Thesium, García & Mucina & Nickrent, 2024

García, Miguel A., Mucina, Ladislav & Nickrent, Daniel L., 2024, A tough nutlet to crack: Resolving the phylogeny of Thesium (Thesiaceae), the largest genus in Santalales, TAXON 73 (1), pp. 190-236 : 226-228

Biogeography of Thesium

The extensive sampling in this study throughout the entire distributional area of Thesium ( Fig. 1 ) allows the reconstruction of the historical biogeography of the genus. Further studies including estimation of absolute divergence times will help to understand the dispersal and speciation events within the genus and allow comparison with the large-scale floristic patterns observed in other groups of African origin.

The monotypic genus Lacomucinaea has a SW African distribution ( Nickrent & García, 2015) and Osyridicarpos is more common in SE Africa with populations extending north to Ethiopia along the Eastern Arc African Mountains (EAAM). Both genera are the extant lineages most closely related to Thesium . Moore & al. (2010), using ancestral range reconstruction, suggested a southern African origin of Thesium in the Late Eocene, with a crown age of 39.1 ± 11.9 million years. Lacomucinaea and Osyridicarpos were not included in their analyses, but their current distribution supports their common origin in southern Africa. The topology of both ITS and chloroplast trees resolves, with the highest support in all the analyses, two main lineages in the genus, the CMPB lineage (clade 2), mostly including the Eurasian species, and the African lineage (clade 26).

Biogeography of the CMPB lineage. — The CMPB lineage has a remarkable disjunct distribution pattern because it includes the Cape endemic species of Thesium subg. Hagnothesium (former genus Thesidium ) nested within Mediterranean, Canarian, and Eurasian species (T. subg. Thesium ), not within the main African lineage. Both chloroplast and ITS phylogenies highly support this relationship. Moore & al. (2010) dated the divergence of both subgenera to the Oligocene (33.7–23.8 million years ago), preceding other plant groups with origin in southern Africa that dispersed north via East Africa with the Late-Miocene uplift of the EAAM ( Pokorny & al., 2015).

The early-diverging lineages in the type subgenus include the NW African Thesium mauritanicum and T. sect. Kunkeliella , endemic to the Canary Islands. Even though T. mauritanicum might be the closest extant relative of T. sect. Kunkeliella , their sister relationship is not resolved in our phylogenies. Our data indicate that T. sect. Kunkeliella is monophyletic and its presence in the Canary Islands is the result of a single dispersal event followed by local in situ diversification in the islands with one species in Gran Canaria, three in Tenerife and one in La Palma ( Rodríguez-Rodríguez & al., 2022).

A second lineage, the Paleoboreal clade (clade 7), is well-supported with both ITS and chloroplast markers and includes the remainder species of Thesium subg. Thesium . Well-supported backbone relationships in this lineage are not always seen, but among the well-supported clades the Core Eurasian clade (12), that includes most of the diversity in terms of number of species, emerges as an independent lineage with a mostly W Asian distribution. The ITS phylogeny shows that T. humile , a circum-Mediterranean and Canarian species, is moderately supported as sister to the Macranthia clade (9), which includes species distributed in the E Mediterranean region and W Asia. A second lineage, the Procumbens clade (11), is present in W Asia and SE Europe and finally, the C Asian T. minkwitzianum whose closest relatives are not resolved. This scenario suggests that after the migration from the Cape region to the Mediterranean, one or several dispersals to W Asia and subsequent speciation generated the extant diversity of species of Thesium with mostly E Mediterranean and W Asian distributions, especially in Anatolia and other regions of the Middle East.

The Core Eurasian clade (12) represents the dispersal and subsequent speciation in Europe that produced many of the endemic species of the continent, especially in southern mountain areas. In the ITS phylogeny, the Linophylla clade (17) together with T. catalaunicum and T. rostratum are the closest European relatives of a group of species of Asian distribution. However, chloroplast markers resolve the Linophylla clade as being more closely related to other European species. Independent of the markers used, the phylogenies within the Core Eurasian clade are congruent with the concept of one to several dispersals from Europe into Asia and subsequent speciation in Siberia, C Asia, China, and the Himalayas. This represents the second Asian radiation and subsequently migration into E Australia ( T. australe ).

Biogeography of the African lineage. — The topology of ITS and chloroplast trees resolves three early-diverging lineages (clades 28–30) including species with current distributions in the Succulent Karoo and Nama-Karoo semi-deserts (clade 29) and Fynbos (clades 28 and 30). One species, Thesium triflorum , is also present in areas of the East African mountains, probably as the result of a later dispersal to the north from South Africa ( Brummitt, 1976). Mountains of the Cape Fold Belt were important for the further diversification of Thesium in the Cape as revealed by the phylogenetic position of T. nautimontanum , an endemic from the Matroosberg Mt. ( García & al., 2018), as sister to the entire African clade in the ITS phylogeny or in the chloroplast phylogeny as sister to the African species (with the exception of clades 28–30 that were resolved as earlier diverging lineages). The rapid diversification in the fynbos between the Great Escarpment and the coast generated the current diversity of many endemic species, all of which are included T. subg. Frisea (clade 32).

Both nuclear and chloroplast phylogenies resolved Thesium subg. Psilothesium (clade 48) as sister to the Core Cape clade, representing the broad diversification of the genus in the grassy biomes of E and N South Africa and the radiation into tropical Africa. The ITS topology resolves the Gnidiaceum clade (49) as sister to the rest of the subgenus. This clade includes species from the SW ranges of the Cape Fold Belt ( T. oresigenum and Thesium sp. nov. 5574) as sister to others from the SE Great Escarpment and E Cape Fold Belt, supporting the biogeographical link between both areas ( Clark & al., 2011, 2012). Although the topology of the chloroplast phylogeny does not place T. oresigenum and Thesium sp. nov. 5574 in the Gnidiaceum clade, both species were resolved as successive sisters to the grassland and tropical African radiation.

Subsequent radiation in the tropics is, in part, represented by clades 52 and 60 that include species lacking an apical beard and mostly distributed in savannas in regions spanning East and West Africa. Our data suggest at least two long-distance dispersals in clade 52, one to South America and at least one to Madagascar. The monophyly of the South American species suggests a single long-distance dispersal from tropical Africa. Although we only have sequence data of Thesium madagascariense , the morphological features typical of the Brachyblast clade (54) suggest that another Madagascan endemic species, T. perrieri , belongs in this group. Because the latter could not be sampled it is unknown whether the presence of these two species in Madagascar is the result of one or two independent dispersal events from the African continent.

Although not well-supported, the topologies of ITS and chloroplast trees suggest an independent diversification of the leafy species of Thesium subg. Psilothesium in the grassdominated biomes of southern Africa with further range expansion to the tropical Africa. The chloroplast phylogenies highly support these species as monophyletic and with lower support its sister relationship with T. sect. Barbata species of the Gnidiaceum clade and other species mostly distributed around the Drakensberg mountains. This area has been considered a “stepping-stone” between the CFR and the so-called Afrotemperate Region ( White, 1978; Galley & al., 2007) encompassing high elevations of mountain ranges of southern and East Africa (and to a minor extent Central and West Africa). Migration northwards is supported by the presence in this clade of species endemic to East African mountains and highlands such as T. kilimandscharicum , T. dolichomeras , T. whyteanum , etc. and its expansion to Ethiopia and SE of the Arabia Peninsula (species such as T. radicans or T. hararensis ). Expansion and diversification took place also in lowland areas of the subtropical and tropical grasslands of SE Africa and the tropical Central and West Africa as well as more recent in situ speciations suggested by the overlap of morphological characters among species in this taxonomically difficult group. Further events of long-distance dispersal are seen in clade 64, one to Madagascar ( T. decaryanum ) and another one to E and SE Asia ( T. psilotoides ). Thesium wightianum (unsampled) from SE India, shares morphological characters with T. radicans and molecular data from this species might reveal another case of dispersal to Asia from NE Africa in this clade (see discussion of clade 64).

Conflict between the ITS and chloroplast datasets

Although gene trees derived from the two subcellular compartments were generally congruent, ITS and chloroplast trees showed remarkable incongruences at the specific or even infraspecific levels. These incongruences are not observed equally along the main lineages. The Eurasian, grassland and tropical African lineages show, in general, high congruence between datasets with particular exceptions previously discussed. However, and in agreement with the results obtained by Zhigila & al. (2020), the Cape radiation ( Thesium subg. Frisea ) is characterized by the high incongruence of chloroplast and nuclear trees. Within the subgenus this incongruence is not detected in the monophyletic T. sect. Frisea (clade 39) because of the lack of internal resolution in this clade but it is interesting that none of the species of this section are resolved among other clades of the Cape radiation on the chloroplast trees except for accession 5460. Biological factors known to cause phylogenetic incongruence in plants include paralogy, hybridization, incomplete lineage sorting, and horizontal gene transfer.

Paralogy. — Paralogy cannot explain the incongruences detected because there is very low ITS intraspecific variation and in those cases with polymorphic ribotypes the paralogous sequences obtained after cloning were closely related to each other. Moreover, clades resolved for Thesium subg. Frisea in the ITS trees largely correspond with morphological features and species are mostly resolved as monophyletic, unlike in the chloroplast phylogeny.

Hybridization. — Hybridization has been identified as one of the most common sources of incongruence between nuclear and plastid phylogenies. Although hybridization probably occurs in Thesium , it has yet to be demonstrated and by itself cannot explain the widespread incongruences found in the Cape clade. Hendrych (1972) discussed three cases described in the literature as possible interspecific hybrids and his conclusion was that hybrid origin could not be proven. We sequenced cpDNA from two or more individuals in 25 species of T. subg. Frisea and 14 of them were resolved as polyphyletic on the chloroplast but not on the ITS trees. They were resolved in two or more well-supported clades together with individuals of species very different in vegetative, inflorescence, floral and fruit morphology, thus ruling out the introgression of nuclear genes. As an example, T. strictum , a common and widespread species in the fynbos, is distinctive given its wand-plant architecture and other morphological features. It was resolved as monophyletic with strong support on the ITS tree (clade 36). In contrast, on the chloroplast trees these accessions were placed among several well-supported clades with species such as T. pubescens , T. pycnanthum , T. hispidulum , and T. prostratum , all morphologically quite different. If hybridization and successive backcrosses were involved, without convergent evolution of the nuclear ribosomal cistron, one would expect to find cases of ITS sequence polymorphisms with ribotypes of the same species resolved on different clades on the ITS trees. In the case of convergent evolution, it should have occurred in all the species and always homogenized towards the paternal progenitor because species are monophyletic on the ITS trees and clades are congruent with morphology.

Stochastic incomplete lineage sorting of ancestral polymorphisms. — Incomplete lineage sorting might explain some incongruences between nuclear and chloroplast phylogenies, especially under high rates of diversification. This biological factor was considered by Zhigila & al. (2020) as the most probable explanation of the incongruences in Thesium subg. Frisea . This hypothesis is supported by the fact that incongruences seem random and not geographically biased. However, it is difficult to use incomplete lineage sorting to explain the incongruence found by Zhigila & al. (2020) of one individual of T. minus (subg. Hagnothesium ) resolved in T. subg. Frisea on their concatenated chloroplast tree. It is more probable that the cpDNA in this individual was acquired after the diversification of the four subgenera. To our knowledge no cases have been documented in plants of maintenance of ancestral polymorphism long after speciation or at least with such a wide sequence diversity among individuals of the same species as we found in T. subg. Frisea . In this work we have detected polyphyletic cpDNA haplotypes among individuals of different populations of the same species, but it remains to be studied if it occurs also within the same population. Although incomplete lineage sorting in T. subg. Frisea or even deep reticulation might be involved, it alone cannot explain the incongruences found in T. subg. Frisea by Zhigila & al. (2020) and in this work.

Horizontal gene transfer. — Horizontal gene transfer (HGT) is recognized to have contributed to the genome plasticity and adaptive evolution of prokaryotes and unicellular eukaryotes, and played an important role in the origin and evolution of green plants ( Chen & al., 2021). HGT has been documented to be frequent in several groups of parasitic plants ( Davis & Xi, 2015). Normally HGT involves particular genes but these transfers can be large, such as the acquisition of entire mitochondrial genomes from three green algae, one moss and fragments of other angiosperms in Amborella ( Rice & al., 2013). Although HGT of mitochondrial genes is frequent, especially in parasitic plants, it only rarely affects plastid genomes ( Sánchez-Puerta, 2014). We sequenced two regions of the chloroplast genome from different locations. The topologies recovered with each of them are largely congruent (see, e.g., suppl. Figs. S8, S11) and similar results were obtained by Zhigila & al. (2020) comparing their trnLF and matK phylogenies. This suggests that whole plastomes, and not just fragments, have been transmitted horizontally between species. Chloroplast capture in plants is typically associated with reticulation events. However, asexual chloroplast capture in the absence of nuclear introgression has been demonstrated to occur through natural grafts ( Stegemann & al., 2012). Plastids can dedifferentiate into motile ameboid organelles and move into neighboring cells through connective pores after cell wall disintegration ( Hertle & al., 2021). In parasitic plants, intimate connections between cells of the host and parasite are formed during haustoria development, which mechanistically resemble grafting. In Thesium , autoparasitism through selfhaustoria and connections between different individuals are known to be common ( Melnyk & Meyerowitz, 2015). Thesium subg. Frisea is greatly diversified, and several species commonly grow in the same area in close proximity to each other. This might allow the formation of haustorial connections among individuals of different species, thus in theory HGT of whole plastids could occur between them. Although this is proposed here as a hypothesis, experimentation could yield significant insights to help explain the observations. Thesium offers an attractive group in which to investigate whether this phenomenon is actually taking place and whether it explains the unprecedented level of incongruence documented here between nuclear and chloroplast gene trees.

Figures

Fig. 1. Map showing the distribution of Thesium worldwide. Data from GBIF were vetted and used for generating the first iteration. Additional information was obtained from specimen records given in various Floras.