Psittacanthus Mart. (Loranthaceae), with ca. 120 species, is the most species-rich genus of mistletoes in the family, ranging from Baja California in Mexico to northern Argentina (Kuijt 2009, 2014). Psittacanthus species occur in almost all vegetation types and parasitize a wide variety of host tree species (Kuijt 2009). Phylogenetic analyses based on molecular characters and a limited sampling (mostly Mexican species) indicate that Psittacanthus is monophyletic (Ornelas et al. 2016), and sister to the Andean genus Aetanthus (Vidal-Russell & Nickrent 2008a, Su et al. 2015). The absence of endosperm in the fruit is considered a synapomorphy of Psittacanthus and Aethanthus that together with its large flowers, anthers thicker than the filaments, and bulky haustorial connections to the host trees, clearly distinguish the clade from the rest of the Neotropical genera of Loranthaceae (Kuijt 2009).
The present study focuses on three species of Psittacanthus, Psittacanthus palmeri (Watson) Barlow & Wiens (Figure 1A), P. nudus (Molina) Kuijt & Feuer (Figure 1B) and P. sonorae (S.Watson) Kuijt, which are the smallest plants in the genus, and only parasitize Bursera species (Kuijt 2009). These three mistletoe species were originally placed in the polyphyletic genus Phrygilanthus Eichler (Standley 1919), which at the time also included many Old and New World mistletoe taxa. Given the heterogeneous assemblage of species with both Old and New World mistletoe species, Phrygilanthus was thus split into ten genera, including Psittacanthus (Barlow & Wiens 1973). Based on the lack of endosperm, inflorescence organization, pollen characteristics and other similarities of flower features, Phrygilanthus nudus, P. palmeri and P. sonorae were then transferred to Psittacanthus (Kuijt 1971, 1973, 1981, Kuijt & Feuer 1982).
Psittacanthus sonorae is easily recognized by its terete and fleshy leaves, and by the dry viscin area forming a sealed, resinous capsule around the developing seedling; these characteristics are unique among Neotropical Loranthaceae and are considered adaptations to the desert environments in northwestern Mexico (Sonora and Sinaloa) and the southern Baja California Peninsula (Kuijt 2009). Plants of P. palmeri and P. nudus are small and share the lateral inflorescences with sessile dyads (Figure 1A-B), characters not observed in other species of Psittacanthus (Kuijt 2009). Further differences between these two species include the inflorescences, in which P. palmeri has two dyads and P. nudus three or more dyads, and the presumed leafless habit of P. nudus. The Honduran P. nudus, only known from the type specimen and possibly a synonym of P. palmeri, was described as a leafless species (Molina 1952), whereas the Mexican P. palmeri is more widely distributed and considered a seasonally deciduous species based on herbarium collections. The leaflessness or deciduousness of both species needs to be confirmed in the field (Kuijt & Feuer 1982, Kuijt 2009); this is relevant, since no other deciduous mistletoes have been reported in the tropics and subtropics, even on deciduous hosts (Glatzel et al. 2017). Interestingly, two records of leafless mistletoes exist between the type locality of P. nudus in Honduras (Kuijt & Feuer 1982) and the range of P. palmeri from Sonora south to Oaxaca (Kuijt 2009). One is a specimen from Chiapas, in the municipality of La Trinitaria near the border with Guatemala, cited as P. palmeri (Davidse & al. 29967; MEXU, MO). The second record is a photograph taken by D. Lorence of a leafless flowering individual growing on Bursera in the area of Juchitán, Oaxaca (Kuijt 2009), a few kilometers away from where specimens of P. palmeri had also been collected (Cedillo T. & Lorence 482; MEXU). With this evidence, Kuijt (2009) discussed that P. palmeri and P. nudus might be conspecific. It is possible that P. nudus does not represent a valid species, but rather is a form of P. palmeri where samples were collected during the leafless period. Alternatively, it is possible that P. nudus is not restricted geographically to Honduras, and that populations of this species exist in southern Mexico and Guatemala.
A previous molecular phylogenetic analysis of Psittacanthus sequences of nuclear (ITS) and chloroplast (trnLF) DNA markers (Ornelas et al. 2016) included 13 species, with three samples of P. sonorae from the Sonoran Desert and three samples of P. palmeri from xeric scrubland in Oaxaca. That study showed that these two species were part of a well-supported clade sister to P. ramiflorus G. Don. However, P. nudus samples have not been included in previous phylogenetic studies. The type locality of P. nudus is Honduras, Departamento de Francisco Morazán, SW of the Yeguare Valley, banks of Río La Orilla at 900 m (Molina 1952). Since its publication, no further material has been collected and the trees from which the original collections were made have long since been felled (Kuijt & Feuer 1982). Plant collections we made (E. Y. Guerrero 1632, CURLA) are 300 m lower than that reported for the P. nudus holotype, near the type location in the Yeguare Valley (also known as Valle del Zamorano): Aldea de Salalica, road from Zamorano to Sanbuena Ventura, Municipio de Nueva Armenia, Departamento de Francisco Morazán at 600 m elevation and about 50 km SE of Tegucigalpa (Table 1). An additional specimen deposited at the herbarium Paul C. Standley de la Escuela Agrícola Panamericana Zamorano (EAP) was collected and photographed in the same area (J. L. Linares & al. 13530, EAP): San Antonio de Oriente, 18 km S of Zamorano, road to Salalica, Departamento de Francisco Morazán (13º 53´ 49˝ N; 89º 04´ 50˝ W at 750 m elevation).
Location | n | Altitude (m) |
Latitude (N) |
Longitude (W) |
---|---|---|---|---|
Psittacanthus nudus | ||||
Honduras, Francisco Morazán, road El Zamorano-Aldea de Salalica | 1 | 600 | 13° 50´ 38´´ | 87° 07´ 02´´ |
Honduras, Francisco Morazán, Nueva Armenia, Aldea de Salalica | 2 | 500 | 13° 52´ 13´´ | 87° 06´ 36´´ |
Psittacanthus palmeri | ||||
Mexico, Jalisco, Barranca de Huentitán, Ixtlahuacan del Río | 3 | 1,770 | 21° 00´ 42´´ | 103° 09´ 17´´ |
Mexico, Oaxaca, Sta. Ma. Tecomavaca, Cañón del Sabino | 8 | 725 | 17º 51´ 53´´ | 97º 02´ 10´´ |
Mexico, Oaxaca, Santiago Matatlán | 8 | 1,830 | 16° 50´ 28´´ | 96° 22´ 26´´ |
Mexico, Chiapas, La Trinitaria | 4 | 1,475 | 16° 04´ 55´´ | 92° 02´ 06´´ |
Psittacanthus sonorae | ||||
Mexico, Baja California Sur, San Bartolo, Bellavista | 2 | 37 | 23° 43´ 55´´ | 109° 47´ 46´´ |
Mexico, Baja California Sur, La Paz, Cerro La Calavera | 2 | 36 | 24° 04´ 15´´ | 110° 16´ 19´´ |
Mexico, Baja California Sur, 2 km NW Pichilingue | 1 | 23 | 24° 11´ 10´´ | 110° 17´ 58´´ |
Mexico, Baja California Sur, road 286, km 9 Los Planes-La Paz | 1 | 262 | 24° 05´ 03´´ | 110° 13´ 41´´ |
Mexico, Baja California Sur, road La Paz-Los Cabos, Rancho San Ignacio | 2 | 57 | 23° 59´ 22´´ | 110° 00´ 44´´ |
Mexico, Sonora, Cañón de Nacapule | 1 | 83 | 27° 59´ 04´´ | 111° 02´ 40´´ |
Mexico, Sonora, Ejido Cruz de Piedra | 1 | 25 | 27° 57´ 25´´ | 110° 40´ 51´´ |
Mexico, Sonora, Paraiso La Manga | 1 | 7 | 27° 53´ 43´´ | 111° 06´ 55´´ |
Mexico, Sonora, Las Cadenas | 1 | 12 | 28° 17´ 48´´ | 111° 25´ 37´´ |
We obtained sequence data from ITS and trnF DNA markers to address the taxonomic status and phylogenetic position of recently collected samples near the type locality of P. nudus in Honduras. Phylogenetic and molecular dating methods were used to (1) infer the phylogenetic position of these samples within Psittacanthus, and (2) to estimate the time of divergence between these and other samples of putative relatives. Our aim is to provide a phylogenetic framework to link morphological traits (e.g., small size, leaf morphology and deciduousness; traits that in theory could be related to drought tolerance) to the host-mistletoe context and the associated ecological conditions at the time of diversification. Specifically, we asked: Are some morphological traits of Psittacanthus associated with the diversification of Bursera? How was the Bursera’s mistletoes speciation? Does time of diversification in both lineages coincide?
Materials and methods
Taxon sampling. A total of 37 samples were collected, three putatively of P. nudus, 23 of P. palmeri and 12 of P. sonorae (Figure 1A-C, geographic information for samples studied in Table 1, voucher information and GenBank accession numbers in Appendix 1). Twelve other widespread Psittacanthus species (52 samples) and other representatives of Loranthaceae (19 samples) downloaded from GenBank (accession nos. in Appendix 1) were used as outgroups according to Wilson & Calvin (2006), Amico et al. (2007), Vidal-Russell & Nickrent (2007, 2008a,b), Díaz-Infante et al. (2016), Ornelas et al. (2016), and Pérez-Crespo et al. (2017).
DNA extraction and gene amplification. For the newly obtained samples in this study, genomic DNA was extracted from silica dried plant tissues using a standard CTAB method (Doyle & Doyle 1987) or the DNeasy Plant Mini kit (Qiagen, Valencia, CA, USA) using the manufacturer’s protocol. Nuclear ribosomal DNA internal transcribed spacer (hereafter ITS) was amplified with the primers ITS-F2-Psitta (5′-TCGCAGTATGCTCCGTATTG-3′) and ITS-R2-Psitta (5′-TCGTAACAAGGTTTCCGTAGG-3′) designed for the genus (Ornelas et al. 2016), whereas for the plastid trnF intergenic spacer region (hereafter trnL-F) we used the universal primers ‘e’ and ‘f’ (Taberlet et al. 1991). Protocols for PCR reactions and for sequencing the PCR products were described in Ornelas et al. (2016).
Alignment and phylogenetic analyses. Edited sequences using Sequencher 4.1.4 were imported into Se-Al 2.0a111 (Rambaut 2007) and aligned manually. For trnL-F, indels were introduced during alignment, but the resulting gaps were not coded. For ITS, the alignment was guided by reference to published studies of Loranthaceae (Wilson & Calvin 2006, Vidal-Russell & Nickrent 2008a,b, Ornelas et al. 2016, Pérez-Crespo et al. 2017), and indels were not coded. Polymorphic sites in the ITS region were coded as ambiguous. It is possible that divergent ITS paralogues may have been amplified in this study, including pseudo genes and recombinants. However, other phylogenetic studies with different Loranthaceae using ITS did not encounter evidence of paralogy or mention evidence of paralogues (Wilson & Calvin 2006, Vidal-Russell & Nickrent 2008a,b, Ornelas et al. 2016, Pérez-Crespo et al. 2017, Lopez-Laphitz et al. 2018). There is almost no missing data in the matrices, with 0.6 % of incompletely sequenced genes in ITS (404), 1.8 % in trnL-F (689), and 1 % in the combined ITS+trnL-F matrix (1,102). Individual gene alignments were saved as NEXUS files and then concatenated using Mesquite 3.01 (Maddison & Maddison 2011). The newly generated sequences were submitted to GenBank (Appendix 1).
The ITS and trnL-F datasets and the combined dataset were analyzed using Bayesian inference (BI), maximum likelihood (ML) and maximum parsimony (MP) methods. For the BI and ML analyses, appropriate substitution models were estimated using jMODELTEST 2.1.7 (Darriba et al. 2012), F81 + G (trnL-F) and GTR + I + G (ITS). The BI analyses were performed using MrBAYES 3.12 (Huelsenbeck & Ronquist 2001, Ronquist & Huelsenbeck 2003). BI analyses were run using the CIPRES Science Gateway (Miller et al. 2010) for the ITS, trnL-F and the combined datasets. Two parallel Markov chain Monte Carlo (MCMC) analyses were executed simultaneously, and each was run for 10 million generations, sampling every 1,000 generations. Bayesian posterior probability values were calculated from the sampled trees remaining after 2,500 samples were discarded as burn-in (Huelsenbeck & Ronquist 2001) to only include trees after stationary distribution was reached. The remaining trees were used to generate a 50 % majority-rule consensus tree, showing nodes with a posterior probability (PP) of 0.5 or more. We consider nodes significantly supported if posterior probabilities were ≥ 0.95 (Huelsenbeck & Ronquist 2001).
ML analyses were performed in RAxML 8.2.4 (Stamatakis 2014) under the general time-reversible nucleotide substitution model (GTR) and 1,000 non-parametric regular bootstraps using the CIPRES Science Gateway (Miller et al. 2010). Bootstrap support values were interpreted as indicating weak (50-70 %), moderate (71-80 %) and strong support (81-100 %).
The most parsimonious trees were obtained using the ratchet strategy (Nixon 1999) in Winclada 1.0000 (Nixon 1999-2002), running NONA 2.0 (Goloboff 1993), with nucleotide characters treated as unordered and equally weighted, 1,000 iterations, holding 10 trees per iteration with 10 % of the nodes constrained, and all other parameters set to default. Branch support was assessed using bootstrap resampling, 1,000 bootstrap-resampled pseudoreplicate matrices were each analyzed using 100 random addition sequences (multi*100). Ten trees were retained during TBR swapping after each search initiation (hold/10) using NONA and performed in WinClada, with the same interpretations of support level as in the ML analyses.
Divergence time estimation. We estimated divergence time under a Bayesian approach using the concatenated ITS and trnL-F sequence dataset. The ingroup comprised all sequences of the Psittacanthus ‘Bursera group’, and sequences from other Psittacanthus species and other representatives of Loranthaceae downloaded from GenBank were used as multiple outgroups (Vidal-Russell & Nickrent 2008a). Divergence time estimation was performed with BEAST 1.6.1 (Drummond & Rambaut 2007) using the uncorrelated lognormal relaxed molecular clock and the nucleotide substitution model GTR+G+I for the ITS sequence dataset and GTR+G for the trnL-F sequence dataset, suggested by jMODELTEST 0.1.1 (Posada 2008). The tree prior model was set using a coalescent approach assuming constant population size. To calibrate the root node, we constrained Nuytsia floribunda (Labill.) G. Don as sister to the aerial parasites based on Vidal-Russell & Nickrent (2008b). The divergence time between Nuytsia and the aerial parasites clade was used as secondary calibration, approximating a median age of 48.9 Ma (normal distribution, mean 48.9, SD 3.9, range 56.5-41.2 Ma) using pollen fossil of Loranthaceae according to Grímsson et al. (2017). The geometric mean of 5.798×10-9 substitutions per neutral site per year (s/s/y) was used to calibrate the tree based on the mean mutation rates of 4.13×10-9 s/s/y for ITS of herbaceous annual/perennial plants (Kay et al. 2006) and 8.24×10-9 s/s/y for trnL-F estimated for annual or perennial herbs (Richardson et al. 2001). The BEAST analysis was run two times for 100 million generations, sampling every 10,000 steps. We combined the log and trees files from each independent run using LOGCOMBINER 1.8.0 (Drummond & Rambaut 2007), then viewed the combined log file in TRACER 1.6 (http://tree.bio.ed.ac.uk/software/tracer/) to ensure that ESSs for all priors and the posterior distribution were > 200, making sure that parameter values were fluctuating at stable levels. Based on these results, the first 10 % of trees were discarded as burn-in, and the remaining trees were annotated and summarized as a maximum clade credibility tree with mean divergence times and 95 % highest posterior density (HPD) intervals of age estimates using TREEANNOTATOR 1.8.0 (Drummond & Rambaut 2007) and visualized in FIGTREE 1.3.1 (http://tree.bio.ed.ac.uk/software/figtree/).
Results
Molecular data. The ITS partition had 645 aligned positions (109 samples) and contained few ambiguous positions. This partition including the outgroup taxa had 348 variable characters (59.5 %), 324 (50.2 %) parsimony-informative, while considering only the ingroup had 102 variable characters (15.8 %), 98 (15.1 %) parsimony-informative. With outgroups included, the plastid trnL-F intergenic spacer region had 377 aligned positions (109 samples) with 173 (46.3 %) variable sites, 117 (31.3 %) parsimony informative. The alignment of this region considering only the ingroup had 17 parsimony informative sites (4.5 %). The concatenated alignment of the 109-taxon dataset consisted of 1020 aligned positions of which 557 (54.6 %) were variable sites and 441 (43.2 %) were parsimony-informative. Considering just the ingroup, the number of variable sites was 119 (11.6 %), of which 115 (11.2 %) were parsimony-informative.
Phylogenetic analyses. Tree topologies for single (ITS, trnL-F) gene partitions of the BI analysis, summarizing the clade support for all methods, are shown in Figure 1. Figure 2 shows the tree topologies using maximum likelihood (ML; Figure 2A-B), and strict consensus maximum parsimony (MP; Figure 2D-E). The ITS dataset showed that Psittacanthus species are monophyletic (BI = 1.0 PP, MLBS = 100 % bootstrap support, MPBS = 100 % bootstrap support; Figure 1D, Figure 2A, D), and that individuals from Honduras, P. palmeri and P. sonorae are closely related, forming a well-supported clade (1.0 PP, 98 % MLBS, 99 % MPBS). In the trnL-F, Psittacanthus was not monophyletic and provided very low resolution within Psittacanthus, with unresolved relationships in the separate trees (Figure 1E, Figure 2B, E). Despite this, the trnL-F trees suggest that samples from La Trinitaria (Chiapas) provisionally determined as P. palmeri and those from Honduras are closely related to individuals of P. palmeri from Santa María Tecomavaca (Oaxaca) and individuals from Ixtlahuacan (Jalisco) by all three methods (BI, ML, and MP; Figure 1E and Figure 2B, E). However, these samples and those of P. palmeri from Santiago Matatlán arise from a polytomy. Although the level of resolution between the ITS and trnL-F datasets and phylogenetic methods differed, the relationships among individuals from Honduras, P. palmeri and P. sonorae were largely congruent in trees generated using ITS and the combined ITS+trnL-F datasets (Figures 1 and 2).
Tree topologies for combined (ITS+trnL-F) gene partitions of the BI analysis, summarizing the clade support for all methods, are shown in Figure 3, and Figure 2 shows the tree topology using maximum likelihood (ML; Figure 2C), and the strict consensus maximum parsimony (MP; Figure 2F), that produced similar tree topologies (Figure 2C, F). The 50 % majority-rule consensus BI tree is shown in Figure 3 also including ML and MP bootstrap support values. All phylogenetic analyses indicate with strong support that the Psittacanthus species in our dataset are monophyletic (1.0 PP, 100 % MLBS, 99 % MPBS), with samples of P. palmeri and P. sonorae forming a strongly supported clade (1.0 PP, 99 % MLBS, 99 % MPBS), named here the ‘Bursera group’.
Within the ‘Bursera group’, samples of P. sonorae form a clade (1.0 PP, 100 % MLBS, 100 % MPBS) that is sister to P. nudus and P. palmeri (1.0 PP, 99 % MLBS, 99 % MPBS). However, phylogenetic relationships within the group formed by P. nudus and P. palmeri, as currently circumscribed, were not fully resolved. Both species, including individuals of P. palmeri from Jalisco, Oaxaca and Chiapas and the samples collected near the type locality in Honduras of P. nudus, formed a strongly supported clade (1.0 PP, 100 % MLBS, 100 % MPBS); the relationship between a well-supported, monophyletic group formed by individuals of P. nudus (1.0 PP, 100 % MLBS, 99 % MPBS) with other members of P. palmeri is supported, with the latter clearly paraphyletic with respect to the former. Within the ‘Bursera group’ two noteworthy structural changes were detected in the trnL-F region. One corresponds to a deletion of 7 bases (5’ GCCTAAAT-------ATGAAAAA 3’), and, within the same span of DNA, the other one to a 62-bp deletion (5’ AAAATAAA-------AACTAATT 3’) shared by all individuals in the ingroup, except individuals of P. palmeri from Santiago Matatlán (Oaxaca).
Divergence time estimation. The BEAST analyses placed the split between Psittacanthus and other Loranthaceae at 10.87 Ma (95 % HPD 18.42-3.94 Ma, Figure 4) and the origin of the Psittacanthus crown group in the Late Miocene, separating the P. ramiflorus clade from other Psittacanthus (8.31 Ma, 95 % HPD 14.14-2.95 Ma, Figure 4). The ‘Bursera group’ formed by samples of P. nudus, P. palmeri and P. sonorae split from the other Psittacanthus at 7.89 Ma (95 % HPD 13.47-2.83 Ma). Within the ‘Bursera group’, the early split between P. sonorae from P. palmeri/P. nudus clade was estimated at 5.62 Ma (95 % HPD 10.13-1.82 Ma) and the crown age of P. sonorae was estimated at 1.17 Ma (95 % HPD 2.38-0.28 Ma). Lastly, within P. palmeri, the age for the group of P. nudus samples from Honduras was estimated at 0.28 Ma (95 % HPD 0.64-0.02 Ma).
Discussion
Phylogenetic methods and markers. Phylogenetic results obtained in this study are consistent with previous phylogenetic analyses: the plastid marker trnL-F showing low resolution at species level in Loranthaceae (e.g., Vidal-Russell & Nickrent 2008a, Ornelas et al. 2016), whilst the ITS sequences have proved to be more informative showing greater resolution at species and genus level (Vidal-Russel & Nickrent 2008a, Zanjanchi & Saeidi-Mehrvarz 2015, Ornelas et al. 2016, Lopez-Laphitz et al. 2018). The trees obtained in this study showed topological differences between different partitions and very low resolution among Psittacanthus species using only the trnL-F dataset, with most of the branches poorly supported. Unlike the plastid tree, the trees retrieved using the nuclear ITS dataset showed better resolution of the phylogenetic relationships among species.
The trees obtained from the Bayesian analyses of the individual molecular markers (ITS and trnL-F) and the combined dataset differ in topology and degree of resolution. The trnL-F gene produced the poorest resolved tree (Figure 1D), and also contained the smallest number of nodes with posterior probability (PP) values above 0.90. The combined ITS and trnL-F dataset produced the most resolved tree, including the highest number of strongly supported nodes (Figure 3). The phylogenetic relationships of the major Psittacanthus clades based on ML and MP analyses of ITS+trnL-F showed some incongruence between the ML and MP topologies and the BI tree topology (e.g., split between P. rhynchanthus (Benth.) Kuijt and other Psittacanthus species), but most of these conflicted relationships received weak bootstrap support.
All BI, ML and MP analyses of the combined dataset resulted in similar tree topologies, showing strong support for the monophyly of Psittacanthus and phylogenetic relationships between species of the ‘Bursera group’ (Figure 3). Given that coalescence occurs four times faster in maternally inherited chloroplast genes than in biparentally inherited nuclear genes (Hudson 1990), this result is not unexpected (see also Amico et al. 2007).
Psittacanthus lineages, phylogeny and the status of P. nudus. The three used analytical methods (BI, ML, and MP) of the combined ITS+trnL-F analyses resolved the conflicts between the single-gene trees, and topologies were more congruent with previous phylogenetic analysis of Psittacanthus, including the unresolved relationships between members of subclade C (P. ramiflorus and allies) and those in suclade B (‘Bursera group’) and subclade A (Ornelas et al. 2016, Pérez-Crespo et al. 2017, Licona-Vera et al. 2018). Based on this we considered the ‘ITS+trnL-F tree’ to be the best estimate of the phylogeny relationships among samples for further discussion and taxonomic considerations, until further phylogenetic analysis includes more species and markers, treating gaps (generally representing an insertion or deletion; i.e., indels) with the input sequence alignment as missing data and coded characters in phylogeny estimation.
The ‘Bursera group’ (subclade B) is composed of two strongly supported clades: P. sonorae and P. palmeri (five populations, from Jalisco, Oaxaca, and Chiapas) plus samples of P. nudus from Honduras. The ITS+trnL-F tree (Figure 3) supports a topology in which the population from Honduras is more closely related to the Ixtlahuacan (Jalisco) population of P. palmeri than to the other population of that species (Oaxaca and Chiapas). Here, the strongest phylogenetic signal for these relationship derives from the ITS partition, as the chloroplast gene failed to distinguish this clade (Figure 1D-E). The lack of congruence between the two gene phylogenies could be due to hybridization events or to incomplete lineage sorting in which reciprocal monophyly has not been reached (Ornelas et al. 2016, Pérez-Crespo et al. 2017, Licona-Vera et al. 2018). Although we cannot exclude any of the two processes, the ITS phylogenetic pattern is partial evidence that the trnL-F undifferentiated populations could be incipiently divergent. A complete geographic sampling of P. palmeri and genotyping using next-generation sequencing (e.g., SNPs) will be necessary to address these issues. Strong support for the paraphyly of P. palmeri and putative samples of P. nudus from Honduras is seen with the nuclear and chloroplast partitions (Figure 1D-E) and in the combined analysis (Figures 2 and 3). However, the combined analysis does not resolve the relationship between Mexican populations of P. palmeri and samples of P. nudus from Honduras.
The data presented herein shows that samples from Honduras are not genetically distinct from those of P. palmeri, suggesting that P. palmeri and P. nudus do indeed represent the same species (possibly a species complex), or that the sampled population in Honduras derived from populations representing part of the distribution of a wider-ranging taxon, P. palmeri. Based on a recent specimen from Guatemala (Mario Véliz Pérez 18578; BIGU, MO), the distribution range of P. palmeri would not be restricted to Mexico. Furthermore, the collector of the holotype (A. Molina) believes that the species no longer exists at the type locality (Kuijt 2009).
The question of whether there could have been two species (P. nudus and P. palmeri) sympatric in this area may only be answered with the inclusion of the holotype or isotypes of P. nudus in the molecular analysis, but samples of the P. nudus holotype were not available. The present study suggests the combination of both taxa. However, a more thorough geographic and molecular sampling to be analyzed using a detailed phylogeographic approach, in combination with species distribution modeling would be needed to delimit geographic lineages.
Molecular dating of Psittacanthus and of the ‘Bursera group’. Most of our estimated ages for clades within Psittacanthus correspond to the Pliocene-Pleistocene epochs. Using a similar dataset that included the same Psittacanthus species and multiple samples for P. schiedeanus (Schltdl. & Cham.) G. Don, Ornelas et al. (2016) identified a mean of 5.41 Ma with a 95 % HPD time interval of 6.75-1.11 Ma for the age of the Psittacanthus crown, while the estimated divergence ages by Pérez-Crespo et al. (2017) using the same Psittacanthus species and multiple samples of P. calyculatus G. Don placed the origin of the Psittacanthus crown clade latter, in the Late Miocene (7.36 Ma, 95 % HPD 11.12-3.75 Ma). Here, samples of Psittacanthus formed a well-supported monophyletic clade, with diversification also occurring during the Late Miocene, with an even older mean age, but a wider HPD (8.31 Ma, 95 % HPD 14.14-2.95 Ma). Differences in age constraints (median age of 48.9 Ma in this study using pollen Loranthaceae fossils, vs. 28 Ma in other studies using substitution rates) for the tree root during dating analyzes can be attributed to discrepancies among these studies.
Within Psittacanthus, the group formed by samples of P. sonorae formed a well-supported monophyletic group, with a split between P. sonorae and P. palmeri/P. nudus samples occurring during the Pliocene (5.62 Ma). Thus, the divergence dates within the Psittacanthus ‘Bursera group’ appear to be related with those of its main host species (De-Nova et al. 2012), where the divergence of the host necessarily precedes the origin of the parasite. Divergence times for Bursera species parasitized by P. sonorae and its corresponding closest relatives were estimated between 7.7 and 4.6 Ma, for those parasitized by P. palmeri between 11.2 and 4.3 Ma, and for the only known host species for P. nudus (B. simaruba L. Sarg.) and its closest relative (B. itzae Lundell) the divergence time was estimated as 8.4 Ma (De-Nova et al. 2012; Table 2).
Mistletoe species | Bursera node | Divergence time (Ma) |
---|---|---|
Psittacanthus nudus | B. simaruba Sarg. / B. itzae Lundell | 8.4 |
Psittacanthus palmeri | B. cuneata Engl. / B. biflora Standl. | 10.15 |
B. heteresthes Bullock / B. fragrantissima Bullock | 6.35 | |
B. infernidialis Guevara & Rzed. / B. penicillata Engl. | 6.55 | |
B. palmeri S. Watson / B. stenophylla Sprague & L. Riley | 7.52 | |
B. bippinata Engl. /B. vejar-vazquezii Miranda-B. submoniliformis Engl. | 7.05 | |
B. sarukhanii Guevara & Rzed.-B. velutina Bullock / B. bicolor Engl. | 5.63 | |
B. copallifera (Sessé & Moc. ex DC.) Bullock / B. excelsa Engl. | 4.63 | |
B. grandifolia Engl. / B. instabilis McVaugh & Rzed. | 5.45 | |
B. multijuga Engl. / B. confusa (Rose) Engl. | 10.95 | |
B. galeottiana Engl. / B. arida Standl.-B. suntui C.A. Toledo-B. morelense Ramírez | 11.17 | |
B. fagaroides Engl. / B. aptera Ramírez | 6.32 | |
B. schlechtendalii Engl. / B. medranoana Rzed. & E. Ortiz | 4.33 | |
Psittacanthus sonorae | B. microphylla A. Gray / B. multifolia (Rose) Engl. | 7.72 |
B. hindsiana Engl. / B. cerasifolia Brandegee | 4.56 | |
B. laxiflora S. Watson / B. filicifolia Brandegee | 7.08 | |
B. fagaroides Engl. / B. aptera Ramírez | 6.32 |
Based on the Bursera phylogenetic tree (De-Nova et al. 2012), P. palmeri and P. sonorae could have colonized during the Late Miocene both ancestors of the Section Bursera and Section Bullockia, after having a significant increases in their diversification from Early to Middle Miocene (25-18 Ma), and ecological shifts from their ancestral seasonally dry tropical forest to xerophytic scrub (e.g., B. microphylla A. Gray). More recent lineages of Psittacanthus (P. mayanus Standl. & Steyerm., P. rhynchanthus, Figure 2) invaded ancestors of the Simaruba clade in the Middle Miocene, after shifting from the seasonally dry tropical forest to the tropical rain forest (De-Nova et al. 2012).
In this context, the divergence and evolution of lineages within the Psittacanthus “Bursera group” could be related to the colonization of new geographic areas, where potential hosts are already occurring (occupation of novel niches).
In addition, the colonization of these sites with different selection pressures (e.g., dry ecosystems) would require key morphological innovations, such as the terete, fleshy leaves unique to P. sonorae, small size (fruit ranges from ca. 7 mm long in P. sonorae, to at least 15 mm in P. macrantherus Eichler and 20 mm in P. acinarius Mart.; Kuijt 2009) and encapsulated cotyledons by dried viscin and enlarged cells of the vesicular viscin tissue (Kuijt 2009), and the seasonal deciduousness exhibited by P. nudus / P. palmeri (Appendix 2). Although morphological similarities between P. palmeri and P. nudus and their differences with P. sonorae are evident (Appendix 2), traits associated with host shifts leading these mistletoes to colonize and specialize on Bursera host species (e.g., haustoria, host-compatibility mechanisms) remain to be investigated.
Evolution of deciduousness and succulent leaves in Psittacanthus. Of all described species of Psittacanthus, P. nudus, P. palmeri and P. sonorae are the smallest (Kuijt 2009), and according to our analyses they share a common origin. The likely conspecific P. palmeri and P. nudus share lateral and umbellate inflorescences and petals with inner vermiform appendages, but differ in the shape of the basal ligule (Kuijt 2009; Appendix 2). According to Kuijt (2009), P. sonorae is succulent, while P. palmeri is possibly seasonally deciduous and P. nudus, previously known only from the type collected in Honduras, is apparently leafless.
The lack of leaves in P. nudus was not confirmed by Kuijt (2009). The leaflessness or deciduousness of P. nudus has been a botanical enigma since its publication (Molina 1952). Based on our field observations of P. palmeri in Oaxaca and Chiapas and a leafy branch of a specimen from Huehuetenango, Guatemala (BIGU, MV 18578; 15º 53´ 23˝ N, 91º 44´ 00˝ W; at 1,088 m) photographed by Mario Véliz (14 June 2007), and a specimen with leaves photographed by Eydi Yanina Guerrero (5 May 2016) near the type locality of P. nudus in Honduras (Table 1, Figure 1B), we confirm that P. palmeri is seasonally deciduous, and based on our phylogenetic studies, argue that P. palmeri and P. nudus are the same species. However, more field observations or phenological studies in these mistletoes throughout the seasons are needed to determine the start and ending of the leaf loss stage.
It is possible that the evolution of deciduousness in P. palmeri and P. nudus and of the terete leaves of P. sonorae are linked to their xeric habitats: seasonally dry tropical forest and xerophytic scrubs. A seasonally dry tropical forest was reconstructed as the ancestral vegetation type of Bursera and the most likely state for most internal nodes, and shifts to other vegetation types always involved a change from seasonally dry tropical forest to a different vegetation type, most frequently to xerophytic scrubs (De-Nova et al. 2012). Another species in the genus, P. divaricatus (Kunth) G. Don, possibly the closest relative of P. cordatus (Hoffmanns.) G. Don (Kuijt 2009; subclade A), occurs in the extremely xeric coastal areas of Ecuador and northern Peru, and has cordate-clasping, bluish-grey (glaucous) leaves (Kuijt 2009). Our observations indicate that P. palmeri is seasonally deciduous. In Figure 1B we show a photograph of an individual in vegetative condition that was taken near the type locality of P. nudus, and we have observed P. palmeri individuals without leaves during the non-breeding season.
The above observations are important because available information on deciduous mistletoes is sparse. According to a recent review on leaf production in mistletoe-host associations (Glatzel et al. 2017), most of the world’s mistletoes are evergreen, regardless of the foliar habit of their hosts. Deciduous mistletoes are rare and confined to a few species in Loranthaceae in Eurasia, to Misodendraceae, and the monospecific genus Desmaria (Loranthaceae) in southern South America. Our observations of deciduousness in P. palmeri and possibly P. nudus challenge Glatzel et al. (2017) conclusion that there are no deciduous mistletoes in the tropics and subtropics. Further comparative phenology of mistletoes in the ‘Bursera group’ is needed to analyze whether leaf deciduousness is synchronized with the deciduousness of their Bursera main hosts.