Monday, August 22, 2016

Melbourne impressions

I spent the weekend in Melbourne on family matters and then today Monday at the Royal Botanic Gardens Herbarium on work matters.


Artwork in our hotel room: A painting of roses that I accidentally put into the washing machine, by anonymous.


Even coming in from the airport we were already wondering what all those masses of yellow, nodding, tubular flowers along the roadsides were. Surely they wouldn't have that many primroses everywhere? Turns out they were rather large-flowered Oxalis. Not sure about the species though, did not study the relevant characters closely enough.


Jellyfish at the Melbourne Aquarium.


Although I have been in the city and even the botanic gardens before this was the first time I actually visited the herbarium. The building consists of two parts, a box that was constructed in the first half of the 20th century, and a round extension that was added in the 1980s. The wall on the left is apparently what used to be one of the facades of the old building, now inside the extension on the right. The latter has two levels of herbarium specimens and a third level housing the library.


Finally, a statue honouring the leaders of Melbourne's ambitious but ultimately tragic 1860-61 expedition attempting to cross the whole continent from south to north, Burke and Willis.

Sunday, August 21, 2016

Monophyletic species yet again: a recent example

Recently I received a publication alert for Ja soon to be published manuscript. It gave me reason once more to write about the issue of "monophyletic" species.

I do not want to give the impression I am deliberately picking on this particular paper. On the one hand, for all I know its data are completely awesome and its conclusions are valid; the issue I am writing about here is somewhat tangential to the paper anyway, its main focus being on genus level phylogeny. On the other hand, this same issue can be seen in many, many other papers in the field, as all too many systematics lectures at universities seem to leave it at "stuff must be monophyletic", without explaining the relevant background like the various relationships that OTUs can have to each other and, crucially, that different classification approaches apply to different relationships. So the occasion here is really only that the present paper showcases the issue in an extremely compact format, all condensed down to a mere three sentences in the discussion.

In full they run as follows:
However, while cladists debate whether higher level taxonomic groups should be monophyletic (e.g. Horandl and Stuessy, 2010; Schmidt-Lebuhn, 2012), it is conceivable that species need not be monophyletic, as different modes of speciation may have different phylogenetic outcomes (Rieseberg and Brouillet, 1994); non-monophyly is an expected intermediate state as taxa diverge (Avise and Ball, 1990). Indeed, in a morphology - based survey of 206 Australian plant species and subspecies (Proteaceae and Fabaceae), Crisp and Chandler (1996) estimated that 21% were paraphyletic. In addition, eucalypt taxonomists generally follow the ecological species concept that allows for hybridisation between taxa (Johnson, 1976), and such reticulation can cause non-monophyly and incongruence between morphological and genetic markers (e.g. Rutherford et al., 2016).
So what do I find problematic about these three sentences? Going through again in order...
However, while cladists debate whether higher level taxonomic groups should be monophyletic (e.g. Horandl and Stuessy, 2010; Schmidt-Lebuhn, 2012),
First, Hoerandl and Stuessy are not cladists. Second, of course cladists do not debate if supraspecific taxa should be monophyletic, because a cladist is defined as somebody who has already decided that supraspecific taxa should be monophyletic. If you still debate it you are by definition Not A Cladist. Compare "vegetarians debate whether they should stop eating meat".  If they are still pondering that question they are Not Vegetarians.
it is conceivable that species need not be monophyletic,
This is where we get to the real issue: monophyly of species. Admittedly we first have to ask, are we talking about sexually reproducing species here? But I assume we are, because the article is about eucalypts, and they are mostly sexual. And the thing is, if that is the case then the concept of monophyly just simply does not apply. It is a word that describes a group of terminals on a rooted tree graph, like so:



A good analogy to describe what is going on is this. Imagine you have a real tree in front of you, and you are taking a group of twigs off using secateurs. To get a monophyletic group, you need to cut exactly once. To get a non-monophyletic group, you need to cut more than once. (If after cutting off a non-monophyletic group you have kept only one piece in your hand, it is paraphyletic; if you have kept several pieces but thrown out what used to connect them, it is polyphyletic. But that just as an aside.)

Within a sexually reproducing species, AKA a breeding group, there is no tree structure but a network structure, as individuals have numerous ancestors in each generation as opposed to one. That looks like this:


How do you get a monophyletic group? Well, you cannot, it is impossible. You could argue that my secateurs analogy would work for paraphyly even in a network, but only by jumping over the "imagine you have a real tree in front of you" part. You don't have a tree, you have a fishnet.

So to me this fragment - and everything that follows - makes as much sense as "it is conceivable that songs need not be yellow with purple stripes". Of course they don't - Amy Winehouse's Rehab, for example, is not yellow with purple stripes and yet it is a perfectly acceptable song. But then again I have no idea how it could be yellow with purple stripes, even if one were to try and make it so.
as different modes of speciation may have different phylogenetic outcomes (Rieseberg and Brouillet, 1994); non-monophyly is an expected intermediate state as taxa diverge (Avise and Ball, 1990).
Opponents of Phylogenetic Systematics regularly make the argument that "non-monophyly is an expected intermediate state as taxa diverge" at all taxonomic levels. At the supraspecific level, this constitutes a rather clear example of circular reasoning. A cladist would argue that a subclade cannot diverge from the larger clade it is part of, ever, because it is by definition part of that clade. That is what the sub- part of subclade means.

At the species level, on the other hand, the above fragment makes sense if we think of incomplete lineage sorting. Barring recombination, the copies or alleles of an individual gene do indeed evolve in a tree-like fashion, and the alleles found in one species will at first generally be paraphyletic to the copies found in its sister species. Only over time will selection or even loss through purely stochastic processes (genetic drift) make the alleles from each species monophyletic on the gene tree, a process known as lineage sorting.

It is possible that this is what the authors are referring to. But to me it still does not mean that it makes sense to call a species paraphyletic, because the components of a species are not gene copies but individuals, and individuals of the same sexually reproducing species stand in a network-relationship to each other, so that the word paraphyletic does not apply.
Indeed, in a morphology - based survey of 206 Australian plant species and subspecies (Proteaceae and Fabaceae), Crisp and Chandler (1996) estimated that 21% were paraphyletic.
Although I would not use the terms as they did (see above), the conclusions of the Crisp & Chandler paper are completely in accord with what I am saying here: species are special, because they are the level at which and from which on downwards it does not make sense any more to try and make stuff monophyletic. That being said, however, the paper does show species as paraphyletic on several morphology-based trees. How did it arrive at that result? Or in other words, given what I wrote earlier, what is the difference in perspective?

First, the terminals on the trees in Crisp & Chandler, the OTUs, are not actually individuals but groups of individuals, such as populations or subspecies; second, the authors conducted phylogenetic analyses on these OTUs. What that means is that the OTUs are forced into a tree-relationship even if the true relationship is net-like, because that is what phylogenetic analyses do. But if we are really talking about structures within a breeding group, within a sexually reproducing species, then in my eyes that analysis was just not appropriate because yes, the true relationship is net-like instead of tree-like. (And if the OTUs are not in a net-like, reticulating relationship, but instead genetically isolated, separate evolutionary linages, then why aren't they recognised as species?)

For example, I can jot down some morphological traits of a bunch of fellow humans, make a data matrix, and do a phylogenetic analysis. Because I do a phylogenetic analysis, the analysis will invariably return a tree. But does that mean that each of my OTUs - individual humans - had only a single parent, and only a single grandparent? Of course not, because we humans do not have a branching, tree-like relationship to each other either. The analysis simply made assumptions that do not hold up against reality.
In addition, eucalypt taxonomists generally follow the ecological species concept that allows for hybridisation between taxa (Johnson, 1976), and such reticulation can cause non-monophyly
Unfortunately it is left unclear what items are forming a non-monophyletic group in those situations. If it is alleles, see a few paragraphs further up; if it is individuals, see the immediately preceding section.
and incongruence between morphological and genetic markers (e.g. Rutherford et al., 2016).
That is true, but we could also mention several other processes, like aforementioned incomplete lineage sorting, meaning that we can have such incongruence even in the complete absence of hybridisation.

To close this post I would like to present a little paragraph that shows how the three sentences discussed above read to me:
However, while opponents of Scottish independence debate whether Scotland should be independent, it is conceivable that citizens need not be independent nations as different ways of acquiring citizenship may have different political outcomes; not being a geographic entity is an expected intermediate state as nations become independent countries. Indeed, in a survey of 206 individual citizens, Doe & Average (2010) estimated that 21% of them were not independent nations. In addition, political scientists usually follow a concept of citizenship that allows dual citizenship, and such reticulation can cause nations not being independent from other nations and incongruence between native language and nationality.
Again, this could be part of a great article on the Scottish independence movement, just like the present paper presents interesting genomic data on its study genus. But does this read as if the author was just a tiny bit confused about the difference between nations and the citizens that nations consist of? Quite so.

A lot of unproductive controversy and confusion among systematists and evolutionary biologists could be avoided if it became a bit more widely known what even ur-cladist Willi Hennig himself had in mind when he came up with the idea of "making stuff monophyletic". He was only arguing that supraspecific taxa should be monophyletic groups of species; the concept of species being monophyletic groups of individuals would not have made any sense to him, as he was very clear on the difference between usually tree-like (phylogenetic) relationships between species and net-like relationships within them.

Reference

Crisp MD, Chandler GT, 1996. Paraphyletic species. Telopea 6(4): 813–844.

Thursday, August 18, 2016

Botany picture #234: Histiopteris incisa


The above fern, Histiopteris incisa (Dennsteadtiaceae) was one of several species we saw on the recent field trip to Kioloa. I am fond of the Denstaedtiaceae, the bracken family, partly because we had bracken in front of our classroom window when I was in lower high school age. But perhaps I also simply find it impressive what large stands these plants can form where they are thriving.

Sunday, August 14, 2016

Greater variance in computer skills than in other skill sets

When I was in what would here be called high school, around the early 1990s, I took informatics as one of my classes. Students in the class had a considerably wider spread of prior knowledge than in any other classes; some people had already dabbled in a programming language, but at the same time others were afraid that their computer would break if they typed in the wrong letter. Perhaps unsurprisingly, the teacher had much more fun interacting with the former group.

When I was an undergraduate, I also took informatics as a minors. The lectures and tutorials had an interesting audience. The percentage of female students was somewhere south of 10%, something I was totally unused to from my biology lectures, where it was more around 60%. More relevant for present purposes, there was again the same massive spread between people who already knew how to program and others who seemed to be touching a computer for the first time.

I particularly remember two fellow biology students who dropped out of informatics after the first few weeks. The whole idea of how you need to make a computer understand what it should do just never clicked for them, down to simple concepts like a variable – when tasked with increasing an integer variable called x by one they would, right to the end, try things like “integer + 1” instead of “x = x + 1”. And while one might argue that the instructors failed these students, who can blame them given that most other students were progressing well?

Given the ubiquity of computers in young people's lives today, I had kind of assumed that these days were gone. The past week, however, I inadvertently put that belief to the test. I gave a practical on using quantitative analyses of morphological data for species delimitation, specifically ordination, hierarchical clustering, and non-hierarchical clustering. My chosen software was R, and although I had prepared a script that the students only had to execute without needing to understand any of the commands, I felt some trepidation.

It starts with the computers themselves. Just a few years ago I had the feeling that very nearly every student owned a laptop suited to such a computer practical. But the past few years have seen a shift in what hardware people are likely to own, so now half of them may only have a tablet; great for reading a PDF or watching YouTube clips, not so great for using a scripting language.

More importantly, I found that the situation with the spread in prior knowledge had not changed in the intervening more than twenty years. About five students raised their hands when I asked if anybody had used R before, something I found seriously impressive for a second year course. At the other end of the spectrum were at least two students who appeared to find it difficult to interpret the meaning of the term working directory / folder.

I want to make it clear that I have no problem whatsoever with the latter. I have no expectation that second year students necessarily have any experience with computers beyond checking eMails and using a search engine. I could happily design a course to give them a good, useful learning experience.

But what I have no idea how to do is how to design a course that gives both them and those who have already used R a good, useful learning experience at the same time. Focus on walking the beginners through every step, and you will find the experienced students getting bored and starting to talk about what to do on the weekend. Focus on engaging the experienced, and you will hopelessly frustrate the beginners, so that they will just give up at some point.

I may be mistaken, but I think there is really no other area where the same thing happens. Yes, students come with somewhat different levels of prior knowledge in every subject, but surely we don't have some of them walk into a molecular genetics practical saying “ah yes, Western Blot, I already did that in summer camp when I was twelve” while others have never even heard of DNA.

So what to do? At the moment I don't have any good ideas.

Thursday, August 11, 2016

The uses of phylogenies

Inspired by a conversation today, I am wondering: What are the uses of phylogenies? I mean, as a phylogeneticist and systematist one would like to understand how species are related full stop. That is the research program. But there are many other uses beyond that, some of them in very different context. Below a potentially incomplete list, which may be updated in the future. It may also have a bit of a botanical bias.

1. Informing biological classification (Phylogenetic Systematics).

2. Understanding evolutionary and biogeographic history of individual lineages.

2.1. Simply describing the phylogenetic relationships between species.

2.2. Understanding when lineage splits or other evolutionary events happened. Example: studying whether the main lineages of mammals already existed before the K/T extinction event or only arose later.

2.3. Inferring diversification rates, in particular shifts in diversification rates correlated with trait shifts or biogeographic events, extinction rates versus speciation rates, etc. Example: higher rates of diversification after colonisation of an empty oceanic island.

2.4. Inferring rates of evolution, in particular shifts in the rate of evolution correlated with trait shifts or biogeographic events. Example: higher speed of evolution in organisms with shorter generation times.

2.5. Inferring ancestral character states. Example: was the ancestor of all eucalypts likely to have been fire-resistant?

2.6. Inferring ancestral ranges and subsequent dispersal and extinction events. Example: on what continents did the perching birds originate, and how did they colonise the rest of the world?

2.7. Studying co-evolution and host shifts. Example: where did the human AIDS virus come from?

2.8. Understanding epidemic dynamics. Example: phylogenetic analysis of virus strains, including historical samples.

3. Studying spatial patterns for biodiversity.

3.1. Inferring hotspots of phylogenetic diversity or related metrics, as opposed to merely counting species, and searching for an explanation for these patterns. Example: a climatically stable area may have accumulated many diverse lineages even as many of them were wiped out in less stable areas.

3.2. Informing conservation management (at least ideally). Example: a hotspot of phylogenetic diversity deserves protection.

3.3. Bioregionalisation, that is the definition and delineation of different biota. This can be done based on nothing but similarity in species content, but can also be done based on phylogenetic similarity metrics, thus taking into account diversity at higher levels.

4. Predicting shared traits from relatedness.

4.1. Bioprospecting. Example: related species are more likely to share similar secondary chemistry.

4.2. Informing plant breeding or gene transfer. Example: finding a wild relative of a crop plant that is very resistant against a disease, and transferring the genes conferring that resistance into the crop.

4.3. Predicting disease or pest susceptibility. Example: wild relatives of the tomato are at greater risk from introduced tomato diseases than totally unrelated species.

Tuesday, August 9, 2016

We have a new toy

This week ALDI in our area is selling little digital microscopes, among other specials. I thought for only sixty dollars it was worth a gamble - wouldn't it be cool to have a microscope for our seven year old daughter? - and I do not regret it.

Obviously we are not talking professional quality here. The picture is a bit blurry compared to a really good microscope, and the images it takes are only 800 x 600 pixels. It also does not have an ocular but runs entirely through a connected computer. But it is still functional to show a child the beauty of nature at smaller scales, and it may even become useful for the odd, quick shot to be integrated into a lecture or presentation, e.g. if I pick something up during a walk on the weekend and don't want to go through the trouble of using our high-end microscopy facility at work for something that trivial.

But for the moment our daughter in particular loves exploring small plant parts or fibres with the microscope.


Above: A moss leaf, the first object I collected to show her. Note the individual cells are clearly visible; what more do you need to demonstrate that organisms are build from cells?

I was a bit worried at first because the microscope comes with software only for Windows and Mac. A staff member at the store told us that "the program will work on Linux, Linux uses the Mac OS anyway, trust me, I am a software engineer", but I knew even then that that was what we might euphemistically call an over-simplification.

But no matter - the software it comes with is unnecessary. It turns out that a random webcam program, in our case the pre-installed Cheese, can run the microscope camera just as well. So far I am not regretting changing over to Ubuntu either.

Saturday, August 6, 2016

Botany picture #233: Prostanthera violacea


Currently at ANU's Kioloa field station with a group of students. I was happy to see the above Prostanthera violacea (Lamiaceae) in flower, as the mint family is still one of my favourite plant groups. I did my PhD on a genus of that family and will thus always have a soft spot for them.