover morphological and geographical variation as much as possible and also included the samples that were already part of the plastid phylogenomic study.The nrITS and nrETS regions were used as they provided some variable and informative characters in a short marker that was possible to sequence with little effort per sample.Additionally, by representing the nuclear genome, the dataset could be used to test for incongruence between trees inferred from different genomic compartments.Plastid regions often applied to assess the tree space of speciose clades (Mansion et al. 2012) were not suitable in the Gynoxyoid clade due to extremely low genetic distances (Escobari et al. 2020).Genomic DNA was extracted using the CTAB method by Doyle and Doyle (1987), with three fractions for each sample as modified by Borsch et al. (2003).PCR amplification of ITS followed White et al. (1990), ETS was amplified with the primers AST-1 (f) and 18-S-ETS (r) (Markos and Baldwin 2001), following Pelser et al. (2010).PCR was performed in a peqSTAR Thermocycler 1107D (PeqLab, Erlangen, Germany).The PCR products were electrophoresed on 1.5% agarose, the bands were cut out and cleaned with the GenepHlow Gel/PCR kit (Geneaid, New Taipei, Taiwan).Samples were se-quenced by Macrogen Europe (Amsterdam, The Netherlands).Sequence files were aligned using MAFFT v.7.394 (Katoh and Standley 2013) and manually edited using PhyDE version 0.9971 (Müller et al. 2010), following the rules of Löhne and Borsch (2005).Indels were coded as binary characters using the simple-indel-coding method (Simmons and Ochoterena 2000) in SeqState version 1.4.1 (Müller 2005).Altogether, 146 ETS and 166 ITS sequences were newly generated and the sequences were deposited in the European Nucleotide Archive (ENA) using the annonex2embl submission pipeline (Grünstäudl 2020) and can be retrieved from ENA under study number PRJEB53579 (https://www.ebi.ac.uk/ena/submit/webin/study/PRJEB53579).
Phylogenetic trees were inferred from the ITS, ETS and a concatenated matrix of both belonging to the corresponding samples in the plastid tree presented in Escobari et al. (2021).A Bayesian analysis was performed with MrBayes v.3.2.6 (Ronquist and Huelsenbeck 2003), using four parallel Markov Chain Monte Carlo (MCMC) runs for a total of 50 million generations under the GTR+G+I model.The convergence of the Markov chains was checked with Tracer v.1.7 (Rambaut et al. 2018).The initial 25% of all trees were discarded as burn-in and the remaining trees were used to summarise the 50% majority consensus tree.

Assignment of all species to genera and evaluation of taxon concept at species level
Despite the extended nuclear ribosomal sequence dataset generated in this investigation, not all species could be included into phylogenetic analysis.This was largely due to the unavailability of suitable material, for example, in species only known from type or historical specimens.We, therefore, used our set of eleven morphological characters with their states in conjunction with the results from ancestral state reconstruction, to assign all species to a genus and, in the case of Gynoxys species, also to informal infrageneric groups of morphologically similar species that can be used as a hypothesis on close relationships.The genera and informal infrageneric entities were described and a taxonomic key for their determination was created.At species level, all protologues were consulted to check for the correct typification of names.Type specimens of all names, with the exception of only a few unavailable ones (indicated in the taxonomic treatment part, below), were examined from high resolution digital images provided by JSTOR Global Plants, GBIF and the herbarium websites of individual herbaria.The digital images of type specimens were referenced in the checklist to the type citation.Where necessary, new combinations were made and names were lectotypified.As a general principle, a morpho-species concept, delimiting species purely based on morphological discontinuities, was applied.Type specimens and additional specimens (see Suppl.material 1) were examined to assess the qualitative differences and possible infraspecific variation with the aim to hypothesise a name as accepted or as a synonym.The citation of authors follows the international standards by Brummitt and Powell (1992); the citation of publications follows BPH (Bridson et al. 2004) and TL-2 (Stafleu andCowan 1976-1986;Stafleu andMennega 1992-2009); the latter was also consulted for actual publication dates.Accepted names were provided with full synonymies and type citations.Type specimens that were online include only the herbarium acronym.Specimens that were physically examined are marked with (!).

Morphological characters of taxonomic relevance on supra-specific level
The evaluation of morphological characters with respect to discontinuities at supra-specific level resulted in a matrix of eleven characters.These characters and their states are defined in Table 1 and, where appropriate, illustrated in Fig. 1.

Morphological characterisation of the members of the Gynoxys clade
Our evaluation for consistent presence and absence of sets of diagnostic character states in Gynoxyoid species resulted in the recognition of four morphologically and phylogenetically defined genera.The morphological matrix with the diagnostic characters applied to the genera and species of the Gynoxyoids represented in the sampling for the plastome tree is given in Table 2.
Table 1.Morphological characters selected for the ancestral character reconstruction analysis with their respective character abbreviation (Abbr.)and character states with a respective abbreviation and definition when needed.
Multicellular hairs: branched or unbranched hairs.Differences between multicellular hairs were avoided since several types of these can be present in a same specimen (Fig. 1F).

Corolla colour CF white (W), yellow (Y)
This character state describes both ray and disc flowers since it is always shared by both flower types in a capitulum.

Outer phyllaries OP absent (A), present (P)
As outer phyllaries were considered all phyllaries attached at the base of the involucrum and not at the peduncle of the capitulum Number of inner phyllaries The following categories are based on the stability of a defined number of phyllaries for the genera

Radiate flowers RF absent (D), present (R)
The states implicitly define the architecture of the capitulum.The absence of ray flowers (0 = A) represents a discoid capitulum (Fig. 1A).A number > 0 represents a radiate capitulum The following categories are based on the stability of a defined number for the genera Ratio corolla lobe/tube length Rat ≤ 0.6 (S), > 0.6 (D) This character describes the opening depth of the corolla.Length of lobes in relation to the length of the corolla tube (shortly vs. deeply lobed corolla) (Fig. 1C).
Anther-base shape AB sagittate (S), obtuse (O) The base of the anthers is defined as obtuse when no appendage can be distinguished (Fig. 1e).
We ignored the difference between acute (small appendages) vs. sagittate (large appendages) since both can be present in a same specimen and this may be unstable depending on the state of the specimen Style branch apex shape The style branch apex is described as acute when the branches tips have a conspicuously pointed tip (Fig. 1D).We use rounded in a wider sense also including an apex described as truncate, as the presence of papillose hairs makes the distinction unreliable The first of these four genera is Aequatorium with all species sharing the combination of multicellular trichomes, radiate capitula, white flowers, a low number of disc flowers (< 8) and an obtuse shape of the anther base.Diagnostic for this genus is the unique combination of white flowers and radiate capitula.
Further genera are Paracalia and Paragynoxys, the species of which are differentiated from the other Gynoxyoid genera by a deep-lobed corolla, white flowers and discoid capitula.Paracalia can be distinguished from Paragynoxys by a scandent habit, absence of outer phyllaries and a central Andean distribution.In contrast, Paragynoxys has a woody habit, an involucrum with outer phyllaries and a north-Andean distribution.
The genus Nordenstamia cannot be delimited morphologically.The presence of stellate hairs by which this genus was originally distinguished from Gynoxys (Lundin 2006) is not only highly variable amongst the Nordenstamia species, but also shared with many Gynoxys species.If Nordenstamia is included in Gynoxys, this genus can be differentiated from all the others by the combination of yellow flowers and a shallowly divided disc corolla.
Gynoxys is notably the most diverse taxon within the Gynoxyoid clade, displaying a wide range of morphological variation.Within the genus, three informal groups can be discerned, based on distinct characteristics, including phyllotaxis, the number of ray flowers and the type of trichomes.The first group encompasses species with discoid capitula.In contrast, the second group comprises species with multiseriate stellate hairs, primarily featuring alternate leaves.Finally, the third and largest group is characterised by opposite leaves, radiate capitula and simple hairs.

Phylogenetic trees inferred from nuclear ribosomal markers
In addition to the trees of the Gynoxyoids, based on a representative set of complete plastid genome sequences (Escobari et al. 2021), this study attempted to provide further phylogenetic evidence from nrDNA, amongst many others also including the same set of samples present in the plastid tree.Three phylogenetic analyses were performed, based on the ribosomal nuclear markers ETS, ITS and a concatenation of both (Suppl.material 2).In contrast to the tree, based on the plastid genome (Suppl.material 2: appendix 2a), the Bayesian ETS and ITS trees are poorly resolved (Suppl.material 2: appendix 2d).In all trees, the members of the Gynoxys clade form a single polytomy.The sister group relationship between the two species of Paragynoxys is the only clear congruence between the two nuclear ribosomal trees (however, with low support in the ITS inference) and is, moreover, in conformity with the plastid genome tree.Only the ETS tree resolved the two species of Paracalia as a (moderately supported) clade (Suppl.material 2: appendix 2c), whereas they were resolved in separate clades in the ITS (Suppl.material 2: appendix 2b) and in the concatenated ETS+ITS tree (Suppl.material 2: appendix 2d).

Character evolution in the Gynoxyoids based on the phylogenetic hypothesis of the plastome tree
Employing the eleven characters of The Gynoxyoids exhibit various evolutionary changes in their characteristics.The shrubby habit was initially considered primitive, but two independent shifts to a scandent habit occurred in the two Paracalia species, while a shift from shrub to tree habit was observed in Paragynoxys and within the Gynoxys clade.Opposite phyllotaxis was revealed as the ancestral state, but shifts to alternate phyllotaxis occurred in Paracalia pentamera, the stem node of Paragynoxys, two (out of three) species of Nordenstamia and Paracalia jungioides.Unicellular trichomes were revealed as ancestral for all Gynoxyoids, but Paracalia pentamera became glabrous.Multicellular hairs emerged in the most recent common ancestor of Aequatorium and Paragynoxys, as well as in certain species within the Gynoxys clade.White flowers were revealed as the ancestral state, retained by the earliest diverging clades (Aequatorium, Paragynoxys and Paracalia pentamera), while yellow flowers appeared at the stem node of Gynoxys and Nordenstamia.A reversal to white flowers occurred in Paracalia jungioides, nested within the Gynoxys clade.A higher number of inner phyllaries was ancestral, but both Paracalia and Paragynoxys species showed a decrease in this number.Radiate flowers were ancestral, but discoid capitula emerged in all Paracalia and Paragynoxys species, with additional losses of ray flowers in some Gynoxys species.et al. (2021).Each pie chart represents a single character and each colour represents a character state which is described in the legend.The actual state of the characters is represented by boxes next to the species names.The pie charts at the stem of the tree show the character abbreviations as mentioned in Table 1.
A high number of disc flowers was the ancestral state, but reductions occurred at the stem node of Aequatorium and Paragynoxys and in all Paracalia species, partially within the Gynoxys clade.A shallow division of the corolla into lobes was revealed as plesiomorphic and retained in Aequatorium and all Gynoxys, but changed in Paracalia and Paragynoxys to a deep division.The style branch apex was rounded ancestrally, retained in Aequatorium and Paragynoxys, but an acute apex appeared in the earliest diverging species, Paracalia pentamera, with further shifts and reversals in Gynoxys, Nordenstamia and Paracalia jungioides.A summary of the BayesTraits analysis of all state shifts for each character in Figs 2-4 is given in Fig. 5. Characters are represented by numbers and states with the codes given in Table 1.A high number of shifts occur in the two species of Paracalia because the genus is retrieved as non-monophyletic in the plastid topology, although its species share most morphological character states.The clade represented by both Paragynoxys species shared all derived characters with Aequatorium jamesonii in addition to five derived characters that characterise the clade.A single character (corolla colour) was retrieved as synapomorphic for the clade containing Aequatorium, Gynoxys, Paragynoxys, Nordenstamia and Paracalia jungioides and even this character shows several reversals at the MCRA of Aequatorium and Paragynoxys and of both species of Paracalia.The analysis retrieved most of the morphological characters as highly homoplastic with the style branch apices being the most variable character throughout the tree at many nodes.A summary of the BayesTraits analysis with each character at each node is given in Suppl.material 3.  2021).Each pie chart represents a single character and each colour represents a character state which is described in the legend.The actual state of the characters is represented by boxes next to the species names.The pie charts at the stem of the tree show the character abbreviations as mentioned in Table 1.Missing data are represented as (-).

Checklist of the Gynoxys clade
The initial revision of the different databases resulted in a variable number of species.

Trees inferred from plastid genomes and putative cytonuclear discordance
In the present investigation, we attempted to compare, for the members of the Gynoxyoid clade, tree reconstructions, based on the plastid genome and on the ITS and ETS nuclear ribosomal regions.Variation amongst the plastid genomes was extremely low (Escobari et al. 2021) and even more so in the nuclear ribosomal DNA.The lack of resolution in the nuclear ribosomal trees severely limits the comparison of phylogenetic signal from the organellar and nuclear genomic compartments.Nevertheless, there are some noteworthy exceptions.One is the unequivocal support for the sister group relationship of the two Paragynoxys members.The second is the missing support for the monophyly of the three Nordenstamia members in all three trees.The most significant result is, however, the gene tree incongruence concerning the two Paracalia species.The plastid tree placed Paracalia jungioides within Gynoxys and far distant from the second species, P. pentamera (Escobari et al. 2021).In contrast, the two species of Paracalia are supported as monophyletic in the ETS tree in conformity with morphology, although not in the ITS nor in the concatenated ETS-ITS tree (Suppl.material 2: appendix 2d).This finding is surprising because Paracalia jungioides is morphologically very distinct from all members of Gynoxys.It is scandent (instead of a tree or shrub), has white (instead of yellow) flowers and an involucre without outer phyllaries (instead of present).Moreover, P. jungioides and P. pentamera are morphologically very similar and the plastome phylogeny would suggest that these species have accumulated a high number of independent parallel state shifts (i.e.scandent habit, alternate leaves, absent outer phyllaries, few inner phyllaries, discoid capitula, few number of disc flowers, deep-lobed corolla, obtuse anther base) (Figs 3, 5).We assume that the incongruence with respect to the position of the two Paracalia species between the morphological data and the ETS topology on the one hand and the plastome tree topology on the other hand, is best explained by a chloroplast capture event.Chloroplast capture occurs when two species hybridise and go through extensive backcrossing to one of the ancestors (Rieseberg and Soltis 1991).The hybridisation event followed by extensive backcrossing swamp out the nuclear signal, but the captured plastid remains (Kandziora et al. 2022).In our case, we assume that P. jungioides, after introgression with a Gynoxys species, has captured the plastome of a member of the latter genus.Nuclear-cytoplasmic incongruences have been reported in several studies within the Asteraceae family at higher and specific levels (Kilian et al. 2017;Pascual-Díaz et al. 2021;Senderowicz et al. 2021), especially in the Senecioneae (Pelser et al. 2007(Pelser et al. , 2010)).It has also been shown by Stull et al. (2020) for the asterids that conflicts between nuclear and plastome trees are a relevant issue even at higher evolutionary scales.Phylogenetic inferences on nuclear data recovered different placements for several asterid lineages compared to topologies on plastid data (Yin et al. 2021;Kandziora et al. 2022).This is of some significance when we consider that current backbones of angiosperm phylogeny are largely based on plastid phylogenies (APG IV 2016).Amongst the principal reasons for these incongruences, horizontal gene flow amongst lineages, introgression, hybridisation and incomplete lineage sorting were suggested (Rieseberg and Soltis 1991;Maddison 1997;Vargas et al. 2017).The inclusion of different markers of different origins in a phylogenetic analysis has the capacity to elucidate signals of such events.Pelser et al. (2010) analysed potential causes for tree incongruences in the tribe Senecioneae comparing two nuclear (ITS/ETS) and six plastid markers.They concluded that hybridisation is a much more likely explanation than ILS, long-branch attraction or sampling error.Lee-Yaw et al. (2018) focused their study on or-ganelle discordances by sequencing whole plastomes and over 1000 nuclear single-nucleotide polymorphisms in Helianthus L. The authors showed that incongruences in this genus can be expected at species level and amongst individuals of the same species.The Gynoxyoid clade is a further example of short molecular distances on plastid and ribosomal markers amongst species.The lack of molecular variability hampers the reconstruction of well-supported clades on this type of data; nevertheless, the great morphological variation enables the definition of morpho-species in many cases.On the other hand, the phylogenetic reconstruction, although with moderate support, can give evidence to support the assignment of morphologically similar individuals to the same entities (i.e.hypothesised species).
Gene tree discordance is expected to be more likely in rapid radiating lineages that can be found in young biodiversity hotspots, such as the Andean Region (Madriñán et al. 2013;Kandziora et al. 2022).The fast succession and accumulation of descendant species are prone to inter-breeding before reproductive barriers develop, increasing the probability of incomplete lineage sorting (ILS) (Vargas et al. 2017).In addition, young radiating groups have shown whole genome duplication and hybridisation events in the tropical high-altitude areas of South America (Lachemilla: Morales-Briones et al. 2018;Lupinus: Nevado et al. 2018;Diplostephium: Vargas et al. 2017;Espeletiinae: Cortés et al. 2018).Hybridisation may be a result of sexual selection, ecological adaptation, pollinator changes (Moreira-Munoz 2020; Kandziora et al. 2022) or due to the dynamic changes in habitat connectivity in this ecosystem with multiple topography changes during the Pleistocene (Flantua et al. 2019) which facilitated the contact between geographically isolated species before exhibiting strong barriers to gene flow (Vargas et al. 2017;Kandziora et al. 2022).Vargas et al. (2017) revealed complex patterns of reticulate evolution at generic and species level of Diplostephium.

Gynoxys clade
Previous generic classifications of the Gynoxyoid group were based on morphological similarities and discontinuities between species assemblages.In this study, we tested these hypotheses by optimising character states on the full plastome phylogeny (Escobari et al. 2021, see also Suppl.material 2: appendix 2a).Morphological differentiation amongst the Gynoxyoids is shallow and limited to comparatively few and often rather subtle characters.The most recent ancestor of the Gynoxyoid had a shrubby habit, opposite leaves and it was vested by unicellular simple hairs.The capitula was radiate, equipped with outer phyllaries and 6-8 inner phyllaries and had up to eight disc flowers.The corolla was whitish and the corolla lobes were remarkably shorter than the corolla tube.Most of these plesiomorphic states (except the whitish corolla) were retained by most of the species of the genus Gynoxys during its evolution.Shifts in the character states are evident in the rest of the Gynoxyoid members.Specially, the switch from whitish to yellowish corolla (which is apparently the only synapomorphy under the given tree inference) resulted as unresolved due to a small difference of the PP values (59% yellow vs. 41% white).All shifts reconstructed under the ancestral character recon-struction were retrieved as highly homoplasious and are, therefore, unsuitable for genera characterisation under the given plastid inference presented in Escobari et al. (2021).

Species diversity of the Gynoxyoid clade
Our taxonomic backbone provides the best estimate of species diversity in the Gynoxyoid clade.Type information has been synthesised here for the first time in a comprehensive way.Further taxonomic knowledge turnover is expected at species level once species limits are tested in an integrative approach in a separate paper.Specially the examination of a reduced group of Bolivian species depicted shallow morphological differences, making the taxonomy complicated and predicting further nomenclatural changes.Additionally, the low number of collections available hinders a full examination of the species limits.
Notes: We exclude Aequatorium venezuelanum from this genus, based on its yellow flowers and distribution and transfer this species to Gynoxys.Note: Although the phylogenetic inferences suggest this genus to be not monophyletic, we kept the circumscription of Paracalia including two species.We substantiate this decision based on shared morphological characters, such as deeply lobed and white-flowered corolla and the central Andean distribution beginning from lowlands (800 m).Paracalia jungioides which is nested in the Gynoxys clade strikingly differs morphologically from the true Gynoxys species and its inclusion in this genus would break the continuity of the morphological characters and altitudinal distribution in this group.A possible explanation for the contradiction between morphological/ecological and molecular data may be chloroplast capture and this needs to be further studied and better understood before further nomenclatural decisions are made.In this context, we think the best practice is to retain the current circumscription of Paracalia and avoid suggesting further possibly wrong hypotheses of relationships of these species.
Note: Gynoxys alternifolia and G. mandonii were described in literature as scandent.This information is certainly erroneous; in the field, we had a chance to trace several individuals of G. mandonii as large trees with thick branches and the type specimen of G. alternifolia also shows thick and erect branches with no sign of a liana-like growth.

Figure 2 .Figure 3 .
Figure2.Bayesian inference of ancestral character state reconstruction of four morphological characters of the Gynoxyoid clade in the consensus plastome tree byEscobari et al. (2021).Each pie chart represents a single character and each colour represents a character state as described in the legend.The actual state of the characters is represented by boxes next to the species names.The pie charts at the stem of the tree show the character abbreviations as mentioned in Table1.Missing data are represented as (-).

Figure 4 .
Figure 4. Bayesian inference of ancestral character state reconstruction of four morphological characters of the Gynoxyoid clade in consensus plastome tree byEscobari et al. (2021).Each pie chart represents a single character and each colour represents a character state which is described in the legend.The actual state of the characters is represented by boxes next to the species names.The pie charts at the stem of the tree show the character abbreviations as mentioned in Table1.Missing data are represented as (-).

Figure 5 .
Figure 5. Summary tree based on the results of the BayesTraits analysis (Figs 2-4) of state shifts in morphological character.A threshold of 0.75 was used to define the character shifts between states.Characters with multiple state shifts (homoplasies) are shown with white boxes, reversals are indicated by * and unresolved shifts are indicated by an open circle.Numbers at the left of the branches represent the nodes in Suppl.material 3.

Table 2 .
Escobari et al. (2021)rs and their states for the species within the Gynoxyoid clade as included in the phylogeny inferred byEscobari et al. (2021).The codes of characters and states are noted in Table1, (?) indicates missing data.