Character- and tree-based delimitation of species in the 'Cichlasoma' facetum group (Teleostei, Cichlidae) with the description of a new genus
2006; Wiley; Volume: 44; Issue: 2 Linguagem: Inglês
10.1111/j.1439-0469.2005.00347.x
ISSN1439-0469
Autores Tópico(s)Evolution and Paleontology Studies
ResumoThe 'Cichlasoma' facetum group is part of the taxonomically complex group of Neotropical cichlid fishes of the tribe Heroini. Many species groups and unplaced species of heroines are still left without a generic name following the revision of the genus Cichlasoma. We describe here the 'Cichlasoma' facetum group as a new genus, Australoheros, and provide evidence for its monophyly based on phylogenetic analyses of morphological and mtDNA characters. Australoheros is morphologically characterized by the lowest values in meristic characters among heroines and by three apomorphic characters in coloration pattern. In addition to the three described species of Australoheros, our results of species delimitation based on a combination of tree- and character-based approaches identify seven putatively new species of Australoheros. Several coding schemes of morphological characters are used to recover the intrageneric relationships within the genus, resulting in very similar topologies. Discovery of additional species within the genus is expected once material from the whole distribution area is studied. Die 'Cichlasoma' facetum Gruppe ist eine der zahlreichen Artengruppen von Amerikanischen Buntbarschen der taxonomisch komplizierten Gruppe der Heroini. Viele Artengruppen der Heroinen haben keinen stabilen Gattungsnamen nach der Revision der Gattung Cichlasoma. Wir beschreiben diese Artengruppe als neue Gattung, Australoheros, und demonstrieren ihre Monophylie anhand einer phylogenetischen Analyse von morphologischen und mtDNA Merkmalen. Morphologisch ist Australoheros durch die niedrigsten meristischen Werte innerhalb der Heroini sowie durch drei apomorphe Merkmale gekennzeichnet. Die Kombination verschiedener Verfahren der Merkmalsanalyse ermöglicht die Identifikation von sieben weiteren Arten neben den drei bereits beschriebenen dieser Artengruppe. Durch verschiedene Kodierungsschemata der morphologischen Merkmale entwickeln wir eine phylogenetische Hypothese der Verwandtschaftsverhältnisse der Arten innerhalb von Australoheros und zeigen, dass verschiedene Kodierungen zu sehr ähnlichen Topologien führen. Cichlasoma facetum is the oldest Neotropical aquarium fish, first brought alive to Europe in 1889 and bred for the first time in captivity before the end of the 19th century (Bade 1897). The first specimens described as Chromis facetusJenyns, 1842, were collected in coastal Uruguay during the famous voyage of Charles Darwin. Chromys oblongade Castelnau, 1855, Heros autochtonGünther, 1862, H. jenynsiiSteindachner, 1869, and H. acaroidesHensel, 1870 are treated as synonyms to C. facetus, known as C. facetum for most of the 20th century. More than two decades ago, Kullander (1983) restricted Cichlasoma to only 12 South American species, most of them described in that revision, and the species is since then referred to as 'Cichlasoma' facetum (Jenyns, 1842). The genus level taxonomy and also phylogenetic relationships of Mesoamerican heroines are poorly understood, and after the restriction of Cichlasoma, more than eight species groups including >20 species are without any applicable generic name. In view of the difficulty of a prompt comprehensive analysis of Cichlasoma sensu lato, Kullander (1996) suggested the provisional use of available generic names, an advice that other workers had already widely adopted, in spite of the earlier opinion by Kullander himself (1983), Stiassny (1991), and Miller (1993), that 'Cichlasoma'– with the quotation marks indicating its informality – should be used for species for which no generic name was applicable. To date, no study has been undertaken to demonstrate monophyletic status of any of those groups. To make the situation even more complicated, new genera are being described for some of the most complex groups (e.g. Allgayer 2001), ignoring results of studies not in agreement (Martin and Bermingham 1998) and without any phylogenetic analysis demonstrating monophyly. Due to collecting efforts and also recent imports of new cichlids via the pet trade, it has been noted that populations of 'Cichlasoma' facetum coming from different areas are frequently distinct and that 'Cichlasoma' facetum likely represents a species complex (Lucena and Kullander 1992; Staeck 1998a,b, 2003; Körber and Stawikowski 1999; Casciotta et al. 2003). Two additional species to 'Cichlasoma' facetum have been formally described to date, 'Cichlasoma' tembe (Casciotta et al. 1995) and 'Cichlasoma' scitulum (Říčan and Kullander 2003). The 'Cichlasoma' facetum group is distributed in the southern half of South America in the Río Paraná–Río Paraguay system and their tributaries and in the Río Uruguay system. To the west, the group reaches the foothills of the Andes and to the east occurs in the Atlantic coastal drainages of Argentina, Uruguay and Brazil. The southern distributional border is shared with Crenicichla scottii making it one of southernmost distributed cichlids on the South American continent. The question of how to recognize species is a heated topic (see e.g. Wheeler and Meier 2000). There is an immense literature concerning what species are and how they are to be discovered. We share the notion of de Queiroz (1998), who suggested that, despite the long history of dispute over species concepts, most species concepts agree fundamentally that species are lineages (Simpson 1961; Wiley 1978; Cracraft 1983; de Queiroz and Donoghue 1988; Frost and Kluge 1994; Baum and Shaw 1995). What previous authors have generally disagreed about are the best criteria for recognizing these lineages (de Queiroz 1998). This trend is particularly apparent when the meager literature on the methodology of species delimitation is contrasted with the extensive body of work on the theory and methods of phylogenetic analysis. Several species criteria or methods for species delimitation have been proposed (e.g. Avise and Ball 1990; Davis and Nixon 1992; Baum and Donoghue 1995; Mallet 1995; Brower 1999; Wiens 1999; Wiens and Servedio 2000; Puorto et al. 2001; Templeton 2001; Wiens and Penkrot 2002), but empirical taxonomists rarely state their criteria for species delimitation explicitly. In this work, we use character-based and tree-based approaches to analyse morphological characters as two tests of species delimitation. Character-based species delimitation involves finding diagnostic character states that represent seemingly fixed differences between the putative species, or differences that are at least non-overlapping. This approach has been formalized as population aggregation analysis (PAA; Davis and Nixon 1992). However, given finite sample sizes, determining with certainty as to whether traits are truly fixed is virtually impossible (Wiens and Servedio 2000). Most systematic studies using the character-based delimitation are based on the assumption that diagnosability itself is sufficient evidence for species recognition. This notion heavily rests on the assumption that there is a strong differentiation between species and weak differentiation within species. Very few studies have tested this assumption, and there are studies which suggest the 'worst-case scenario' with striking differentiation within species and limited differentiation between species (Wiens and Penkrot 2002). Many systematists utilize statistical analyses of quantitative morphological characters to test species boundaries, often evaluating the extent to which individuals of a putative species cluster together using principal components or canonical variates analysis. This approach is very useful but lacks the clear relationship to estimated patterns of gene flow that the phylogenetic component of the tree-based approach offers. Tree-based delimitation with morphology, although advocated by some authors (e.g. Baum and Donoghue 1995), has rarely been used by empirical systematists (e.g. Hollingsworth 1998; Wiens and Penkrot 2002). A precise methodology for its use has recently been proposed by Wiens and Penkrot (2002), facilitated by methods that allow continuous quantitative characters and polymorphic characters to be included in phylogenetic analyses with little loss of information (e.g. Thiele 1993; Wiens 1999, 2001). Populations rather than individuals are used as terminal units (following Hollingsworth 1998; Wiens and Penkrot 2002) because using individuals will inappropriately treat all polymorphisms shared between populations as homoplasies rather than potential synapomorphies (Wiens 2000). The tree-based approach provides the parsimonious solution of character distribution, a homology hypothesis, and presents monophyletic groups, which are compared with results of the character-based approach. This two step system, combining character- and tree-based approaches, has multiple advantages over a single step system. Empirical studies show that monophyletic groups without diagnostic characters can receive strong support in the tree-based approach, demonstrating that species as lineages do not need to be defined by obvious character states (Wiens and Penkrot 2002). On the contrary, cases in which intraspecific branch lengths are similar to interspecific branch lengths can be reconciled using the character-based approach providing a priory diagnostic characters if these do not conflict with the tree topology. Most characters are based on alcohol preserved museum specimens with notes on live coloration if available. Measurements and counts were taken as described by Kullander (1986). Measurements were taken with digital calipers to 0.1 mm and are made point to point except for head length and snout length, which are projections from the anterior tip of the premaxilla to the orbital margin and the posterior margin of the gill cover, respectively. Scale rows are numbered as described by Kullander (1990), i.e. the horizontal row including the lower lateral line is designated as row E0, and the rows are counted as E1, E2, etc. dorsally, and H1, H2, etc. ventrally. Dorsal and anal fin rays and vertebrae were counted on X-radiographs. Vertebral counts include the last halfcentrum. Colour marking terminology follows Kullander (1983, 1986. Bars are counted and numbered in postero-anterior succession (e.g. Kullander 1983; Kullander and Silfvergrip 1991), in this case because of their more stable number in the posterior part of the body. The number of specimens is indicated in parentheses. Institutional abbreviations are as listed in Leviton et al. (1985) and Leviton and Gibbs (1988). We have used statistical analyses of quantitative morphological characters to test species boundaries, evaluating the extent to which individuals of a putative species cluster together using principal components analysis. The putative species have been formulated based on (i) possession of unique characters/character states; and (ii) based on a unique combination of character states. Statistical analyses were done in the Systat package (SPSS 2000) and the constrained principal components analysis (RDA) has been performed using the Canoco for Windows program supplemented with CanoDraw to visualize results of the analyses (TerBraak and Šmilauer 2002). The RDA analysis has been performed as described in Lepš and Šmilauer (2003; p. 245–252). Data were log transformed prior to analysis and only specimens of a similar size range were included (>50 mm SL). The multivariate analysis has been performed in an iterative approach. Species separated in the first round of the analysis are removed, and the analysis is repeated without these already separated species. The resolution should improve as the most divergent species condensing the axes of the first analysis have been removed. Characters used in the character-based delimitation include all the following (colour pattern characters not included; HL: head length; SL: standard length): HL/SL, snout L/HL, body depth/SL, orbital diameter/HL, head width/HL, interorbital dist./HL, preorbital dist./HL, caudal peduncle L/caudal peduncle depth, pectoral fin L/SL, ventral fin L/SL, last dorsal fin spine L/SL, and the following counts: scale counts (E0, L1, L2, scales between anterior insertion of the dorsal fin and the upper lateral line, scales between the posterior end of the upper lateral line and the dorsal fin, cheek scale rows), lower ceratobranchial rakers, caudal vertebrae, caudal peduncle vertebrae, anal pterygiophores anteriorly from the first haemal spine, anal fin spines, anal fin rays, anal fin total, dorsal fin spines, dorsal fin rays, dorsal fin total, pectoral fin rays. A discriminant analysis has been performed using Statistica (StatSoft, Inc. 2000). Data sets. We have constructed two kinds of data sets for the tree-based approach of species delimitation. In one, we use populations as terminal units (PTU) in tree building to test as to whether the character-based species are monophyletic units on the resulting cladograms. The second data set includes species as terminal units (STU). Cladograms resulting from this matrix are used to evaluate possible differences in topology compared with the PTU analyses and thus to test the robustness of character coding on the phylogenetic hypotheses (see below). All characters that have been found during the study were included into the parsimony analyses. For the STU analysis, 21 characters have been scored and all multistate characters are ordered. See Appendix 1 for details. In the PTU analysis, 35 characters were scored, with most multistate characters ordered. See Appendix 3 for details. The total number of characters is lower in the STU analysis than in the PTU analysis, as some characters are autapomorphies of one species only or because some character complexes could not be split into character states under the coding methods used in the STU analysis. The terminology for the coding methods used follows Campbell and Frost (1993) and Wiens (1995, 1999). Qualitative characters were coded using the majority approach. Two meristic characters in the STU analysis have been coded using the majority coding. Some characters, such as the number of abdominal bars have been coded using the scaled coding (Campbell and Frost 1993). The states are ordered under the assumption that traits pass through a polymorphic stage between absence and fixed presence. The scaled method is advantageous in that it allows polymorphisms to act as synapomorphies. Quantitative characters in the PTU analyses have been coded using two coding methods. The first was a modified gap weighting (GW) method of Thiele (1993). Thiele's implementation of GW involves finding, for a given character, the mean value of the trait in each species in the analysis, the range of mean species values among taxa (i.e. the species with the greatest mean value and the species with the lowest), and then dividing this range into smaller ranges or segments equal to the maximum number of character states allowed by the phylogenetic software program (e.g. 32 for paup*). We have used a less fine grained spacing, thus having in most cases 32 for paup or paup*), which does not allow analyses of large numbers of taxa with unique trait means. If the number of taxa with unique means is too large, the gap-weighting method of Thiele (1993), or it's modification as used here is the best approximation. These methods use less-fine-grained information, but have no limits on the number of taxa that can be coded. We have used the between-state scaling (Wiens 2001) to weight quantitative characters against qualitative characters. This weighting scheme assigns transformations between species with fixed and adjacent values of meristic variables (e.g. 13–14 vertebrae) the same weight as changes in binary variables (0 to 1), and species with intermediate mean values (e.g. 13.5) receive proportionally intermediate weights. Data matrices and descriptions of characters are in Appendices 1–5. Sequences of the mitochondrial cytochrome b (cyt b) gene of 49 species (including outgroup taxa) were obtained from Genbank. Additionally, 29 taxa (representing 24 species, including the first published cyt b sequences of the 'Cichlasoma' facetum group) were sequenced during this study in order to have a complete coverage of all lineages within heroines, and most lineages of cichlasomatines, the sister group of heroines. The sequenced specimens are mainly from aquarium stocks, except some of the Australoheros sequences (Table 1). DNA was extracted from small pieces of muscle or gill (10–25 mg) using the DneasyTM Tissue Kit (Qiagen). The entire cyt b gene (1.3 kb) was PCR amplified with primers GLuDG.L-TGA CTT GAA RAA CCA YCG TTG (Palumbi et al. 1991) and H15915-AAC TGC CAG TCA TCT CCG GGT TAC AAG AC (Irwin et al. 1991). PCR reactions were carried out with initial denaturation at 94°C for 5 min, followed by 30 cycles with denaturation at 94°C for 1 min, primer annealing at 45–50°C for 40 s and primer extension at 72°C for 1 min. PCR was finished by final extension at 72°C for 5 min. PCR products were purified by ethanol precipitation or using Microcon PCR Filter Units (Millipore) and directly sequenced on an automated DNA sequencer using BigDyeTM Terminator Cycle Sequencing Kit v.3.1 (PE Applied Biosystems). Sequencing reaction products were cleaned by ethanol precipitation or with DyeEx 2.0 Spin Kit (Qiagen) were resolved on ABI Prism 310 Genetic Analyser (Perkin Elmer). Except the amplification primers, the following additional primers were used for sequencing: modified L14952 of Lydeard et al. (1995; TCA TCC GTC GCC CAC AT), modified L15162 of Taberlet et al. (1992; CCA TGA GGA CAA ATA TC), and L15299 (Lydeard and Roe 1997). Chromatograms were assembled and checked by eye for potential mistakes using SeqMan II of the DNAStar software package (http://www.dnastar.com). Edited sequences were aligned using the default settings in ClustalX software (Thompson et al. 1997). The alignment was manually revised in BioEdit (Biological sequence alignment editor v5.0.9, http://www.mbio.ncsu.edu/bioedit/bioedit.html). The alignment includes no gaps. The phylogenetic analyses were performed using paup* 4b.10 (Swofford 2001) and the STU analysis also with NoNa (Goloboff 1993) run from the Winclada interface (Nixon 1999), which was used to map character states onto the tree. Maximum parsimony (MP) analyses were performed with same search strategy both with paup* and NoNa/Winclada (100 random sequence additions, 10 trees kept per addition, search on the saved trees to find all the shortest trees). Bootstrap analyses were done using the same approach, with five random sequence additions per one bootstrap. Bootstrap analyses were run with 1000 replications. As the sister group of Australoheros is not known, we have used Heroina isonycterina as the outgroup based on several lines of evidence. First, it is the geographically closest heroine species in a clade of heroines related to Australoheros. Secondly, it is morphologically plesiomorphic in most characters within that clade. Thirdly, when a hypothetical outgroup is reconstructed for the clade containing Australoheros based on the commonality principle, Heroina approaches this reconstructed outgroup most closely. For a few characters, the plesiomorphic condition among the clade of heroines containing Australoheros cannot be decided (characters 1, 9, 12, 15, 17, 19 and 20 in Appendices 1 and 2), and these characters are represented by question marks in the character matrices. To reconstruct the phylogenetic hypothesis testing Australoheros monophyly using cyt b sequences, we have used paup* with the same search strategy and commands as described above. We have weighted all positions and Ti/Tv equally and robustness of the hypothesis has been assessed with bootstrap (1000 pseudorelications). The analysis included all major lineages of heroine cichlids, as well as cichlasomatines and outgroup taxa. The alignment of the 1143 nucleotide positions of the cyt b gene for 78 species contained 492 parsimony informative characters. The recovered phylogenetic hypothesis strongly supports the monophyly of the genus Australoheros (Fig. 1; length, 4708; n, 36; CI, 0.19; RI, 0.49; bootstrap support 100%). The monophyly of Australoheros is supported by 39 unambiguous nucleotide substitutions. Results of the analysis also strongly support Australoheros as a heroine cichlid genus, in agreement with morphological characters (Fig. 1; see below; BS, 79%). Based on the cyt b phylogeny, Australoheros is nested within the predominantly Mesoamerican heroine clade (Fig. 1; BS, 76%), to which it is more closely related than to the South American genera Pterophyllum, Hypselecara, Hoplarchus, Mesonauta, Uaru, Symphysodon and Heros. The cladogram supports the distinctiveness of the five included species, i.e. Australoheros scitulus, Australoheros facetus, Australoheros tembe, A. sp. jacutinga and A. sp. uruguai. Both A. scitulus and A. facetus, for which multiple individuals from different localities were sequenced form strongly supported clades. The average uncorrected pairwise sequence divergence between the five species is 5.5%. The highest pairwise divergence is between A. scitulus and A. sp. jacutinga (7.0%), the smallest between A. sp. jacutinga and A. sp. uruguai (4.2%). The five included species are diagnosed by 18 (A. facetus, A. sp. uruguay) to 33 (A. scitulus) unambiguous nucleotide substitutions. Cladogram depicting relationships among heroine cichlids and showing monophyly of the genus Australoheros. The cladogram shown is the strict consensus of 36 maximum parsimony topologies obtained from a parsimony analysis of the complete cyt b gene. Bootstrap support values shown above nodes (only values >70% are shown). Asterisks below branches show two nodes at which is the cyt b partition strongly in conflict with the morphological partiotion in combined analysis (see Discussion). Combined analysis topology of Australoheros is identical to cyt b topology. A monophyletic group of heroine cichlids having the following synapomorphies: lowest scale counts (modally <25 in E1 scale row); scales on chest of comparable size to flank scales; lowest counts of vertebrae (13 + 13 – 14); unique breeding coloration characteristic in the interruption of the abdominal bars in their middorsal part (Fig. 2); juveniles with distinct xanthophore dots at the base of the caudal fin (Fig. 3); Most species of Australoheros develop four abdominal bars (vs. three), an apomorphic condition among heroines (Říčan et al. in press). Australoheros facetus pair in breeding coloration showing the genus specific apomorphy of interruption of the dorsal portion of the abdominal bars. Australoheros facetus juvenile showing the genus specific xanthophore dots at the base of the caudal fin. The genus is part of the heroine lineage (Fig. 1). Heroines are morphologically diagnosed with the following characters: (i) a single palato-ethmoid articulation (Kullander 1998); (ii) five or more anal fin spines (vs. 4 or less in cichlasomatines and most other Neotropical cichlids; Kullander 1996, 1998); (iii) the palatine bone of heroines is shifted away from the head of the vomer (Kullander 1996, 1998); (iv) midlateral blotch of heroines develops in the fourth ontogenetic bar (vs. in the fifth in all other Neotropical cichlid groups), which is an apomorphic condition (Říčan et al. in press). From the latin word australis, meaning southern, and the name Heros, after the nominotypic genus of the heroini tribe. Australoheros facetus (Jenyns, 1842). Australoheros tembe (Casciotta et al., 1995), Australoheros scitulus (Říčan and Kullander, 2003), and at least seven undescribed species (Australoheros sp. forquilha, Australoheros cf. facetus, Australoheros sp. jacui, Australoheros sp. jacutinga, Australoheros sp. paraguay, Australoheros sp. pirapo, and Australoheros sp. uruguai). The description is based on specimens over 50 mm SL. Body generalized in proportions (see Říčan and Kullander 2003), with variation between the species, variation also in head and mouth shape (short to long snouted, with inferior to superior mouths). American type lips (Kullander 1986: Fig. 12). Scales on head and chest not distinctly smaller than on flanks. Scales in E1 row 23–26 (95% range 24–25). Upper lateral line scales 15–19 (95% range 16–18), lower lateral line scales 7–10 (95% range 8–9). Scales between upper lateral line and dorsal fin 3–5 anteriorly, one large one small to two large one small posteriorly, scales between lateral lines two. Circumpeduncular scales 16. Cheek scale rows 3–4 (5). Lower lateral line continued on caudal fin by 1 or 2 scales. Scale cover of the dorsal and anal fins varies between the species, as does the extent of scalation of the soft parts of the unpaired fins. Pectoral and pelvic fins without scales. Caudal fin densely scaled, scales ctenoid; interradial scales in single rows. Fins: Anal fin meristics typically heroine: V–IX, 9–6. Anal fin pterygiophores 11–14, with one to three pterygiophores anteriorly of the first haemal spine. Dorsal fin: XV–XVII, 11–8. Pelvic fin base below pectoral fin base. Pectoral fin rays 12–14. All teeth caniniform, slightly recurved. Outer row teeth increasing in size symphysiad, upper jaw anterior teeth longest, lower jaw anterior teeth subequal. Jaw teeth with a distinct second cusp in all species examined (contra Říčan and Kullander 2003). Gill rakers externally on first gill arch 2 epibranchial, 1 in angle, 5–9 ceratobranchial. Microbranchiospines on external side on second to fourth ceratobranchials. Lower pharyngeal tooth plate of varying shape, ranging from very robust (see Říčan and Kullander 2003), over intermediates (e.g. A. sp. paraguay, A. sp. jacui) to very shallow (e.g. A. sp. forquilha), including marked differences in dentition. Thirteen abdominal and 13–14 (15) caudal vertebrae, with the combination 13 + 13 typical for most species. Caudal peduncle moderate to short, containing none to three vertebrae. Colour pattern: Coloration includes one of the best distinguishing characters of Australoheros and is also one of the characters showing relationships of the genus. A midlateral blotch of variable size, vertical flank bars, an interrupted midlateral stripe and the caudal fin blotch make up the principal coloration markings. Vertical bars can range from wide to relatively narrow, partly depending on their number in the abdominal area. The majority of Australoheros species have four abdominal bars (i.e. between the bar bearing the midlateral blotch and the opercular cleft), which is an apomorphic condition among heroines. Four species (A. scitulus, A. sp. pirapo and A. sp. jacutinga, A. sp. jacui) do always have only three abdominal bars, while three are the norm in a fifth (A. sp. paraguay). The degree of completeness of development of the four bars varies among the species. Two bars posteriorly from the midlateral blotch above the anal fin, the posterior one at the border of the caudal peduncle. The caudal peduncle bar can be fused with the caudal spot bar in species with very short-caudal peduncles. Horizontal markings include an interrupted midlateral band from the opercular cleft posteriad to the midlateral blotch. In some species, the bar continues in the same scale rows posteriad, while in A. sp. pirapo and to a lesser extent in A. scitulus, A. sp. jacutinga and also A. sp. uruguai, the stripe bends upwards posteriorly from the midlateral blotch. Various species specific spotting patterns are observed inside the genus, including an opalescent stripe below the eye in A. sp. forquilha, spotted opercular bones and anterior part of the body, in A. scitulus, or checkerboard spotted membranes of unpaired fins (A. sp. forquilha). Australoheros juveniles show two very distinct xanthophore dots on the base of the caudal fin, dorsaly and ventraly bordering the caudal fin blotch (Fig. 3). The dots develop at the size of 9 mm TL, are best visible at 11–14 mm TL and become covered by melanophores before 25 mm TL. These spots are known from all three species for which we have complete ontogenetic series (A. facetus, A. scitulus and A. sp. jacutinga). Such spots are not known from any other heroine cichlid (Říčan et al. in press). Examination of the variation encountered while examining the material of Australoheros supported the notion that more species than the three described can be diagnosed. Most of the hypothesized new species can be diagnosed based on unique combinations of characters, but at least one putative species also possesses several unique characters. Two tests of this a priori delimitation have been used, i.e. the character- and the tree-based delimitation. Character-based delimitation: The results of the multivariate RDA analysis support all putative species. The first separated groups are A. sp. forquilha, A. tembe and A. sp. jacui (Fig. 4). The be
Referência(s)