Res.I.P. – an unprofessional science (and other things) blog: Trivial but illogical – reconstructing the biogeographic history of the Loranthaceae (again)

In 2007, a short but nice paper by Vidal-Russell & Nickrent provided a scenario for the unfolding of the Loranthaceae, a plant family of mostly epiphytic tree parasites. Recently, they teamed up with a Chinese group (Liu, Le et al. 2018) to provide a new, and totally unexpected hypothesis.
In this post, I’ll be using this standard-made study (pitted with some unfounded and wrong statements) as an example to show:

Common sense would have sufficed to come to the same conclusion that Liu et al. on the background of the relatively few but long-known unambiguous phylogenetic relationships within the group (Wilson & Calvin 2006; Vidal-Russell & Nickrent 2008a) — we don't need explicit biogeographic inference to support trivial observations.
Address a few shortcomings regarding the analysis, commonly found in similar studies — mainly, should one use a European Eocene fossil to inform the minimum age of the most-recent common ancestor of a clade that is reconstructed as surely Australasian?
Map the result of the authors' top-down biogeographic analysis on palaeogeographic maps showing the known fossil record — maps we (Grímsson, Kapli et al. 2017) published five months before Liu et al. submitted their paper, and in an open access/open data journal (and promoting peer review transparency; Liu et al. opted for the opposite).
Outline what could (have) been done.

The basic situation – fast ancient radiation(s) but geographic lineage sorting

The Loranthaceae [Wikipedia/APW] are not an easy group for molecular phylogenetic analyses (Grimm 2017 [PDF]; Grímsson, Grimm & Zetter 2018 — open access/open data). Nevertheless, some morphologically defined groups (see Nickrent, Malécot et al. 2010 for the systematics) found readily support from the first molecular trees (Wilson & Calvin 2006; Vidal-Russell & Nickrent 2008a). Unfortunately, those first studies show only cladograms and their data matrices have never been published. And the group appeared to be deserted. So, when a colleague asked me in 2014 to generate some molecular phylogenetic framework for their palynological analysis of the family (Grímsson, Grimm & Zetter 2018; submitted December 2016, reviewed by two of the authors of Liu et al., online September 2017), I had to harvest the gene banks and re-do the analysis with what I could find. Which was not optimal, but served our needs.

The maximum likelihood tree shown in Grímsson, Grimm & Zetter (2018), using genus consensus sequences generated for two nuclear and three plastid gene regions (one non-coding). See my last post for the corresponding figure in Liu et al. (2018)

As you can see, some clades are straightforward (the tree is rooted with the most distinct of the three surviving root parasites, Nuytsia, following the results of the 2006 and 2008 papers): long root branches with high support. Others, like the Erytrantheae clade make sense, but lack support (as before, but there is also no strong contradictory signal in the data). A few, e.g. the placement of Aetanthus, are simply or partly wrong (missing and potentially mediocre data, according to the researcher that uploaded these data). The "backbone", the inter-tribal (-eae)/ -subtribal (-inae) relationships, is largely unresolved (mostly: lack of consistent or discriminating-at-all signal). The only difference between our and the 2006 and 2008 trees (and the 2015 by Su et al.) is that our doesn't mask the extent of ambiguous signal.

We were just interested in placing pollen into a systematic-phylogenetic framework, and for that, the bootstrap support consensus network showing the competing alternatives (when existing) served well.

The approach in our first Loranthaceae paper (Grímsson, Grimm & Zetter 2018). Background: a reduced (terminal clades represened by triangles) bootstrap support network rooted with Nuytsia; the images show the often characteristic pollen for each genus. Green backgrounds – the three root parasites, orange – the strongly different pollen of Tupeia showing a type shared by many other Santalaes instead of the Loranthaceae-typical triangular grains (which we considered as alternative root, of course "absurd" as reviewer #1 informed us; but see Grímsson, Kapli et al. 2017).

Just by looking at the above Nuytsia-rooted and geographically annotated tree (and the competing alternatives seen in the boostrap consensus network), the biogeographic history appears clear (from a modern-day perspective).

The family's origin and initial radiation must have been in Australasia (blue signatures), the home of the supposedly first diverging root parasite (Nuytsia), many Elytrantheae – an early diverged subtribe, generally shorter branched (note their relative phylogenetic closeness to the root node) –, an early, very distinct but small branch of Lorantheae ("Clade H" in Vidal-Russell & Nickrent 2008a, the Ileostylinae), and, finally in one ("Clade I") of the two main groups within the core Lorantheae ("Clade I+J"). Any other scenario would require multiple dispersals into Australasia, failing Ockham's famous razor. Notably the oldest records of Loranthaceae were all from Tasmania, and the otherwise least diverged modern genus (genetically), the palynologically odd Tupeia (and only non-South American member of the Psittacantheae) is found across (warm temperate) New Zealand. A neat fit.

Noting the low support along the backbone, the first radiations were a quick and dirty thing.

The slower-evolving (genetically), all other Psittacantheae and the root parasite Gaiadendron, went west. Vidal-Russell and Nickrent (2007) had the hypothesis of a "proto-Tupeia" (note the naming!) connecting the Gondwana fragments Australasia, Antarctica and South America. Only two South Americans are placed apart from the rest in our tree (similar in earlier trees), but when one explores the support patterns (when branch support is ambiguous, do exploratory data analysis!) it becomes clear that there is no reason to reject the hypothesis of a single dispersal (note the low support for alternatives, in particular the one seen in the tree; but see also the accD tree below).

A tree based on the accD plastid gene data included in Su et al.'s (2015) up-to-seven-genes data set (Grimm 2017, Fig. S6-3). Note the position of the two root parasites and the generally high branch support (less taxa, higher support). I wonder why Liu et al. opted against adding more data from this gene region ... can't Americans and Chinese to get their hand at least on a few Eurasian taxa?

[In 2008 they inferred (much too) young crown ages and a "basal" root-parasitic grade (ingroup-outgroup long-branch attraction, here inevitable: even sister families of the Santalales are very distinct, hence I didn't include them for our analyses to avoid outgroup-induced branching artefacts). So they modified their theory to the one re-booted in the 2018 paper, first the root-parasites radiated and migrated from Australasia, then the once-evolved aerial parasites followed their footsteps.
PS Taking the young dates as granted, some palynologists wrote (because certain reviewers forced them, too??) that most fossil Loranthaceae in the Southern Hemisphere must hence be extinct root parasitic lineages. Palaeontologists, when will you ever learn to read and judge the quality of chronograms? If you can't do it, just ask somebody who can. Inter-disciplinary exchange, you know...]

A second main lineage, the faster evolving Lorantheae, expanded into Eurasia (note the position of the purely East Asian genera), with one sublineage reaching Africa. Before doing so, their common ancestor(s) obviously went through some bottleneck situations (note the very prominent root branch). Bottlenecks are important for biogeographic analyses, because if something migrates or disperses over long distances, it probably will pass the one or other. Each bottleneck can increase genetic uniqueness, the result are longer branches. A (rough) rule of thumb is the farther I go, the more distinct I become (genetically). This is why probabilistic biogeographic inferences outperform the old parsimony-based ones.

The range expansion of the Lorantheae from Australasia happened more than once. "Clade G" (the Loranthinae) were the first spreading into Eurasia from Australasia. Some point later "Clade I+J", which diverged about the same time than its sibling clades, split into a staying-home (Australasian) and going-abroad (into East Asia) party. And while the staying-home party did follow later (the yellow dots in Clade I), the East Asian party ("Clade J") expanded into western Eurasia and Africa, and a few went back to the old home. Loranthaceae are largely tropical-subtropical groups (with a few, but interesting outposts in the temperate zone) with high dispersal capacity (also found across South Pacific Islands), such groups can freely migrate between the modern-day Indomalayan and Australasian realms (essentially no barriers; see also two of the three examples I used for my Median network promotion post: Ixora and Hoya).

Distribution of main Loranthaceae lineages. Pink = New World Psittacantheae with an outpost in New Zealand, Tupeia, the genus without Loranthaceae-typical pollen (fields within: the mono- or ditypic root parasite Gaiadendron); yellow = Old World-Australasian Lorantheae; blue margin = Elytrantheae; light green, Nuytsia, the genetically most distinct genus and representing the putative first-diverging lineage.

Because it's trivial from a modern-day perspective, but not so when looking at the fossil record [Palaeogene/Neogene], we (Grímsson, Kapli et al. 2017) did several molecular dating experiments (using and testing different rooting scenarios and including only ingroup fossils) but no explicit phylogeographic inference when describing new Loranthaceae pollen from the Northern Hemisphere. Because the needed data were missing. Liu et al. re-produced our dating results (re-using two of our fossil) and filled this gap in knowledge.

Liu et al.'s explicit biogeographic reconstruction

Of course, common sense as applied above (demeaned as "explanatory" in today's pseudo-hard biological sciences) is not enough. Only explicit inferences can tell us how it is (was). So, what did Liu et al. find instead using a much better data set (I assume, the authors are not allowed sharing their data, see my last post [PDF with the correspondance]) and being guided by the nigh-expert on these plants? It can be seen in their graphic abstract (the paper itself is paywalled, and the used data matrix naturally not accessible).

Including the sister families as outgroup (and a possibly artefactual outgroup-based root) and up-to-date analyses methods they find ... the same.

The Loranthaceae all come from Australasia (here defined using Wallace's famous line, which has never really been a great barrier for plants), and while some stayed (two of the root parasites and the Elytrantheae), others went early into the Americas via Antarctica (Gaiadendron, Psittacantheae).

Not long after this first radiation, some of the Lorantheae escaped Australia into Asia (and maybe back again more recently) to colonise Africa ("after" 30 Ma) and Europe ("after" 10 Ma).

Trivial situations will give trivial reconstructions.

When all members of a lineage are African, any method, no matter how fancy, will reconstruct Africa as the ancestral area of this lineage (see also this open access paper, showing similarly surprising results, there are many more of this kind — don't be disappointed, the title is a bit misleading).
When the earliest diverged and least diverged members of a focal group are Australian, it's probable that Australia was the home of the common ancestor.
If two regions are mixed (e.g. Australasians and Asians), the point of origin is more probable the one found in the first diverging sublineages, unless they are much more distant to the according root node.

And where it's not trivial, well, then the reconstructions will struggle to decide, too (see the reconstruction for the deeper and deepest nodes in Liu et al.'s area-chronogram). The only but important difference is that probabilistic methods will give you values to play with, alternatives we can discuss (rarely done), whereas the old parsimony-based methods just produced a large question mark or something unreasonable (fun-fact: we once submitted a paper to show that, a decade ago using our beloved Fagus as a case, but it got lost in the Forest of Reviews). Plants are generally trickier than animals regarding biogeography, they don't court to mate, and many don't bother about water bodies (including large ones, see e.g. Denk, Velitzelos et al. 2015). But if there's e.g. not enough rain during growing season, or too much snow and frost in winter, it's directly no-go area (see my post and D. Morrison's The Wine Gourd post about Köppen and vine).

[PS Renner (2004) provides a nice review on trans-oceanic dispersal in tropical plants with some relevance for Liu et al.’s discussion. They cite it and use the concept to explain the “long-distance dispersal” and “disjunction of Loranthaceae between continental Africa and Madagascar” – hardly long distance … but getting from Australia to Asia in the Eocene is a different thing.]


The lineage-through-time plot in Grímsson, Kapli et al. 2017, the paper we published five months before Liu et al. submitted theirs where this result (Scenario 4 equals their preferred topology) is reproduced (see below) and discussed (broadly) as genuinely new.

Illogic assumptions and interpretations

The devil lies in details. Liu et al. did (like so many others) not show a phylogram, only a cladogram and chronograms. So, we have little idea about the primary branch lengths, the amount of change within and between the clades. I can't show a phylogram, either, because their expert co-author advised the corresponding author to not send me the data matrix they used.

[Maybe a phylogram can be found in the online supplement, but it's also paywalled. Thank you, Elsevier, for your dedication to keep science under the lid.]

Let's first look at the used constraints to fix minimum ages (see also my comment on ResearchGate, How not to make a phylogeographic study and The most common errors regarding node dating. Liu et al. write in Material & Methods:

Most fossils recognized as Santalales are represented by pollen grains of Cretaceous and Tertiary age (Vidal-Russell and Nickrent, 2008b). We gave two fossil calibrations for the outgroups of Loranthaceae. The crown age of the tribe Anacoloseae of Aptandraceae was constrained to 70 Ma (95% HPD: 66.0–72.1 Ma) based on the fossil of Anacolosidites recorded since the Maastrichtian (Malécot and Lobreau-Callen, 2005). The fossil pollen of Misodendrum (as Compositoipollenites)
was recorded from middle Eocene (ca. 45 Ma) (Zamaloa and Fernández, 2016). Hence the crown age of Misodendraceae [Misodendron is sister to all others] was constrained to 45 Ma (95% HPD: 41.2–48.6 Ma). The stem age of Loranthaceae was constrained to 70 Ma (95% HPD: 69.4–72.6 Ma) based on the fossil of Cranwellia (Mildenhall, 1980; Macphail et al., 2012) [as in Vidal-Russell & Nickrent 2008, on its own leading to much too young estimates for the Loranthaceae; Grímsson Kapli 2017, hence...] The crown node of the tribe Lorantheae was constrained to 42.8 Ma (95% HPD: 37.8–47.8 Ma) according to the fossil pollen Changchang MT identified as Taxillus, Scurrula and Amyema (Grímsson et al., 2017b). The crown node [= most-recent common ancestor, MRCA] of the tribe Elytrantheae was constrained to 39.6 Ma (95% HPD: 38–41.2 Ma) according to the fossil pollen Profen MT3 (Grímsson et al., 2017b).

Mostly following our suggestions: for each pollen type described, we included a Use-as-age-constraint paragraph. We also used the Changchang MT as constraint for a minimum stem (root) age of the core Lorantheae (clade I+J), which in Liu et al.'s tree equals the Lorantheae crown node.
Same for the Profen MT3 and its Elytrantheae-like counterparts, which Liu et al. however decided to put one node up (reason may be stated in the paywalled supplement).

Use as age constraint
Here we used Profen MT3, MT4 and MT5 to constrain the root age of the Elytrantheae, i.e. the minimum age of the MRCA of Notanthera and Elytrantheae (scenarios 1–3). Further studies of modern pollen of Elytrantheae at and below the genus level and more genetic data are needed [hasn't been done] to decide whether the Profen MT3, and the related Profen MT4 and MT5, are already indicative for a first divergence within the Elytrantheae and can be placed more decisively within the Elytrantheae subtree.

Since we have no idea about the exact age of the stratigraphic unit in which we found these pollen grains, we didn't constrain the mid-points 42.8 (late Lutetian) or 39.6 Ma (mid-Bartonian) but drew a random age from the possible time interval: 47.8–37.8 Ma for the Changchang MT and 41.2–38 Ma for the Profen MTs (details provided in our openly accessible, doi-ed Supplement File S1 — the beauty of publishing in PeerJ and not Elsevier's dustbins!).

But according to Liu et al.'s biogeographic inference, the MRCA of the Elytrantheae, fixed to ±43 Ma, was probably Australasian, but the pollen is from exactly the other side from the world! Quite an enigma.

But easy to solve:

Either one includes this information about past distribution in the analysis: the Elytrantheae were not Australasian as reconstructed, but quasi-ubiquitous at 43 Ma (we would not force them to be just European, because there are equally old Loranthaceae pollen in Australia that – when studied using high-resolution SEM microscopy may well include Elytrantheae).
Or, we interpret the European Elytrantheae as an extinct sister lineage of the surviving, originally Australasian Elytrantheae; adding a shadow node (at 43 Ma) and accordingly coded OTU (near-zero tip) into our ultrametrised tree.

For the Changchang MT, we have the same problem, only that it is not so bothersome. The MRCA of Lorantheae is probably Australasian (and discussed as such), but also may be Asian with a probability of ~ 33% according the reconstruction (something the authors generally ignore). This is a good example why one should rely on a probabilistic framework in biogeography, but still should keep in mind the structure of the tree.

All three main Lorantheae lineages, the diffuse (geographically) Loranthinae, the Australasian Ileostylinae, and the probably original Asian Core Lorantheae have prominent roots.
The tips of the subclades differ: although Loranthinae and Ileostylinae are a bit closer to the Lorantheae-MRCA (phylogenetically, i.e. root-tip distance via the tree), they are not closer than the Asian members of the Core Lorantheae clade, but their Australasian and even more their African siblings are generally longer branched.
The second node in the Lorantheae subtree (the MRCA of Loranthinae and Ileostylinae) is very short-branched and lacks support.

The maximum likelihood bootstrap consensus network for Su et al.'s Loranthaceae matK data (likely the same than used by Liu et al.) This plastid gene enforces the basic tree topology, green edges refer to those seen in the "7-gene" gene tree, competing or matK-preferred alternatives (Grimm 2017, Fig. S6-4). Note the amiguity regarding primary relationships within aerial parasites, including the relationship between the three main Lorantheae groups.

If the out-of-Australia hypothesis is a good one for the Lorantheae, this means that the common ancestors of all three Lorantheae lineages underwent a massive bootleneck, before they could re-radiate, including those that stayed back home: a single or two species survived and one or two escaped.

Or, the common ancestors of all Lorantheae underwent a quick, large radiation spreading across Eurasia and Australasia, then their subsequent areas broke up and remained isolated for quite a time: subtropical-temperate (western) Eurasia – Loranthinae, tropical (South and East) Asia – Core Lorantheae, Australasia – Ileostylinae.

Which of those hypotheses is true, cannot be decided by top-down molecular inferences (both can explain the basic molecular differentiation patterns), but would require a better grasp on the actual fossil record of the three lineages. Liu et al. decided to ignore this entirely (it didn't fit their trivial reconstruction), but our Eocene/Oligocene pollen in Europe includes forms that are possibly very early Loranthinae, and surely Lorantheae (including their precursors, i.e. stem Lorantheae), and Loranthinae pollen can be found still in the Miocene. In contrast to what Liu et al. discuss, the Loranthus-lineage could have easily survived the Plio-Pleistocene climate fluctuations in mild-temperate/subtropical refuges like their main hosts (Fagaceae: oaks, beeches and relatives) until today (Loranthus has a warm temperate Köppen signature, probably not so different form e.g. oaks of sections Cerris and Ilex; cf. Denk et al. 2013, 2017; Grímsson et al. 2016).

Another thing that completely run unchecked by the confidential peer review is the general misconception of the authors that node dating estimates fixed by fossils represent exact dates or even maxima. In the entire discussion the authors refer to "at" and "after" xxx Ma, but what they have are minimum estimates. For instance, they write "Our results show that African Loranthaceae are the younger members of the family, having originated after 28.47 Ma (Fig. 3), i.e. the Chattian of the Oligocene" — no, they don't. They show that the African Loranthaceae diverged latest in the Oligocene (mean 28.47 [median would be better]; 23.95–33.54; Liu et al.'s Table 2). See also our 2017 paper.

Our 2017 dated Loranthaceae tree ...

... and Liu et al.'s (2018) dated Loranthaceae tree (inlet their lineage-through-time plot). Make your pick (it doesn't really matter, does it?)

We hence wrote before discussing our (very similar regarding the estimates) results:

"Due to the data-related limitations regarding both the molecular data and the fossil record, our dating analysis set-up can only provide absolute minimum estimates for divergence ages in the Loranthaceae."

In contrast to many of my fellow palaeobotanists, I have a certain love for molecular dating, but don't pretend it's an exact science. Any estimate is as good as the data behind it, and in the case of Loranthaceae those data needs a lot more work.

This is especially true for the African loranths and the Lorantheae in general. Being a subtropical-tropical group living on other trees, Loranthaceae probably didn't jump into Africa when subtropical and tropical forests were on the retreat, but before that (during the much warmer Eocene). But it may have been the time of their isolation from their South and East Asian (and potential extinct western Eurasian) relatives. (PS The so-far lack of Loranthaceae pollen for Africa is likely a sampling artefact; when using high-resolution SEM imaging, they can be found also in African pollen assemblages; Grímsson, Xafis et al. 2018 — open access; animal-pollinated plants can be tricky to find, even when they have very unique and easy to identify pollen grains; and the tropics are generally not optimal for finding plant fossils.)

Mapping into the past – a litmus test for biogeographic inferences

Let's assume Liu et al.'s biogeographic inference depicts the truth (technically, it's fairly correct) and map it on the past Earth. Since this is not a scientific paper but a post showing why we should do such things, I'll just use the maps of Grímsson, Kapli et al. (© Ron Blakey) with the fossil record shown, for a 1:1 comparison.

By the Paleocene, Nuytsia was already isolated according to Liu et al., but the fossil records doesn't go that deep, so we skip the Paleocene.

Modified from Grímsson, Kapli et al. (2017), fig. 10.

If we take Liu et al.'s (matching ours) estimates as minima, we see quite a fit with the actual fossil record (determined: coloured stars, and undetermined: white stars). The main shortcoming of the explicit biogeographic reconstruction as performed here is that the modern distribution patterns (of the survivors) are likely not comprehensive for their lineages. Liu et al. silently accepted this for the Elytrantheae (by using our European fossil as age constraint for the Elytrantheae MRCA), but decided to not discuss this and to ignore our other Northern Hemispheric pollen of the Psittacantheae/Psittacanthinae and the likely first representatives of the Australasian-East Asian (at the time) Lorantheae: "The major extant lineages of Loranthaceae differentiated in Australasia and South America between 30 and 44 Ma, but during this time the ancestor of Loranthaceae that occurred in North America, Greenland and Europe became extinct". Case closed.

It's southern action only, why again can we use two northern hemispheric fossils as the only ingroup constraints? And what again makes you think that something can jump from subtropical-temperate Australia into tropical Asia, but doesn't go all the way along the closing Tethys into Europe (like all the other tropical stuff)? This is why you publish something like this in a journal titled: Molecular Phylogenetics & Evolution, where palaeobotanical reviewers are rare, and easily outruled by the phylogenetic experts judging your work (well, we don't know if it just a rumour or true, because the peer review process is confidential).

Liu et al. discuss Antarctica as bridge, and especially the Nothofagaceae, as the carriers for dispersal into South America. But Nothofagaceae are a subtropical-temperate group (distinctly the latter in South America). Antarctica was mild but not tropical during the Eocene climate optimum. So, the huckpack-loranths went into the temperate zone to colonise the neotropics. And why shouldn't they do the same on the Fagaceae, according to Liu et al. another important host? Fagaceae were already there in the northern hemisphere, and nearly everywhere. They are at least 80 Ma old, and especially the tropical-subtropical members have a vast fossil record from the end of the Cretaceous into the Paleocene (we published a nice, open access paper in this context in 2016). Which would explain, why one finds so many unique southern hemispheric pollen types in the Eocene of Central Europe. It was hot subtropical to tropical, and full of hosts (see supplements to Grímsson, Kapli et al. 2017, Grímsson, Grimm & Zetter 2018).

In cools down in the Neogene, and so does the amount of known Loranthaceae pollen, but Australia is moving steadily towards nicely warm latitudes.

Modified from Grímsson, Kapli et al. 2017, fig. 11. PS A series of very detailed palaeotopographic world maps can be found in Christopher Scotese's ResearchGate project entitled Earth History: The Evolution of Earth Systems.

Another odd thing is that the reconstruction assumes rather dramatic cross-ocean dispersal events in the past (repeatedly between a still very southern Australia, which for some reason was always a diversity hotspot for the group, and distant Asia), which stopped in the Neogene. Although the South Americans went into North America and possible Europe, they never managed to get to Africa or expand into (East) Asia (in contrast to some of their hosts). No cross-oceanic actual long-distance dispersal from India west-wards. Once established in Africa, this sublineage of the core Lorantheae never went back, not even during the mid-Miocene climatic optimum. Neither did other Eurasian lineages follow their example. Which is quite a difference to other subtropical-tropical groups with high dispersal capacity, where we can find genetically (near)identical species in eastern Africa and southern Asia; and the birds mentioned so prominently in the title (you have to wait till the last section of the discussion to see why, p. 209; I couldn't help laughing out loud when reading there: "Additionally, our dating results indicated that aerial parasitism arose ca. 48–50 Ma" – something we showed already in our dating paper, but were cautioned against jumping to quick conclusions; you say potato...)

Did the Loranthaceae lost their drive? Maybe. Alternatively, what we see today is mainly old established areas (Eocene and older, and possibly including Africa), not so different from the old theories Liu et al. discard without providing any counter-evidence. Not classic Gondwana Breakup, but something that roots in a global distribution (and link) of Loranthaceae by the end of the Mesozoic, dawn of the Cainozoic; a time when angiosperm-dominated forests were on the steep rise. With the World recovering from the Big Chill (the "K-T-Event"), this basis allowed them to rapidly diversify during the Paleocene-Eocene climate optimum. Globally. Or to take the opportunity to move into the tropics. Keep in mind: minimum ages. And given their particular biology as epiphytic parasites depending on animals to disperse, I tend to think that when we find a pollen the according lineage is already there with full force and maybe even for a bit of time (personally, I still fancy the original scenario put up by Vidal-Russell and Nickrent in 2007 with slight modifications).

In this context a coarse but very interesting palaeo-Köppen map produced in course of a French study (Chaboureau et al. 2014; but not published in the paper), that we also used for a supplementary figure to Grímsson, Grimm et al. 2018.

A palaeo-Köppen map for the Earth 70 million years ago, latest Cretaceous before the Big Chill (provided by A. Franc, generated in course of the research to Chaboureau et al. 2014). This is the time when Loranthaceae supposedly originated in Austral(as)ia (constrained in Liu et al. 2018 and Vidal-Russell & Nickrent 2008b using Cranwellia). Tropical = Köppen's A-climates, Arid = B-climates (dry climates), Temperate = C-climates (warm temperate climates), Cold = D-climates (snow or boreal climates), Polar = E-climates. See also this post explaining the (in)difference between subtropical and (warm) temperate.

Alain Franc, a co-author who introduced me to these maps (unfortunately still unpublished in this form; their reviewers preferred them showing the biomes not the Köppen climate zones in the paper, which I find a bad decision), told me the reconstructions tend to be too cold towards higher latitudes and the farther one goes back in time. But it's safe to say: If the first, late Cretaceous loranth was Australian, it was not a tropical rainforest species. And when they became tropical (late Paleocene, as we wrote: estimates should be treated as absolute minima) it would have been easier to step from South America into Africa than from (still) warm temperate Australia via putative (tiny) islands into tropical Asia. But once they established in tropical Asia, there would have been no barrier to migrate into Europe (and Africa). Like an Elytrantheae in the Eocene of Europe. Or stem Lorantheae. While the Psittacantheae came from the west the one or other way, depending whether it was tropical-subtropical (cross-oceanic) or subtropical-temperate (via North Atlantic Land Bridge) lineages. A lot is possible and could explain the modern-day molecular differentiation patterns.

Where to go from here

Biogeographic histories, like any historical research, greatly benefit from historical evidence (something leading biogeographers of the last century strongly objected against). But there are too many white stars in our maps. When we really want to understand how the Loranthaceae lineages radiated and migrated in the past and how this correlates to the expanding and shrinking tropical and subtropical forest biomes (their main home), one first needs to analyse also the southern hemispheric pollen record using high-resolution microscopy. Which pollen type do the oldest finds in Australia and Tasmania show? Do we see a diversity matching our dated trees (which predict several Loranthaceae, root and aerial parasitic lineages being present at the time)? There are quite a bit Gothanipollis in literature, but at the moment we just can say that they are Loranthaceae. Studied using SEM, we may be able to say much more (Grímsson, Kapli et al. 2017; Grímsson, Grimm & Zetter 2018, and palynological literature cited therein). In the best case, this will provide one with enough fossils to do a family-wide fossilised birth-death dating (which is a great thing for plants, too, when you have enough fossils) and (if one feels, it's necessary) actual ancestral areas that could be included in an explicit biogeographic inference.

With respect to the imbalanced branch-length patterns and often long terminal branches, one will need African fossils (one has already been found; Grímsson, Xafis et al. 2018), and, in general, fossils closer to the tip of the Loranthaceae tree. I would make any bet that the currently Oligocene stem age for the African clade will be the first to tumble. The results for the Loranthus/Loranthinae lineage shows that it currently doesn't add up for the leaves of the Loranthaceae tree.

Molecular data-wise, one should reduce the number of taxa and try to increase the number of conservative (ideally nuclear) gene regions.
Conservative genes because it is a relatively fast evolving group, and it makes little sense to include outgroups with gene regions extremely different from those used for the ingroup (personally, I never saw the point in adding an outgroup at all, when the aim is to study ingroup diversification).
Ideally nuclear, because the plastid matK already has an idea about the maternal lineage, but the signal in the combined 18S and 25S rDNA data, biparentally inherited gene regions from the same functional, multi-copy unit, is far from being conlusive.
Adding leaves (OTUs) very distant from the nodes one can constrain using fossils, is counterproductive as well. What's the point including 20/30 OTUs with partly substantially different terminal branch-lengths and inflicting not few ambiguous signals for a geographically homogenous clade when the divergences, we are looking at happened 40 or more million years ago? One should make sure not to add data that just increases the terminal noise. Because one thing is likely in such an active group of plants: the closer we get to the leaves, the more we have to deal with non-dichotomous evolutionary processes, hence, reticulate, tree-unlike signals.

A sort-of-tanglegram we showed in our 2017 paper to make the reader aware of the molecular data situation. Note that although there are no high-supported conflicts regarding basic relationships, it becomes quite chaotic towards the leaves. Pity, that there are no common standards on data documentation and presentation. I'd have loved to see the corresponding graph based on Liu et al.'s better data set (same gene regions).

A likely profitable palaeopalyno+genetic project would be an in-detail study of the Psittacanthinae, the main clade in the Americas. They have very diverse pollen; different types can be found even within genera Feuer & Kuijt (1979, 1980, 1985). Knowing when and where they pop-up (we found unambigous Tripodanthus-pollen in North America and Greenland, but it may be the primitive type of the entire lineage) will eventually allow going beyond minimum age node dating (allow a fossilised-birth-death dating; Heath, Huelsenbeck & Stadler 2014) and possibly infer a detailed history of the Loranthaceae in the New World. Revisiting the American Loranthaceae pollen record (and seeking out more of them) will be crucial. But it only makes sense when this subtribe gets finally some substantial attention by molecular phylogeneticists, and across its entire geographic range. In Liu et al.'s tree (and its very similar ancestors), the most pollen-diverse genus, Psittacanthus, is represented by a single accession; Passovia by a species with unusual (for the genus), likely derived pollen (and a species representing another genus, Phthirusa, according Kuijt 2011, which makes pollen-wise sense), and two genera with quite interesting and partly overlapping pollen morphologies, Cladocolea and Struthanthus, with one and two representatives. A reason maybe that if more speces are included to infer molecular phylogenetic relationships within this so far well-resolved group, some of genera may desintegrate/need to be modified. But maybe molecular lineages agree better with the pollen morphology (studied nearly 40 years ago...) than current taxonomy. And pollen is the only feasible ticket into the past. Such a study could be a blue-print for cross-disciplinary research: studying genes and pollen of the same extant material, to facilitate the latter's use in dating and biogeographic studies when popping up in the fossil record.

Last but by far not least, one should give palaeo-Köppen (my preference)/ palaeo-biome (PNAS peers' preference) maps a try. Characterising the modern Loranthaceae lineages (two examples: Grímsson, Potts et al. 2018 — pyrite open access; Rubel et al. 2018 — gold open access), and then see where the reconstructions show the corresponding climate niches/ biomes at (before when using node dating) the times when the various main lineages diverged (evolved).

It strikes me a group to boldly go where no-one has gone before.

And worth the effort because judging from Liu et al. (2018) and the papers before in the last decade, the nigh-expert (American professor emeritus) currently controlling this research is not ready to evolve as much as his study objects (just check out the cited papers). And even though he may keep his group's matrices away from the public, the data will eventually be uploaded to gene banks.

Cited literature
Chaboureau A-C, Sepulchre P, Donnadieu Y, Franc A. 2014. Tectonic-driven climate change and the diversification of angiosperms. Proceedings of the National Academy of Sciences 111:14066–14070.
Denk T, Grimm GW, Grímsson F, Zetter R. 2013. Evidence from "Köppen signatures" of fossil plant assemblages for effective heat transport of Gulf Stream to subarctic North Atlantic during Miocene cooling. Biogeosciences 10:7927–7942. Denk T, Velitzelos D, Güner HT, Bouchal JM, Grímsson F, Grimm GW. 2017. Taxonomy and palaeoecology of two widespread western Eurasian Neogene sclerophyllous oak species: Quercus drymeja Unger and Q. mediterranea Unger. Review of Palaeobotany and Palynology 241:98–128.
Denk T, Velitzelos D, Güner HT, Ferrufino-Acosta L. 2015. Smilax (Smilacaceae) from the Miocene of western Eurasia with Caribbean biogeographic affinities. American Journal of Botany 102:423–438.
Feuer SM, Kuijt J. 1979. Pollen evolution in the genus Psittacanthus Mart. Fine structure of mistletoe pollen II. Botaniska Notiser 132:295–309.
Feuer SM, Kuijt J. 1980. Fine structure of mistletoe pollen III. Large-flowered neotropical Loranthaceae and their Australian relatives. Annals of the Missouri Botanical Garden 72:187–212.
Feuer SM, Kuijt J. 1985. Fine structure of mistletoe pollen VI. Small-flowered neotropical Loranthaceae. Annals of the Missouri Botanical Garden 72:187–212.
Grimm GW. 2017. ‘Quick-and-dirty’ re-analysis of the Su et al. (2015) data of Loranthaceae and their sister groups. File S6 to Grímsson et al., Evolution of pollen morphology in Loranthaceae. Grana 57 [Online supplement].
Grímsson F, Grimm GW, Meller B, Bouchal JM, Zetter R. 2016. Combined LM and SEM study of the Middle Miocene (Sarmatian) palynoflora from the Lavanttal Basin, Austria: Part IV. Magnoliophyta 2 – Fagales to Rosales. Grana 55:101–163.
Grímsson F, Grimm GW, Potts AJ, Zetter R, Renner SS. 2018. A Winteraceae pollen tetrad from the early Paleocene of western Greenland, and the fossil record of Winteraceae in Laurasia and Gondwana. Journal of Biogeography 45:567–581. Grímsson F, Grimm GW, Zetter R. 2018. Evolution of pollen morphology in Loranthaceae. Grana 57:16–116.
Grímsson F, Kapli P, Hofmann C-C, Zetter R, Grimm GW. 2017. Eocene Loranthaceae pollen pushes back divergence ages for major splits in the family. PeerJ 5:e3373 [e-pub].
Grímsson F, Xafis A, Neumann FH, Scott L, Bamford MK, Zetter R. 2018. The first Loranthaceae fossils from Africa. Grana 57:249–259.
Heath TA, Huelsenbeck JP, Stadler T. 2014. The fossilized birth–death process for coherent calibration of divergence-time estimates. Proceedings of the National Academy of Sciences 111:E2957–E2966.
Kuijt J. 2011. Pulling the skeleton out of the closet: resurrection of Phthirusa sensu Martius and consequent revival of Passovia (Loranthaceae). Plant Diversity and Evolution 129:159–211.
Liu B, Le CT, Barrett RL, Nickrent DL, Chen Z, Lu L, Vidal-Russel R. 2018. Historical biogeography of Loranthaceae (Santalales): Diversification agrees with emergence of tropical forests and radiation of songbirds. Molecular Phylogenetics and Evolution 124:199–212.
Nickrent DL, Malécot V, Vidal-Russell R, Der JP. 2010. A revised classification of Santalales. Taxon 9:538–558.
Renner SS. 2004. Plant dispersal across the tropical Atlantic by wind and sea currents. International Journal of Plant Sciences 165:S23–S33.
Su H-J, Hu J-M, Anderson FE, Der JP, Nickrent DL. 2015. Phylogenetic relationships of Santalales with insights into the origins of holoparasitic Balanophoraceae. Taxon 64:491–506.
Vidal-Russell R, Nickrent DL. 2007. The biogeographic history of Loranthaceae. Darwiniana 45:52–54.
Vidal-Russell R, Nickrent DL. 2008a. Evolutionary relationships in the showy mistletoe family (Loranthaceae). American Journal of Botany 95:1015–1029.
Vidal-Russell R, Nickrent DL. 2008b. The first mistletoes: Origins of aerial parasitism in Santalales. Molecular Phylogenetics and Evolution 47:523–537.
Wilson CA, Calvin CL. 2006. An origin of aerial branch parasitism in the mistletoe family, Loranthaceae. American Journal of Botany 93:787–796.