Encyclia

Carnevali, Germán, Tamayo-Cen, Iván, Méndez-Luna, Carlos E., Ramírez-Morillo, Ivón M., Tapia-Muñoz, José L., Cetzal-Ix, William & Romero-González, Gustavo A., 2023, Phylogenetics and historical biogeography of Encyclia (Laeliinae: Orchidaceae) with an emphasis on the E. adenocarpos complex, a new species, and a preliminary species list for the genus, Organisms Diversity & Evolution (New York, N. Y.) 23 (1), pp. 41-75 : 51-65

Phylogeny of Encyclia

We present the most comprehensive Encyclia phylogeny to date, including 106 species accounting for approximately 60% of the taxa and featuring the broadest and more geographically representative sampling of the genus when compared to previous studies (e.g., 70–80 species in Bastos, 2014 with emphasis in Brazil; ca. 60 species in Leopardi-Verde et al. (2016) who focused on the Northern Hemisphere; Tamayo-Cen et al., 2020 with a broad sampling). This study includes all the currently known species of the Encyclia adenocarpos clade upon addition of three recently described species, including the one here proposed as new. In the analyses presented herein, the combination

◂ Fig. 4 Encyclia species. a. E. acapulcensis (GAR) . b. E. acutifolia (MS) . c. E. adenocarpos (GC) . d. E. adenocaula (GC) . e. E. aenicta (GC) . f. E. alata (WCI) . g. E. altissima (ITC) . h. E. ambigua (GC) . i. E. andrichii (GC) . j. E. angustiloba (RL) . k. E. archilae (GC) . l. E. argentinensis (DH) . m. E. aspera (JZ) . n. E. asperula (GC) . o. E. atrorubens (GC) . p. E. betancourtiana (FL) . q. E. bipapularis (GC) . r. E. bocourtii (GC) . s. E. bracteata (AC) . t. E. bractescens (GC)

of all the molecular information allowed us to recover topologies with both morphological and geographical congruence, although not always with high support, a fact that is most likely due to paucity of data and plastid and nuclear topological incongruence. Similar results have been recently obtained in studies of Encyclia (e.g., Higgins et al.,2003; Higgins, 2005; Bastos, 2014; Leopardi-Verde et al. 2016; Tamayo-Cen et al., 2020), other Laeliinae (e.g., van den Berg et al., 2009; Cardoso-Gustavson et al., 2018), orchids in other subtribes (e.g., Batista et al., 2014; Carnevali et al., 2013), and totally unrelated plant groups (e.g., Calonje et al., 2019; Pelser et al., 2019; Ramírez-Morillo et al., 2018). For the purposes of the ensuing discussion, we have adopted the chronology derived from the analysis (with secondary calibrations) of Leopardi-Verde et al. (2016).

Several larger clades, particularly toward the spine of the tree, have poor support or lack it altogether. Thus, we must be cautious at this time regarding their true internal structure and relative position, and we have thus refrained from effecting any nomenclatural proposals. For the purpose of this study, we have selected several clades or species aggregates for discussion, regardless of their support because they show morphological and biogeographical coherence and most likely constitute true linages. Figure 9 features images of species representative of the several species groups or clades discussed below. Below, we will use a few informal names to refer to such aggregations or clades. Most of them have been also identified in previous phylogenies of the genus with approximately similar internal structure and similar or different relative positions (and the same or equivalent names) ( Bastos, 2014; Leopardi-Verde et al., 2016; Tamayo-Cen et al., 2020).

As in previous analyses, there is strong support for the hypothesis that Encyclia bractescens was the first member of the genus to diverge and is thus retrieved as sister to all other Encyclia clades. It is particularly noteworthy that this species is vegetatively like Oestlundia (the former Encyclia subgenus Leptophyllum ) in its small vegetative size, narrow leaves, and perianth segments. Although Oestlundia luteorosea (A. Rich. & Galeotti) W.E. Higgins goes as far south as Peru, the genus is otherwise restricted to Megamexico, like most of the smaller genera of the Encyclia clade other than Encyclia itself and Prosthechea (Leopardi-Verde et al., 2012) . The two largest genera of the clade, Encyclia and Prosthechea , also had major diversifications in Mexico, including the earliest divergent lineages, pointing to a Megamexican origin of the clade and posterior colonizations of other areas, particularly the Antilles and South America. Encyclia bractescens occurs only in Megamexico. The reconstruction of ancestral areas strongly suggests that the stem group and the earliest diverging lineages of Encyclia originated in Megamexico, and these most likely originated in the middle Miocene, circa 12.43 mya ( Leopardi-Verde et al., 2016), at a time when high temperatures and more humid conditions than in the present ( Lunt et al., 2008) must have favored the diversification of mesic clades such as Prosthechea and other members of the Encyclia lineage.

The next lineage to diverge is the Encyclia microbulbon clade, whose stem group must have arisen in Megamexico according to our reconstruction of ancestral areas during the mid Miocene, 9–10 mya ago. Its crown group apparently diverged about 3.9 mya, during the Pliocene when Earth’s temperatures were beginning to cool down, favoring the expansion of dry forests. This clade was described by Leopardi-Verde et al. (2016) and is represented in our study by E. microbulbon (Hook.) Schltr. , E. ovulum (Lindl.) Schltr. , E. adenocaula (Lex.) Schltr. , and E. kennedyi (Fowlie & Withner) Hágsater. We have been unable to obtain sequences of E. guadalupeae R. González & Alvarado. All of these species occur in the mountains of western Mexico, from Sinaloa south into Oaxaca, always in wet, cool montane forests. The extremely rare and poorly known E. lorata Dressler & G.E. Pollard probably belongs in this group judging from overall morphology, geography, and ecological preferences, but no material has been available for DNA extraction and analysis. We had problems in obtaining clean sequences of E. tuerckheimii Schltr. and decided against including it in the analyses. However, as in the case of E. lorata , we hypothesize it belongs in the E. microbulbon clade based upon overall morphology, geography, and ecological preferences. In one of our previous studies ( Leopardi-Verde et al., 2016), this species was retrieved as a member of a South American clade in the ITS analysis reconstruction and as sister to a ( E. adenocarpos “complex”+ E. microbulbon “complex”) clade in the plastid analysis. Clearly, more work is required to understand the relationships of this unusual species. Members of the Encyclia microbulbon clade share with the E. adenocarpos clade the narrow leaves and the verruculose ovary but occur at a lower elevation zone. Based upon ITS sequences, Leopardi-Verde et al. (2016) proposed that the E. adenocarpos clade was sister to the Aptera clade (as the E. meliosma complex), a result that is geographically broadly congruent but lacks statistical, morphological, or ecological support. In the plastid analysis of that study, the Encyclia adenocarpos clade was retrieved as sister to a ( E. microbulbon Clade+ E. tuerckheimii ), a result much more congruent with our analysis.

The Encyclia ceratistes Alliance is composed mostly of species from humid to montane environments, ranging from Chiapas in Mexico through the mountains of the Central American Isthmus into the three cordilleras of the

◂ Fig.5 Encyclia species. a. E. calderoniae (GC) . b. E. candollei (GC) . c. E. ceratistes (JZ) . d. E. chiapasensis (GC) . e. E. chloroleuca (JZ) . f. E. conchaechila (GAR) . g. E. cordigera (Gcas) . h. E. cyperifolia (PH) . i. E. delacruzii (FA) . j. E. dichroma (CvdB). k. E. dickinsoniana (GC) . l. E. diota (GAR) . m. E. diurna (CB) . n. E. dressleri (GC) . o. E. dutrae (CvdB). p. E. elegantula (GCas). q. E. enriquearcilae (GAR) . r. E. fehlingii (MS) . s. E. fowliei (GC) . t. E. fucata (ITC)

Andes in Colombia and Venezuela and the Cordillera de la Costa in northern Venezuela. The stem of this Alliance diverged approximately 5.15 mya ago, during the late Miocene whereas its crown group diverged ca, 3.32 mya ago, during the Pliocene ( Leopardi-Verde et al., 2016). According to our reconstruction of ancestral areas, its MRCA occurred in the composite M + CAI area, where most of the species currently occur, although quite a few others inhabit northern South America. In this species group, the ovaries are verruculose, the columns are provided with short to well-developed stelidia (rarely apparently lacking or small and inconspicuous as in E. stellata ), and in several species the main body of the column is arching upwards, in a 20–40° angle to the ovary ( Fig. 9). This Alliance is particularly well-represented from Panama to Chiapas and Veracruz in Mexico and includes such species as E. ceratistes , E. stellata , and E. dickinsoniana (Withner) Hamer. There are a few, distinctive species complexes included here. One is composed of 2–3 closely related taxa that are fundamentally centered in the Central Valley of Costa Rica, related to E.ncyclia mooreana Pupulin (see Pupulin, 2005, 2006). Another is made up of three species from the Chiapas Highlands and the Sierra de los Chuchumatanes in Guatemala characterized by large flowers with verruculose labella, centered around E. ambigua (Lindl.) Schltr. ( Carnevali et al., 2018b) .

The NHC is distributed in Megamexico, the Antilles (basically Cuba and the Bahamas) and then south into northern South America, particularly along the Caribbean coast and the western slopes of the Andes. It is thus almost restricted to the Antillean Subregion, the Mesoamerican Dominion, and the Pacific Dominion of Morrone (2014). There are a few taxa scattered elsewhere, particularly in the Boreal Brazilian Dominion such as Encyclia granitica . There are no obvious morphological features holding together the three constituent species groups, but they are internally coherent morphologically and geographically.

The Encyclia diurna species group is composed in our analysis of 11 species that are mostly restricted to northern South America. Except for the morphologically aberrant Encyclia cordigera , all have columns with well-developed stelidia ( Fig. 9). Its ancestral area is reconstructed in our analysis as South America with a probability of 74%, whereas in the Leopardi-Verde et al. (2016) chronogram the stem node of this aggregate (not named there) is dated approximately 4.17 mya ago during the early Pliocene whereas its crown node existed also in South America 2.54 mya ago.

A well-supported clade inside it (PP = 0.96; BS = 75%) is composed of the species pair (PP = 1; BS = 90%) Encyclia chloroleuca – E. bohnkiana V.P. Castro & Campacci , the first ranging from Megamexico into the Amazon Basin, the second restricted to NE Brazil, both featuring green flowers and a smooth ovary ( Fig. 9); the remaining species pair is composed of the morphologically dissimilar E. diurna (Jacq.) Schltr. and E. cordigera , the first with verruculose ovary and stelidia-bearing column, the second with a smooth ovary and stelidia-less column and strikingly different flowers. Another highly supported (PP = 0.99; BS = 70%) species pair is that conformed by the Guianese and northern Amazonian species Encyclia ivonae Carnevali & G.A. Romero and E. granitica . This group roughly corresponds with the Northern South America (“Norte da America do Sul”) clade of Bastos (2014) and to the South America 2 clade of Leopardi-Verde et al. (2016).

The Caribbean clade consists of many species and several natural hybrids (57 taxa according to our estimates), mostly restricted to Cuba and the Bahamas. Its ancestral area is reconstructed as the M + CAI combined area with a relatively low probability of 56%. This clade is sister to the western Megamexican Aptera clade. The reconstruction of ancestral areas suggests that the ancestor of both clades occurred in Megamexico from whence it split in two lineages approximately 6.1 mya, one of which dispersed to the West Indies approximately 3.43 mya during the mid Pliocene. It does not feature much internal resolution, but several clades or species groups are identified. One of them, henceforth the Mainland Subclade (PP= 1; BS = 80%, recovered in the parsimony analysis as sister to the rest of the clade) includes three species from the Atlantic watershed of Megamexico, one ranging into Costa Rica and Panama. The distribution of the ancestor of the crown group is reconstructed as the M + WI composite area and most likely occurred approximately 3.43 mya ago where it arrived through a vicariant event at a time when the ancestral composite area was fragmented as sea levels rose during an interglacial warming. The species of this clade are Encyclia alata , E. guatemalensis (Klotzsch) Schltr. , and E. parviflora (Regel) Schltr. The island taxa comprise more than 50 species (at least 23 in Cuba only, Vale et al., 2014) with several natural hybrids recorded. Representative species are E. plicata (Lindl.) Britton & Millsp. , E. phoenicea (Lindl.) Neum. , and Encyclia altissima . The three mainland species and most of the island species feature rugulose ovaries and well-developed column stelidia ( Fig. 9). Furthermore, most of the species are low-elevation taxa, mainly from xeric or coastal habitats. Several are lithophytic or subterrestrial, often on coastal dune associations, and several have

◂ Fig.6 Encyclia species. a. E. gallopavina (AC) . b. E. garcia-esquivelii (GC) c. E. gonzalezii (AC) . d. E. grahamii (ITC) . e. E. guatemalensis (GAR) . f. E. halbingeriana (GC) . g. E. hamiltonii (GC) . h. E. hanburyi (GC) . i. E. howardii (AV) . j. E. huertae (GC) . k. E. inaguensis (GC) . l. E. incumbens (JM) . m. E. inopinata (GC) . n. E. insidiosa (WC) . o. E. ionosma (CvdB). p. E. isochila (DH) . q. E. ivonae (CB) . r. E. jenischiana (AC) . s. E. leucantha (GC) . t. E. linearifolioides (GC)

relatively large, showy flowers. A group of species related to Encyclia altissima and represented in the study only by E. inaguensis are mostly subterrestrials in coastal dunes and feature elongated, narrowly conical or fusiform pseudobulbs, unlike other encyclias, as well as smooth ovaries and are likely to conform a clade when further data are incorporated in the analyzes.

The Aptera clade had also been recovered in previous studies ( Leopardi-Verde et al., 2016; Tamayo-Cen et al., 2020), and it is the only natural aggregation in the genus entirely composed of species characterized by a column lacking stelidia and a smooth ovary ( Fig. 9). Leaves in the Aptera clade are coriaceous and relatively broad and oblong or oblong elliptic. In our analysis, its ancestral area is reconstructed as Megamexico, and, as discussed by Leopardi-Verde et al. (2016), it must have diversified during the cool, dry periods of the late Pliocene-late Pleistocene when the dry forests expanded in Megamexico and associated with the complex orography, opened opportunities for allopatric speciation. As retrieved in our analysis, it comprises about 20–22 species endemic to the Pacific watershed of Megamexico, ranging from Nicaragua northwards into Nayarit in Mexico, with solely Encyclia candollei (Lindl.) Schltr. distributed on the Atlantic watershed in Veracruz, Hidalgo, Puebla, Chiapas, and Oaxaca. Most of the species occur as epiphytes or lithophytes in humid forests at (50–)300–1700(– 2400 m). This species-rich clade includes the E. meliosma (Rchb.f.) Schltr. , E. candollei , E. diota (Lindl.) Schltr. , and E. hanburyi alliances of Leopardi-Verde et al. (2016), although several of these species-complexes are not recovered here as monophyletic. The Aptera clade is sister to the highly supported (PP= 0.99; BS = 76%) Caribbean clade from the Antilles and the Atlantic watershed of Megamexico. Encyclia nematocaulon is recovered in our analyses as sister to the Aptera clade but is extremely different in vegetative and floral characters, as well as in its geography; thus, we have chosen to leave it out of our circumscription of the clade.

In our analysis, the SHA has little internal resolution, but several species groups are consistently recovered in our analyses. Despite the low supports, we have chosen to discuss several of these species aggregations, because they show high degree of morphological and geographical coherence and are likely to be supported in further analyses. These species aggregations are henceforth called Brazil I (which is highly supported and thus constitutes a true clade), Brazil II, Brazil III, and the Andean Clade. The SHA includes species from several biogeographical areas from the western slopes of the Andes (the highly supported Encyclia aspera (Lindl.) Schltr. + E. microtos (Rchb. f.) Hoehne clade), the eastern slopes of the same mountain range (e.g., E. rhizomatosa and E. pflanzii Schltr. ), and many areas of the Brazilian Shield into SE Brazil. The ancestral area of the SHA is not reconstructed in our analysis with any certainty, but the ancestral areas of the internal aggregates, those of the stem and crown groups of the Brazil I clade, the Brazil II, and Brazil III species groups are reconstructed as Eastern South America with high support in all cases, whereas the ancestral area of the Andes clade is reconstructed as the Andes with 100% probability.

The Brazil III species group roughly corresponds to the “ Encyclia osmantha aliança” of Bastos (2014) and is composed of 18 species in our study. These are taxa that occur mostly on the Brazilian Shield, mainly in Cerrado or in the Mata Atlantica as epiphytes and as lithophytes in campos rupestres, more rarely in Amazonian rainforests. Plants are extremely variable (Fig. 10), but many have relatively broad leaves (> 15 mm), column with stelidia, and verruculose ovaries. Two species that occur mainly in the periphery of the Brazilian Shield, E. linearifolioides (Kraenzl.) Hoehne ( Bolivia, Paraguay, and several areas of southern Brazil) and E. conchaechila (Barb. Rodr.) Porto & Brade (northern Amazonian Basin) are anomalous in having columns devoid of stelidia (or these poorly developed) and narrow leaves, as well as a black, deeply bilobed anther. Surprisingly, E. leucantha Schltr. from the Llanos region of Venezuela and Colombia is apparently not related albeit featuring morphologically similar flowers, probably the result of convergence. It is retrieved in our study in the Brazil I clade. The Brazil II species group is also composed mostly of Brazilian taxa, with some species extending their distributions beyond the Brazilian borders into the Andean foothills, mostly in Bolivia and Peru. It includes the “ Encyclia andrichii aliança” of Bastos (2014), characterized by a thick, 3-partite callus. Plants are variable, but the members of the “ Encyclia andrichii aliança” have smaller plants with narrow leaves. Ovaries are warty (excepting E. seidelii Pabst ), and columns have well-developed stelidia (excepting E. cyperifolia ). The Brazil I clade is represented in our study by eight species distributed along the southern section of the Mata Atlantica and the Andean foothills from Peru to Venezuela. Plants are epiphytic, have mostly narrow leaves and verruculose ovaries (excepting E. bracteata ), and the columns display conspicuous stelidia.

The highly supported (PP = 1; BS = 100%) Andean clade is represented in our study by two species ( Encyclia aspera (Lindl.) Schltr. and E. microtos ). We hypothesize that E. angustiloba Schltr. and E. parkeri Reina-Rodr. & Leopardi ,

◂ Fig. 7 Encyclia species. a. E. mapuerae (AC) . b. E. meliosma (GC) . c. E. microbulbon (GC) . d. E. microtos (LOH) . e. E. moebusii (GC) . f. E. naranjapatensis (JZ) . g. E. nematocaulon (WCI) . h. E. oestlundii (GC) . i. E. oncidioides (AC) . j. E. osmantha (AC) k. E. oxypetala (GC) . l. E. papillosa (GC) . m. E. parkeri (JZ) . n. E. parviflora (GC) . o. E. patens (AC) . p. E. peraltensis (GC) . q. E. phoenicea (GC) . r. E. pollardiana (GC) . s. E. powellii (MS) . t. E. profusa (JZ)

all from the Western slopes of the Andes in Colombia, Ecuador, and Peru and whose morphology is concordant, will prove to belong in this clade [Tamayo-Cen and Carnevali, no published] but there was not material available to us for DNA studies. Encyclia betancourtiana Carnevali & Ramírez , retrieved in our analyses (but poorly supported) as sister to the Brazil I + Brazil II species groups, was expected to fall into this clade of Andean taxa.

Phylogeny and biogeography of the Encyclia adenocarpos complex

Recently, we published a comprehensive revision of the Encyclia adenocarpos complex ( Carnevali et al., 2018a), accounting for six species, two of which ( E. enriquearcilae and E. acapulcensis ) were new at the time. This new phylogenetic study includes all currently known species of the Encyclia adenocarpos clade, among them the new species proposed herein. In this multilocus phylogenetic study, the Encyclia adenocarpos clade is recovered as monophyletic and strongly supported (PP = 1; BS = 100%) as sister to a large clade containing ca. 90% of the species of the genus. The biogeographic analysis herein performed reconstructs the ancestral area of this clade as Megamexico with a 0.99 probability, and its historical biogeography includes an event of dispersion of E. papillosa from Megaméxico to the Antilles, an event that is most likely very recent. The divergence of this group is very recent, having occurred during the early Pleistocene, 1.5–2 mya ago, during the climatic changes associated with the glaciations, when the planet was much drier and the dry forests expanded, opening new habitats and opportunities for speciation for the members of the clade ( Leopardi-Verde et al. 2016; Ramírez-Morillo et al., 2018).

The new entity herein proposed, Encyclia mariaeugeniae , fills a distributional gap in the E. adenocarpos clade. The evidence at hand suggests again that the clade is made up of a series of closely related, fundamentally allopatric species aligned along the Pacific coast of Megamexico. The clade reaches its highest diversity in areas of coastal Oaxaca and Guerrero, where five of the seven species occur ( E. rodolfoi , E. enriquearcilae , E. mariaeugeniae , E. acapulcensis , and E. adenocarpos ).

In our analysis, the Encyclia adenocarpos clade breaks up into several species clusters that mostly lack support. Two apparent relationships are noteworthy. The three northernmost species are retrieved as a basal grade, suggesting a northern origin. These are E. trachycarpa , E. adenocarpos , and E. acapulcensis ( Carnevali et al., 2018a) . The southern species of the complex fall into two highly to moderately supported species pairs. Two species that feature a bright yellow column and a white labellum, E. papillosa and E. enriquearcilae are highly supported (PP = 99; BS = 80%) as sister taxa. Then, E. rodolfoi and E. mariaeugeniae , both with yellow, purple streaked labella, are in a poorly supported relationship (PP= 0.82). Both occur in the general area of the Tehuantepec Isthmus in dry forests.

Because Encyclia is a genus of relatively recent divergence ( Bastos, 2014; Leopardi-Verde et al. 2016), there is not much sequence divergence in the loci conventionally used for phylogenetic inferences ( Naciri & Linder, 2015). More powerful methods of massive sequencing, such as genotyping-by-sequencing data (GBS), have recently been used to resolve phylogenetic relationships in species complexes of recent cladogenesis, which typically lack molecular and/or morphological characters to reconstruct their history and are thus difficult to resolve phylogenetically ( Pérez-Escobar et al., 2020). These methods should help resolve the relationships within the E. adenocarpos complex and other species complexes (e.g., the Aptera clade). Also, structural characters, which remain to be thoroughly explored such as leaf, stem, and flower anatomy, as well as pollinia structure, could help clarify the internal relationships of the genus.

Encyclia diversity, biogeography, and broad evolutionary patterns

Our analyses are largely congruent with those of Bastos (2014), Leopardi-Verde et al. (2016), and Tamayo-Cen et al. (2020) in the number and composition of the clades and species groups identified, reflecting the fact that most of the sequences employed in our analyses were drawn from those earlier studies, although the relative positions of some of the clades and species groups may vary somewhat. However, it also indicates that lineages of the genus are mostly geographically restricted, a pattern that is starting to emerge repeatedly in several plant clades and that is evident in several recent phylogenetic studies in Encyclia and other plants ( Leopardi-Verde et al., 2016; Pérez-Escobar et al., 2017b, Pessoa et al., 2018, in orchids; Ramírez-Morillo et al., 2018 in bromeliads; Calonje et al., 2019 in Zamia L.; Pelser et al., 2019 in Rafflesia R.Br. ). The high variability in habits, floral, and vegetative structures found within these geographically restricted clades of Encyclia strongly supports the notion that the genus colonized novel areas, and upon establishment, underwent adaptive radiations that allowed them to occupy a variety of altitudinal, substrate, and, putatively, pollination niches. Most of the larger clades

◂ Fig. 8 a. E. pyriformis (GC) . b. E. randii (GC) . c. E. remotiflora (GC) . d. E. replicata (JZ) . e. E. rhizomatosa (LOH) . f. E. rodolfoi (GC) . g. E. rzedowskiana (GC) . h. E. sabanensis (GC) . i. E. seidelii (GC) . j. E. selligera ((JM). k. E. silverarum (GC) l. E. stellata (GC) . m. E. suaveolens (GC) . n. E. tampensis , albino form (JZ). o. E. trachycarpa (GC) . p. E. trachychila (GE) . q. E. tuerckheimii (GC) . r. E. withnerii (DH) . s. E. × padre ortizii (GC) . t. E. yauaperyensis (JH)

identified in our analyses are geographically restricted. For example, the Caribbean clade is restricted to our Antilles region and to the Antillean Subregion of Morrone (2014).

Although very little is known about the pollination biology of Encyclia , there are a few studies that strongly suggest that differences in color, shape, and size may reflect the use of different pollen vectors ( Braga, 1977; Janzen et al., 1980; Diaz, 2001; Sauleda, 2016; Krahl et al., 2017; Tamayo-Cen, 2020; see also https://www.facebook.com/ g roups/293608830793530/permalink/1853524064801991/). Differences in fragrance composition may also play a role in species isolation ( Del Mazo & Damon, 2006 versus Kaiser, 1993; see also Soto Calvo et al., 2017). In all documented cases, the pollinarium was placed on the frons or top of the head of the insect vector.

Krahl et al. (2017), working at the Reserva Biológica de Campina, north of Manaus, studied the pollination of Encyclia mapuerae (Huber) Brade & Pabst. The species was assumed to deceive its pollinators but was found to produce as reward small but detectable amounts of nectar in the cuniculus. It attracted several species of hymenopterans, but pollination was only successfully performed by a species of Centridini, Centris varia Erichson , which appeared to follow the nectar guides provided by the longitudinal, purple-colored nerves of the central lobe of the labellum. The same species had been studied by Braga (1977), who identified the vespid wasp Stelopolybia cf. pallipes as a pollinator, suggesting the pollination of E. mapuerae is not species-specific. Sauleda (2016) observed the megachilid bee Megachile mendica Cresson successfully pollinating Encyclia tampensis in Florida, a species of medium-sized flowers, with a variable but always conspicuous purple blotch in the midlobe of the labellum. More recently one of us, Tamayo-Cen (2020), documented the pollination of E. nematocaulon near Mérida, Yucatán, Mexico. Encyclia nematocaulon has one of the smallest flowers in the genus, and it is visited by several small bee species but successfully pollinated by the halictid bee Augochlora sp. and the apid bee Exomalopsis sp. both of which are of reduced dimensions. We have observed an unidentified species of Centris successfully working and effecting pollination on flowers of E. guatemalensis under cultivation near its natural habitats in Yucatán, Mexico. The flowers of this species closely resemble in overall size, coloration, and shape those of the unrelated E. mapuerae , which is also pollinated by a species of Centris . In all documented cases, the pollinarium was placed in the frons or top of the head of the insect vector.

Differences in fragrance composition may also play a role in species isolation. The human olfact can recognize several distinctive fragrances of species in the genus, but we know little of their composition and role, if any, in pollination. Del Mazo and Damon (2006) reported the fragrance of what probably is E ncyclia . papillosa (as E. adenocarpos ) as hexadecene, E-nerilidol, and phenylformide, whereas Kaiser (1993) reported that of an unidentifiable species of Encyclia as composed of butyl and benzyl caproate, hydroquinone dimethyl ether and beta -ionone, strongly suggesting that different species produce different fragrances. Soto Calvo et al. (2017) documented the pollination of both E. hircina and E. fucata (Lindl.) Britton & Millsp. , which are sympatric in areas of Cuba. The two species are similar and differ little in the morphology of the small flowers, but feature different fragrances, the first emitting an odor of mammals, probably of deer, and attracting the tabanid Chrysops variegatus (De Geer) . The second species has a sweet fragrance and attracted a bee that could not be captured for determination. The locally sympatric and closely related E. guatemalensis and E. alata have flowers of approximately the same size ( E. alata usually somewhat larger), and whereas E. alata has a deliciously and powerfully sweet fragrance, E. guatemalensis emits no odor detectable by the human nose. Although a few cases of plants intermediate between the two species are known, both largely retain their identities across their extensive area of sympatry.

It has long been thought that Encyclia species deceit their pollinators (e.g., van den Berg & Carnevali, 2005). However, there is some recent evidence that some species do offer rewards: E. mapuerae contains small amounts of nectar in its cuniculus ( Krahl et al., 2017), and most Encyclia species we have examined possess a variously developed cuniculum, so it would be reasonable to assume that at least some of them may offer nectar to reward their pollinators.

Most of the clades and species groups identified in these studies include species that are florally distinctive in shapes and colors, suggestive of the use of different pollinators. Furthermore, they are equally variable in their use of substrates (epiphytes, lithophytes, or subterrestrials) and altitudinal ranges (Fig. 10). As an example, the Aptera clade includes species such as Encyclia rzedowskiana Soto Arenas that grows almost at sea level in tropical rainforests in southern Oaxaca ( Mexico), whereas its close relative, E. huertae Soto Arenas grows as high as 2400 m as a lithophyte in drier associations in Michoacán ( Mexico). These two species feature a labellum where the lateral lobes are partially fused with the central lobe, web-like; other species of the clade, instead, have lateral lobes that are totally free from the central lobe. Within the same clade, E. atrorubens (Rolfe) Schltr. has extremely dark purple, almost black flowers, whereas its relative E. chiapasensis Withner & D.G. Hunt has pale green flowers. In the Brazil III clade, the lithophytic or subterrestrial E. dichroma (Lindl.) Schltr. and E. jenischiana (Rchb.f.) Porto & Brade display entirely purple flowers, whereas most of its relatives, mostly epiphytic taxa, have green or yellow perianth segments.

Furthermore, some of these clades and species complexes are composed of vegetatively and florally similar species, differing in minor details of perianth (mostly labellum) structure or coloration, occurring in close parapatry or allopatry, and using different elevational belts; often they are frustratingly difficult to unambiguously refer to a particular species. This pattern strongly suggests of a group of species of recent evolution, still in a state of fluid evolutionary diversification, with some species apparently grading into each other and possibly still exchanging genes occasionally. In Megamexico II, this is evident in the Aptera clade where the limits of members of the Encyclia meliosma species complex ( E. meliosma , E. spatella (Rchb.f.) Schltr. , E. huertae , E. halbingeriana Hágsater & Soto Arenas , E. rzedowskiana , etc.) are tenuous at best along their distribution for Oaxaca to Nayarit along the Pacific slope of Mexico (e.g., Leopardi-Verde et al., 2016). The same phenomenon is found in the E. adenocarpos and Caribbean clades (e.g., Vale et al., 2014).

According to recent estimates ( Leopardi-Verde et al., 2016; Tamayo-Cen and Carnevali in process), Encyclia diverged from its most recent common ancestor approximately 12 to 7 million years, during the late Tertiary climatic fluctuations (Miocene). During this time, the planet suffered a gradual cooling and consequent drying that enabled the expansion of dry and seasonally dry ecosystems such as those preferred by Encyclia and other plant groups, such as Hoffmannseggella Schltr. ( Gustafsson et al., 2010) , Hechtia Klotzsch ( Ramírez-Morillo et al., 2018) , and others. Judging from the distribution of the closest relatives of the genus (the Prosthechea clade, formed by Alamania Lex. , Amoana Leopardi & Carnevali , Euchile (Dressler & G.E. Pollard) Withner , Oestlundia W.E. Higgins , and Prosthechea Knowles & Westc. ) ( Higgins et al., 2003; Tamayo-Cen and Carnevali in process) and of the earliest diverging clades of the genus ( E. bractescens , the E. microbulbon and E. adenocarpos clades), we can suggest a Megamexican origin for the genus. This hypothesis is supported by the results of our reconstruction of ancestral areas, which clearly places the distributions of these early nodes of the Encyclia + Pr osthechea clade as all having resided in Megamexico.

Furthermore, a preliminary analysis of molecular clock with the inclusion of more species, as well as additional molecular information (e.g., External Transcribed Spacer, ETS) (Tamayo-Cen and Carnevali in process), suggests that the invasion of South America by the SAC (or the South American species group) must have occurred about 3.3 mya when the Panama isthmus was almost completely closed ( Bacon et al., 2015), thus allowing for the expansion of Encyclia southward. Later, in Pliocene times, at least a second wave of Encyclia taxa of the Northern Hemisphere Clade invaded northern South America, where some members of the E. ceratistes Alliance such as E. ceratistes and E. stellata are found. This is in accordance with our findings where the ancestral area of the E. ceratistes Alliance is reconstructed as a combined Megamexico-CAI area.

From the composition and distribution of the clades and species groups identified in this analysis and from the results

◂ Fig. 10 Vegetative variation in Encyclia . a. Encyclia garcia-esquivelii showing shortly creeping rhizomes making subcaespitose plants (GC). b. E. inaguensis showing the elongated, narrowly pyriform pseudobulbs on abbreviated rhizomes found in several coastal dune specialists in the Cuba and the Bahamas (GC). c. E. oncidioides showing elongated, creeping rhizomes found in several, mostly lithophytic species in several clades (AP). d. E. nematocaulon , one of the vegetatively smaller species of the genus, showing crowded, suborbicular pseudobulbs making densely caespitose plants. e. Lithophytic plant of E. adenocarpos growing at the northern limits of the genus near Alamos, Sonora, at approximately 27°N; note roots growing directly over the rock (photograph from http://swbiodiversity.org/seinet/index.php, probably by Mark Dimmit)

of our reconstruction of ancestral areas, it becomes likely that most biogeographical areas where encyclias grow were colonized by the genus just once or only a few times. For example, the Antilles (our West Indies Biogeographical area) has 57 Encyclia taxa of which apparently all but five belong to the Caribbean clade. Of these, E. papillosa is a member of the western Megamexican E. adenocarpos clade. Another, E. nematocaulon , is in its own Megamexican clade. Others, such as Encyclia monticola (Fawc. & Rendle) Acuña appears to be a member of the E. gravida (Lindl.) Schltr. group of species in the E. ceratistes Alliance , whereas E. isochila (Rchb. f.) Dod is the oldest name in a group of taxa of uncertain affinities and taxonomic composition (none has been available for molecular analyses) but that could belong to the Caribbean clade (or even have South American relationships). Encyclia monticola and E. isochila need to be included in our phylogenetic analyses to ascertain whether the Antilles were colonized more than twice. It is noteworthy that the Lesser Antilles are extremely species-poor and east and south of Puerto Rico, the genus appears to be completely absent, only to reappear in Trinidad, where only Encyclia of South American affinities occur (e.g., E. cordigera , E. chloroleuca , E. guianensis Carnevali & G.A. Romero ).

Encyclia nematocaulon is consistently retrieved as sister to the Aptera clade and appears to be yet another event of long-range dispersal, most likely from the Yucatan peninsula to Cuba, from where it was first described and where it has recently been rediscovered ( Soto Calvo et al., 2020). In fact, Cuban plants of E. nematocaulon are morphologically similar to populations of the species from the dry northern area of the Yucatan peninsula, with larger flowers, and large, pure white labellum with conspicuously raised nerves. The Bahamian E. withneri has never been available for DNA sequencing and, judging from floral and vegetative morphology, appears to belong to the E. adenocarpos clade or to the Mainland Subclade of the Caribbean clade, next to E. guatemalensis , with whom it may even be conspecific. Considering how close Cuba is from mainland Mexico (approximately 200 km from extreme NE Yucatan peninsula to the tip of western Cuba), it is surprising how few species in common there are between these areas, with only two extant Encyclia species shared ( E. papillosa and E. nematocaulon ) and evidence for probably 1–2 additional events of long range distance, which are rare but are documented to occur with frequency enough to act as a driving force in the diversification of the Orchidaceae ( Givnish et al., 2015) . Our biogeographical analysis identifies a dispersal event for E. papillosa as the explanation for its arrival to both the Yucatán Peninsula and western Cuba. Along with the inclusion of some of these missing taxa in the phylogenetic and biogeographic analyses, a more rigorous reconstruction of ancestral areas with much more finely split areas, at least in the northern hemisphere, is required to discern how many vicariant/dispersal events are necessary to explain the current distribution of the West Indian Encyclia taxa. However, available chronograms indicate that most of these events and the explosive diversification of the genus in the northern West Indies must have happened starting in the late Pliocene through the late Pleistocene. The large number of nothotaxa within the Caribbean clade also suggest recent divergence.

Encyclia bractescens , sister to all remaining encyclias, is restricted to the Atlantic drainage of Megamexico. The same holds true of western Megamexico (provinces 1, 3, 4, 13 of Morrone) where all the members of the E. microbulbon complex, and all species of the E. adenocarpos clade (excepting the single Yucatecan and Cuban populations of E. papillosa ), are restricted.

Biogeographically, the most complex areas for Encyclia are the Central American Isthmus, the Northern South American region, and the northern section of the Andes Region, where several Encyclia biotas and clades meet. It is noteworthy that the few encyclias of the Pacific coast of Colombia and Ecuador (e.g., E. aspera and E. angustiloba ), belong to South American and not Megamexican or Central America clades. The complex paleotectonic history of northern South America, the Central American Isthmus, and the Chocoan area are the most likely explanation for these mixed Encyclia biotas (e.g., Pérez-Escobar et al., 2019). This has been equally documented for plant groups such as the aroid Philodendron Schott ( Canal et al., 2019) , the orchid genus Cycnoches Lindl. ( Pérez-Escobar et al., 2017b) , and others.