Some thoughts on oral presentations

The Systematics 2017 conference in Adelaide was great, and as always I learned a great deal and enjoyed interacting with colleagues. Also this was simply the first time since I came to Australia that I saw South Australia, and it was the last state that I had not seen so far.

Looking back over the talks (oral presentations) I heard over those three days I wonder, however, about the different aspects that people decide to focus on in those talks. There are five types of talks that I find particularly odd.

The overly introduction focused talk

The average conference talk in science is structured like the average scientific paper: (1) introduction providing background information and leading into the aims of the study, (2) methods, (3) results, and (4) a discussion putting the results into context and explaining what they mean. What I call the introduction focused talk is when the speaker spends so much time telling us how cool their study group is and what had already been known about it decades ago or perhaps what they are trying to achieve that by the ten minute mark I wonder if we will ever get to hear what they have actually found out.

Now this is of course perfectly fine if the speaker is a first year graduate student who has only just started their project, but it is somewhat less understandable if a more senior researcher actually has lots of interesting data but, due to their misplaced sense of priorities, only manages to flash one tantalising results slide for twenty seconds before their talk is cut off by the session chair. In such cases something is off about the balance of the talk, just saying.

Relentless wet lab wonkery

One of the frustrations of the last few years, clearly driven by the rise in high-throughput sequencing, is the increasing amount of lab method wonkery in conference talks. People who could otherwise be great and engaging speakers go through slide after slide with little lines that are meant to be DNA fragments, explaining at length how those fragments are produced, barcoded, amplified, size selected, pooled, and sequenced, often for approaches that have been around for several years.

Maybe I am wrong, but I think most people do no want to hear, at least primarily, what somebody did on the bench, they want to hear for example how the study plants or animals evolved and what that means biologically. The lab wonk talk is like going shopping for a used car and finding that every salesman first walks you through the way a combustion engine works.

Relentless bioinformatics wonkery

This is the same as the previous, only with a focus on how the sequence reads are quality controlled, contigs are built, alignments are made, etc. Except in the context of a methods workshop it is perhaps even less helpful than lab wonkery. Most biologists in the field can at least easily visualise what happens to the DNA fragments the lab wonk is talking about (even if they don't really care and want to see the biologically meaningful results), but the bioinformatics realm is so full of impenetrable jargon that ironically only those who don't need to hear the talk will be able to understand it anyway.

Selling well known facts as great new insights

Another frustration that I have are those speakers who present as their awesome new insight something that every marginally competent audience member has been aware of for years.

High-throughput sequencing gives us more data than Sanger sequencing did? Who knew? So, museum specimens have degraded DNA? You don't say. GBIF exists, and the public can download specimen location data from it? Wow. We should be taking photos of herbarium specimens and putting them into online databases? Quick, somebody invent JSTOR Plants!

I am not saying that these are not points that can be made as part of the introduction to one's topic, to show how far we have come in a very short time. But if the entire point of the talk is something to the effect of "future directions in our field" or "where we need to be in 2028" one does expect something that did not already happen a decade or so ago.

The self-promoting and frustratingly off-topic keynote

Finally, I am starting to notice a certain type of keynote or plenary talk, where a hotshot scientist is invited to provide a broad overview of developments in their field, usually to frame the more focused and detailed talks that will follow after it in the program.

Keynotes and plenaries are always more like review articles than research articles, and I just have to admit that I do not go to conferences primarily to hear them. Nonetheless I have heard great and engaging plenaries, including at this recent conference, and can enjoy some of them even when they are largely about historical developments. It just depends on the choices made by the speaker.

What I really find frustrating are speakers who see these talks largely as opportunities for self-promotion. Their talks show at least several of the following features:

• Nearly everything that is mentioned is the speaker's own work and that of their students, virtually ignoring contributions from others in the field.
  
• Accordingly, many of the images in the slides appear to be photos of their numerous students and postdocs, usually goofing around in the field or looking awkward in front of a computer screen.
• A good part of the remaining images are scans of the tops of the speaker's own articles, with the journal name showing prominently ("look, I have a Nature paper!").
• The content of the talk is at best tangential to the title that it was advertised under, presumably because the speaker is re-using the same slides as for the last four such talks they gave under different titles and on different occasions.

Of course I would not expect everybody to agree with me, but my favourite type of talk is still a standard contributed presentation that is well-balanced between background and results, with methodology largely limited to a few informative bullet points.

IBC and Shenzhen impressions

Seen at my hotel.

Shenzhen Convention and Exhibition Centre.

Exhibition area, with poster area in upper right corner of picture.

One of the larger lecture halls.

After dark.

Poster area.

Tea break in the hall between the smaller lecture rooms.

International Botanical Congress 2017, Shenzhen

A few thoughts after the first two full conference days of the IBC 2017. (Sunday was only the opening and a few public events, and today Wednesday is a bit of a break without symposia).

Shenzhen has really pulled out all the stops to host the conference. It is huge, it has a huge presence in the city, and everything is well organised. The city itself is also a good location. I am not personally a fan of skyscrapers, but of course only a large city like this has the infrastructure for a congress of more than 6,000 delegates. The subway system in particular is amazing.

What I am less convinced of at the moment is the way the scientific program has been organised. In what is apparently custom for IBCs, all the symposia were set in advance, and for each symposium the organisers were required to arrange four set talks and fill the remaining two slots competitively from non-invited submissions.

I find that unfortunate. Why not set perhaps half the symposia, call for free submissions, and then group all the latter thematically into additional symposia? If you get twelve talk submissions for palaeobotany, for example, simply arrange for one of the more accomplished speakers to chair the whole thing. I have not yet heard an argument against such an approach beyond "but that means that they would have less time to organise the symposia", but other meetings do not seem to have any problems with it, from what I hear even very large ones.

I have also heard from many others, and observe myself, that there are too many parallel symposia, and often symposia that would clearly attract the same clientèle are placed in parallel to each other. One of the reasons this happens to such a high degree is that only a small part of the day - four hours in the afternoon and, again, even excluding today - is reserved for the general symposia, while the entire morning is given to keynotes and plenaries.

Now I may be in the minority here, but I am not to be counted among those who primarily go to a conference to hear a famous person give a keynote lecture. At their best they are the oral presentation equivalent of review papers, but I see the benefit of meetings mostly in networking and learning about new results and methods, the oral presentation equivalent of research papers. And for this the general symposia are the places to be. I am consequently wondering if it would not be more productive to increase current symposium time slots by 50% and to cut the number of keynotes and plenaries in half. This would also mean that less symposia would have to run in parallel.

Overall I have seen some very amazing talks, but also some that appeared a bit uninspiring and could have benefited from a more explicit explanation of what the research question even was and why the audience should care. I have heard of software that I should look into and learned about tools that have become more powerful since I last tried them, but also groaned about "paraphyletic individuals" and the assumption that the traits of the smaller of two clades automatically represent the ancestral states. And of course I have met colleagues from overseas who in many cases I haven't seen since the last IBC in 2011.

Now I just have got to get my own talk down from a rambling 25 min to 15-20 and all will be good...

Weird or bad talks I have seen

Two days ago I saw seven great scientific talks presented by students. However, I prefer not to blog about things involving personal names except when discussing a published paper or a public conference; in this case the students had to give the talks, and the symposium was somewhat internal.

At any rate, when I hear a good talk I tend to remember the science. That is just the point, isn't it? A good talk gets the message across. On the other hand, what stays with me from a bad talk is not the message but rather the way in which it grated.

With scientific talks on mind, I consequently thought I would write about some of the really weird or bad talks I have seen over the years, without giving any identifying information. None of these talks, mind you, were presented by students; all speakers were at least postdocs and often professors.

Mostly this is about what one might call slide design choices rather than content or structure, and I am not saying that all of them are horrible. Some were just strange in an amusing way. Again, whatever has really stuck in memory.

1. A classic that I have complained about in the past was one where the speaker had a black slide background with the text and even the figures (!) in various bright colours. Headache inducing after the first few slides. I cannot really recommend this approach.

2. Not quite as bad but not really helpful either was the choice by another speaker years later to show their crucial mathematical formulas in light colours on a light grey background. Although it is generally considered not the done thing to openly criticise a speaker in this way, this design choice actually caused comment even during the talk.

3. Rather distressingly popular in the first few years after PowerPoint became available but now less often seen is the overuse and abuse of its animation functions. I once saw a talk where the speaker used a different way of making text appear for every. single. bullet. point. One floated in from the right, one spiraled in from the left, one was assembled by each letter dropping in individually from the top, and so on. The talk was not helped either by the speaker's lack of confidence in their English. They interrupted themselves after every few words with a kind of loudly barked "uh-huh?" towards the increasingly concerned audience, apparently asking for constant feedback whether everybody had understood the preceding sentence.

4. One of the weirdest talks I have ever seen consisted entirely of numerous diagrams hand-written onto math paper that were subsequently scanned and pasted into PowerPoint.

5. The audience of an international conference I participated in was fairly amused by a talk whose slides consisted of 80% small font walls of text, with a far too tiny figure pressed onto the side here or there. The real kicker was that the speaker had asked a colleague to translate the slides from English into the local language, and the translator had left numerous comments in the text on the lines of "I don't know this term", "is this what you meant?", or "you need to look at this again". Apparently the speaker had not looked at the presentation at all after getting it back from the translator.

Australasian Systematic Botany Conference 2015, third day

On the third day I unfortunately had to miss several talks. The main topic of the day was Integrated Floras, Electronic Floras, and Online Keys.

In the morning session chaired by Zoe Knapp, Ilse Breitwieser's keynote lecture gave an overview over the various types of floras or flora-like publications, starting with her first use of the Schmeil-Fitschen field flora of Germany as a student. She explored the definition of a flora, of an integrated flora, and of an electronic flora - not just a digitised book but a flora curated and updated in an electronic data management system. She then covered a variety of 'learnings' from the eFlora of New Zealand including the problem of managing authorship rights and copyrights in an increasingly collaborative, dynamic and openly accessible workspace.

Australasian Systematic Botany Society Conference 2015, second day

Writing about the second day of the ASBS Conference 2015 I should first mention that I should now really be considered biased because I chaired the morning session and gave my own talk in the afternoon.

The day started with the session on Genomic Data in Plant Systematics. In his keynote Craig Moritz, the director of the ANU/CSIRO Centre for Biodiversity Analysis, set the scene by providing an overview of next generation molecular and analytic methodologies and their use cases. He also discussed the new possibilities arising in what he called a post-phylogenetic world - not in the sense of 'evolutionary' systematists of course, but in the sense of having achieved more or less complete taxonomic description and the availability of decent backbone phylogenies for at least some taxonomic groups. In particular, he mentioned the integration of population genetics and phylogenetics, the study of speciation mechanisms, and cryptic diversity.

Next, Todd McLay presented results form his research on species delimitation and phylogenetics of the grass tree genus Xanthorrhoea. I was particularly intrigued by a new PCR based reduced representation library preparation method that he developed, because it seemed to be comparatively simple and thus perhaps feasible even in shorter student projects. Benjamin Anderson then gave a very concise talk on this work in the arid zone grass Triodia.

Australasian Systematic Botany Society Conference 2015, day one

This week our herbarium is hosting the annual Australasian Systematic Botany Society Conference. For me this is the first time to be involved in conference organisation, specifically the scientific program (deciding on conference theme and major sessions, inviting keynote speakers, organising workshops, deciding on abstract acceptance, grouping talks into sessions, etc.).

Yesterday started with a long session on collections science chaired by Sarah Mathews. Keynote speaker Vicki Funk of the Smithsonian stressed the importance of natural history collections in the genomic age and identified three game changers for the near future of collections science: open access to data and images, improved sequencing of degraded DNA from old specimens, and linking collection specimens with phylogenies and climate data. I was also amused by an anecdote she told of discussing phylogenetic systematics with an older, very influential traditional taxonomist who argued that not being allowed to recognise taxa based on plesiomorphies amounted to "throwing half my data away". Vicki then tried to explain that the data are used, only deeper in the tree, where those character states were apomorphies themselves. The amusing aspect is of course that 'evolutionary' taxonomists still haven't grasped that a generation later.

The collection science session continued with a talk by Jill Brown on crowd sourcing of data entry at the herbarium of the University of Melbourne. They are involving volunteers through the ALA DigiVol portal, and the project appears to be quite successful. Liqin Wu then presented an analysis of environmental lead accumulation in historical lichen specimens, an interesting use case of collections. She also identified possible lead sources (soil/rock, leaded petrol, coal burning) based on isotope ratios.

Conferences, pottery and botany picture #219: Actinobole condensatum

Yesterday morning I had the pleasure of presenting a talk on the daisy family Asteraceae to the Biennial Conference of the Australian Native Plant Society Australia* (ANPSA). It was the first time I have been at a conference that wasn't either political or scientific, and I found it interesting how it compares to our scientific society meetings such as the upcoming annual conference of the Australasian Systematic Botany Society (ASBS), which will also take place here in Canberra.

At the ASBS, we will have around 50 speakers over three days, nearly all of them out of their own initiative, and nearly all of them will have only 15 min. At the ANPSA, they have 21 speakers over five days, many talks appear to be invited, and speakers are alloted 45-60 min.

At the ASBS, we will have one field trip on the day after the conference. At the ANPSA, they have several field trips (excursions) every day throughout the week, although admittedly they do not go as far away.

Clearly priorities are slightly different - native plant enthusiasts can hardly be expected to get together every two years and not prioritise looking at native plants.

Anyway, at least one of the members must be into ceramics and pottery. Each invited speaker was given a little customised vase; at least two of us (including me) did not even remember that the society contact asked us months ago what our favourite plant was. It is hard to pick just one, but I now dimly recall that I nominated Actinobole condensatum, because this is what the artist, known to me only as Franzi from her signature on the vase, did:

And here, for comparison, a picture of Actinobole condensatum I took during a field trip to Western Australia in 2012:

It is a great impression and abstraction, getting across a good likeness of the species. And the species is really a super-cute desert ephemeral. The plants are very tiny and short-lived, but they produce a dense, attractive cluster of small but typical paper daisy heads. The red sand on which they like to grow does its part to make for an aesthetic impression.

Footnote

*) Yes, that extra Australia seems a bit odd at first sight, but I understand that it is meant to differentiate the national society from the local chapters, as in Australian Native Plant Society Canberra Region, and not from a hypothetical Australian Native Plant Society Botswana.

CBA Conference 2015, last day

The final day of the 2015 CBA conference on species delimitation featured another set of very interesting talks.

Sally Potter presented her and colleagues' exome capture pipeline for the study of skinks. Admittedly this was already so specific and applied that one would not expect to walk away with many concrete ideas for one's own work unless one wants to use pretty much the same method, but she also mentioned a species tree method that had so far escaped my attention.

ASTRAL is supposedly very fast for large numbers of loci; it is said to be "improving on MP-EST and the population tree from BUCKy, two statistically consistent leading coalescent-based methods". Statistical consistency sounds nice, but when I played around with them for a bit back when those precursor methods didn't really convince me. Still, ASTRAL may be worth a try at some point in the future.

CBA Conference, day 2

Continuing with the 2015 CBA conference on species delimitation.

Today's first speaker was Sasha Mikheyev, who presented genome sequencing data on hybridisation zones in social insects. Most of his talk focused on Africanised bees in America, where he was lucky enough to find collaborators who had a time series of samples in the freezer capturing exactly the moment when the African genes swamped populations in the southern United States. Important research, but the results were still preliminary, so he could not go into what he ultimately wants to find out about the dynamics of hybridisation.

He ended his talk with a really weird story about an ant species in which queens are clones of their mothers, males are clones of their fathers (how does that work - is there a zygote but the female chromosomes are discarded?), and only the workers are half-half. This means that the males and females of the same "species" are genetically completely isolated, and indeed the male lineage is more closely related to a different species than to the females they have sex with.

You just can't make anything up that is more bizarre than what reality holds in stock for us.

CBA Conference 2015, day 1

Most of this week I am at the 2015 CBA conference Species delimitation in the age of genomics.

Although the title suggests that you'd die of alcohol poisoning in short order if you took a drink every time you heard somebody claim that genomic data are going to solve all our problems, the actual talks happily do not fit that stereotype. Maybe people have seen enough genomic data now to dispense with the hyperbole.

Today started off with a talk by the Australian philosopher of science John Wilkins. He argued that the traditional approach taken by most people who are explicitly grappling with the species problem is to declare a theory-based species concept, try to force it onto reality, and then consider the species-ness to be an explanation for what we see in nature: This individual is the way it is because it is a member of that species. In Wilkins' opinion, that is precisely the wrong way around.

Methods overload

The conference I attended this week was a very methodological one, with lots of modellers in attendance. Looking back over the talks I heard and the workshops that were offered, two things have occurred to me.

First, while everybody appears to assume that we are in the era of big data, especially due to genomic sequencing and the increased availability of  biodiversity databases, it seems more to me that we are instead currently drowning in methods.

Conference today

I am visiting a conference this week, Understanding Biodiversity Dynamics Using Diverse Data Sources. But as it takes place in the city where I work I don't have to go far - the talks are just down the hill from my office. Some of the talks today were inspiring and utterly fascinating.

Two in particular I want to mention very quickly, although for very different reasons. The first was, from my perspective, perhaps the highlight of the day, but more importantly it was highly relevant to my recent discussion of the merits of phylogenetic diversity. Among other things, Professor Marc Cadotte talked about the claim, already advanced by Charles Darwin, that, all else being equal, a biome containing more phylogenetic diversity (PD) is more productive than a biome with less.

And he presented an ingeniously simple field experiment to test this claim quantitatively: They planted numerous experimental plots with one, two or four species of varying evolutionary relatedness, taking care to replicate close relatedness in various groups (two grasses, two Lamiaceae, and so on). The result was a beautiful and clear positive correlation of grassland productivity with PD.

As I wrote above, this is highly relevant to the question of whether PD is correlated with feature diversity, only in this case there is another mental step involved. The underlying idea is that a community of four distantly related species would be more productive than one of four closely related ones because it would have higher feature diversity in the sense that the species are more divergent and can thus utilise more different resources than the same number of supposedly more similar, closely related species.

The positive result of the experiment can thus be read as another vindication of PD, and surely the unseen characters that allow these plants to complement each other in the utilisation of local resources are "conservation relevant", to use the words of the PD critics I discussed a few days ago.

The other talk that I briefly want to mention was sadly somewhat less sparkly. The thing is, there are many snobbish molecular and evolutionary biologists who consider natural history collections and the research they do as mere "stamp collecting" and a tedious description of patters instead of something that increases our understanding of relevant processes.

I will admit there are some colleagues who fit that description, who would go and bore the socks off an audience by showing them half an hour worth of insect mouthparts or leaf shapes without betraying any underlying research question.

But bizarrely, at the moment I have examples showing the exact opposite. In my backpack is a paper I am peer reviewing where the authors examined a highly topical but supposedly hard to test evolutionary question through the simple expedient of measuring a few macroscopic characters on dusty old herbarium specimens and doing some simple statistics. Conversely, today at the conference I had to sit through a 30 minute talk by a scientist working with cutting-edge genomics and bioinformatics techniques who filled virtually the entire time with one "and this transcription factor has this effect on the insect wing" after the other. It was purely descriptive, and I never understood why anybody should care to know this stuff.

That just goes to show that it is not what tools you have but what you do with them.

Back from conference

I spent Tuesday and Wednesday at Systematics without Borders in Sydney, a joint conference of the Australasian Systematic Botany Society, Invertebrate Biodiversity & Conservation, and the Society of Australian Systematic Biologists. Although I sadly had to miss the first of its three days I enjoyed the meeting immensely. It is always good to see all the science that is being done in the field and to catch up with people.

A few highlights for me:

Phil Garnock-Jones received the ASBS' Nancy Burbidge medal for his work in plant systematics. From his prize lecture I learned many astonishing facts about plant sexuality, especially about that of mosses. When we talked a bit in one of the coffee breaks I also learned that he runs a very interesting botany blog, Theobrominated. Check it out!

Quite a few of the talks most relevant to my work were from New Zealand, actually, such as Rob Smissen presenting work on gene flow between different species of southern beeches and Ilse Breitwieser talking about the difficulties of circumscribing species in their native clade of Craspedia.

Having done my PhD on a Lamiaceae and being somewhat interested in pollination ecology, I also particularly enjoyed hearing about Trevor Wilson's work on the awesome Australian mint bushes (Prostanthera, Westringia and relatives). It turned out that quite a few generic limits will have to be redrawn because the stamen characters used for the traditional definition of groups have evolved several times in parallel. I guess I should post more botany pictures of these attractive plants in the future.

In addition there were many others, for example on biogeography and plant-insect interactions, too many to do them justice.

Just a pity they decided to have the conference in the first week of December, just when our summer students start...

Bush Blitz Symposium, one final time

All the talks from the Bush Blitz Symposium that I blogged about last month have been made available online, including my own. In case you are interested, find them via this finely crafted link.

Perhaps more conferences should think about doing this, although admittedly it would strongly discourage scientists from presenting unpublished results.

The taxonomic impediment and DNA barcoding

Further to yesterday's conference, there was also a talk about DNA barcoding. First, what is that?

Some time ago I did a post on the taxonomic impediment, the problem that the number of taxonomists worldwide is too small (and, due to short-sighted funding policies, shrinking) to deal with the massive amount of biodiversity still to be described and classified, a situation that impedes downstream studies in ecology, evolutionary biology, conservation management etc relying on an accurate description and classification of life.

The idea behind DNA barcoding is to allow an easier determination of species by sequencing a few carefully chosen gene regions for as many species as possible and depositing that information in a searchable database. So if you found, for example, a plant in a rainforest plot and you needed to know what it is, you don't collect a specimen any more and send it to a botanist for determination, but instead you extract DNA, sequence the right gene regions, and then use the database to find out, say, that there is an 98.9% likelihood that the plant you are dealing with is Ruellia brevifolia.

Optimists could say that this is a good solution to take superfluous work off the shoulders of overtaxed taxonomists so that they can get on with describing new species. Cynics might think that the idea is ultimately to de-fund taxonomic research and to replace taxonomists with shiny DNA sequencers.

There are good reasons to believe that barcoding cannot work in principle, at least not at the species level. (Hint: the most important one starts with "i" and ends with "ncomplete lineage sorting".) But the talk I heard yesterday raised a different issue.

You see, the colleague who presented it had earlier been at a different conference dedicated specifically to barcoding, and one of the things she learned there was that the vast majority of barcode reference sequences in Genbank are what is called "level 0", meaning that nobody knows what species they are from because the people who produced them were unable to find a taxonomist who could confidently determine the specimen they got the DNA out of.

Is that not just awesome? Of course barcoding, even discounting other technical problems, can only work if there is a high quality, reliable reference database. Maybe at a minimum one should have waited with starving taxonomic expertise to death until such a high quality database was in place.

Bush Blitz Symposium 2013

All day today I was participating in the Bush Blitz Symposium 2013, a one day meeting here in Canberra at Old Parliament House.

What is Bush Blitz? In the words of the eponymous website:

Bush Blitz is Australia’s largest nature discovery project - a three-year multimillion dollar partnership to document the plants and animals in hundreds of properties across Australia’s National Reserve System. Since the program began in 2010 Bush Blitz has discovered about 600 new and undescribed species and has added thousands of species to what is already known - providing baseline scientific data that will help us protect our biodiversity for generations to come.

Perhaps most importantly, the initiative organizes targeted surveys to document all species of certain focus groups (one of them being vascular plants) in defined areas, often properties that are being transformed into nature reserves. In every case, botanists and zoologists with different specialties are brought together for one or two weeks to inventory the biodiversity of the place, with all the logistics provided by the Bush Blitz team. The scientists collect samples, process them in the base camp, determine them to species and ultimately compose species lists. The Bush Blitz team then writes a report on the survey summarizing the results.

I was lucky enough to participate in two Bush Blitz surveys, one in New South Wales and one in central Tasmania. While I did my part to document the local plant diversity, both field trips also allowed me to obtain samples for my research and to learn more about the Australian flora. One of the most interesting aspects of those surveys is the close interaction with biologists studying other groups of organisms.

In addition to the surveys, the initiative also provides grant money for taxonomic research resulting from the survey work. The symposium today presented results from the activities conducted so far and additional talks on topics relevant to the Bush Blitz project. Topics of the talks were extremely diverse, including teaching and outreach, collaboration with the traditional custodians of the land, hard scientific analyses as well examples of species discovered as new to science, and conservation planning. (I presented a talk on collecting biases - field botanists collect mostly close to where they are and along roads, and they under-sample spiny, very small, ugly and non-native plants. Bush Blitz addresses all these issues by targeting neglected areas and by inventorying all species of a group, not only the conspicuous ones.) The talks were very interesting and the speakers were really enthusiastic about biodiversity. All in all a great meeting.

In summary, the Bush Blitz is a fantastic and globally unique institution; if you want to learn more, you could do worse than reading through their website.

Biodiversity Genomics Conference, last day

Today I participated in a workshop on phylogenomics. It was a somewhat mixed bag; on the one hand, there were a few really useful elements, on the other hand one of the presenters merely repeated, virtually one to one, a talk that he had already given on Wednesday. Ah well.

The conference was very rewarding and clearly a great success. Apparently significantly more people wanted to register than there were places. The participants hailed from various continents and represented many different fields of research - from US American evolutionary biologists over German entomologists and Australian soil researchers to New Zealand conservation biologists. I have heard from many colleagues how much they enjoyed it and how much they got out of it, and even that several people have suggested to have a conference like that every year, especially considering how fast the genomics field is evolving.

I cannot help, however, to end on a somewhat skeptical note considering the advances in genomics and Next Generation Sequencing from the perspective of my research interest in phylogenetics. One of the presentations today drove home the point just how mind-bogglingly, unmanageably and ridiculously huge the amounts of data are that are being produced. Again, the 1KITE project sequences the transcriptomes of 1,000 insect species, that means all genes that the insects had expressed at the moment they were sampled. And then they want to use these data to produce a better phylogeny of the insects. Admittedly, they can probably use the same data to do a couple dozen other things in addition, but from a phylogenetics perspective and considering that many other people are doing the same, does sequencing entire whatever-omes really, if we come right down to it, make any sense?

• Nobody appears to know even where to put all these data. That is true on the level of the individual researcher, who buys a cutting edge external hard drive only to run out of space one project later, but also on the level of the scientific community as a whole. One participant actually asked that question on Wednesday: Genbank accepts annotated traditional Sanger sequences of individual genes and they are already struggling to keep up, but where do I submit a terabyte of genomic DNA sequences to fulfill the requirement that it is publicly available to colleagues who want to be able to reproduce my results? This is getting out of hand pretty quickly.
• Nobody seems to know how to analyze them appropriately (for phylogenetic purposes). A major topic discussed controversially in today's workshop was the data analysis. I don't want to go into details, but for massive amounts of genomic data for numerous samples the only chance currently seems to be to concatenate all of it and to use phylogenetic tools that trade sophistication for insane speed. At the same time the people generating all these data are keenly aware that what they really want and need are complex models of DNA evolution, complicated partitioning of the data, and species tree methods, but the software that could do those analyses falls over if you try to do them with even just 10% of the available data.
• And here is the kicker: For phylogenetic purposes, nobody actually needs that much data. Alan Lemmon was entirely correct when he said that 400-500 independent loci are more than enough for phylogenetic analysis. But even that might already be overkill, as Leaché & Rannala (2011) found that 10-100 loci are generally sufficient even in difficult cases, and as few as 10 in simple ones. In other words, do we really need to expend a lot of effort and money on sequencing entire genomes and produce reams and reams of data if using 0.5% of those data would already give us precisely the same result? Don't get me wrong, this all makes sense if you are interested in exploring signals of adaptation in the genome and suchlike, but at the moment it appears the phylogeneticist is presented with a shiny new nuclear bomb and told that this is a good way to kill the flies in their house. What we need would be a labor- and cost-efficient way of capturing, say, 40-50 loci for a few hundred samples (preferably from low amounts of extracted DNA) instead of laborious ways of producing insane amounts of useless data for a few samples. They same goes for population geneticists, by the way.

Biodiversity Genomics Conference, third day

Again a lot of talks today, but again I could not catch all of them. However, this time I heard the session of so-called "lightning talks", seventeen talks in one and a half hours, each of them about five minutes long and without discussion time. The experience has left me doubtful about the idea of lightning talks. Such a short time is not at all adequate to present scientific results. It may work for putting ideas out there, for presenting projects still at their beginning to ask for feedback or collaborations, or for advertising new methods or software tools as long as they are not too complex. But in those cases a question or discussion time afterwards is all the more important.

Most speakers managed to finish their talks in time but one of them had a ridiculously large number of slides prepared which looked as if they would have been hard to cover in a standard twelve minute talk. The topics varied widely - from identifying prey items by genotyping predator feces or stomach contents to the exploration of the invertebrate fauna of Antarctica - and were mostly interesting despite few of them being directly related to my own research interests.

Biodiversity Genomics Conference, second day

Lots of talks today. After the opening words, Leo Joseph of the Australian National Wildlife Collection gave an inspiring talk about the potential of combining the resources of biological collections (animal skins, needled insects, herbaria, etc.) with Next Generation Sequencing. Specimens in these collection provide not only genomic data that can be extracted from them but also usually spatial data (where collected), temporal data (when collected) and phenotypic data that could be combined with the genotype. He also pointed out that conserved specimens with their degraded DNA are actually better usable with genomic tools than with traditional Sanger Sequencing because the former are designed to read short DNA pieces. Next, Justin Borevitz (Austr. Natl. Univ.) presented a very dense talk on his research in Arabidopsis and Australian Pelargonium, focusing on attempts to correlate genetic variation with phenotype (such as flowering time) and geographic locality to understand local adaptation.

The next session started with Graham Coop (UC Davis) who presented recent progresses in providing null models of genetic variation. Admittedly much of it was too mathematical and went over my head, but I got the general idea. When you want to know whether some genetic data signals local adaptation, the null hypothesis cannot simply be that the genes are distributed randomly. Instead, you have to account for covariance or in other words the possibility that the relevant gene copies are locally frequent due to the shared history of the local populations. For example, are Europeans light-skinned because they just happened to descend from a light-skinned ancestral population or because it is an adaptation to something? The solution is to produce a population by population covariance matrix with lots of genetic data which can be assumed to be mostly neutral, and then to compare the genetic variants of interest against it.

John Novembre (Univ. of Chicago) gave a really clear and well presented talk on four methods of increasing sophistication for spatial assignment of samples. The idea is that you have genetic information for a representative number of samples of a species across its range, e.g. all African elephants. Now customs seize a shipment of ivory from a black market in South East Asia. What analyses could you use to infer the country of origin from genetic information extracted from the seized tusks? The four methods he presented are PCA based, logistic frequency surfaces, Gaussian process based ("SCAT") and a joint analysis of genomic and isotopic data with SCAT. The first are simple but imprecise, SCAT is sophisticated but computer intensive.

Rose Andrew (Univ. of British Columbia) presented her research on incipient speciation in sunflowers. She used genomic data to examine patterns of divergence and what they mean for local selection, gene flow and isolation. The last speaker of the session was Ke Bi (UC Berkeley) who used genomic data from ca. 100 year old and contemporary chipmunk specimens to test the impact of climate change on their genetic structure; a very nice example of how museum specimens can provide the basis for cutting edge research!

In the afternoon session, Sonal Singhal (UC Berkeley) showed the pattern of genetic variability across five contact zones between species that have diverged at different times, with a clear negative correlation of introgression and time since divergence. Alan Lemmon and Emily Moriarty Lemmon (Florida State Univ.) advertised their recently published "anchored hybrid alignment" technique for rapidly producing data from hundreds of loci for ca. 100 samples for phylogenetic analysis. This was probably the talk that was most relevant for my own future research interests but, well, they have only done it for vertebrates and insects so far. Bernhard Misof (Univ. of Bonn) provided an update of the 1KITE project which is close to reaching its goal of sequencing the transcriptomes of 1,000 species of insects. The amounts of data produced in the course of the project are just mind-boggling.

Finally, Tariq Ezaz (Univ. of Canberra) presented some of his work on the evolution of sex chromosomes in lizards. Sex determination is really odd if you think about it. There are at least three systems in land vertebrates: XY, as we humans have it and where the females are XX homozygous, ZW, where the males are ZZ homozygous, and temperature dependent. In other groups of organisms, it gets even weirder, with some animals changing sex as they age, some choosing a sex depending on what the current sex ratio in their population is, and hymenopterans having haploid males. After that session ended, there were fifteen very short "lightning talks", but I had to get a few things done and missed them.

To be honest, I feel pretty overwhelmed by all the lab method wonkishness going on here. I am more interested in the phylogenetic analyses and, well, the actual organisms themselves, and my ideal situation would be to outsource the lab work part. On the other hand, I am probably the only person at the conference who could tell Pogonlepis stricta apart from Angianthus micropodiodes, or Odixia angusta from Ozothamnus rodwayi, so there is that. We all have our areas of expertise and it would be pretty boring if we all tried to be sequence capture experts...

Finally, from a purely technical perspective, the talks were generally at a very high level. Certainly variable in quality - the major problems I saw were some very overcrowded slides and the odd overly rushed presentation - but there were definitely no bad or uninteresting ones. It was also interesting to note that while one usually sees much more variability on conferences, nearly all of the speakers today used slides with a very simple black font on white background template and without any fancy logos or corporate identity nonsense. I appreciate that. One talk was white font on black background, which is harder on the eyes, but in this particular case it might be considered somewhat justified because most of the figures it showed were microscopy images with a black background.