A new dynamic null model for phylogenetic community structure

Pigot & Eitenne 2015 A new dynamic null model for phylogenetic community structure. Ecology Letters 18(2) 153-163

Figure 2 from Pigot & Etienne. Plots of likelihood of community membership under low (top) and high (bottom) rates of local extinction. Or, if you prefer, a series of variously coloured and not-coloured phylogenies.

Figure 2 from Pigot & Etienne. Plots of likelihood of community membership under low (top) and high (bottom) rates of local extinction.


Will Pearse

As if by magic, the ‘new’ approaches I hoped would appear last post have done so. It’s almost as if I know the posting schedule ahead of time!… Alex Pigot and Rampal Etienne have produced an analytical framework within which we can distingush between speciation, extinction, and colonisation in structuring an assemblage’s phylogenetic structure. Beyond that, they have developed a method that uses phylogeny to its true potential: not a proxy but unique data that helps us estimate evolutionary (speciation and extinction) and ecological (migration) processes of interest.

This is an important contribution because the problem of dispersal is one that has vexed many for some time, yet (with notable exceptions) received relatively little attention. Dispersal from a wider source pool provides an important link between ecological and evolutionary time-scales that we need to model. A species ‘appearing’ can either have an evolutionary (speciation) or ecological (dispersal) origin, and that their DAMOCLES model can at least start getting at that distinction is important. It is of little surprise to anyone who has followed these sorts of studies that phylogenetically overdispersed communities can result from something other than competition, but linking its origin to evolutionary and dispersal events outside a community is interesting.

There are, of course, additional complexities that could be built on top of this model, and I’m not going to bore you by rambling on about traits because I think you all know where we want that to be going. However, I think it’s important to explicitly consider the meta-community (or source pool, if you prefer) from which these species are being drawn. Focusing on one assemblage is useful, but the reality is that speciation and extinction dynamics are happening at biogeographic scales, and we desperately need to link community-scale models such as these with those. Considering multiple assemblages undergoing these kinds of dynamics could be a good place to start. I wonder if the limiting factor may be finding something analytically tractable; while simulating individual communities linked within a wider system is reasonably feasible, doing so analytically (as the authors seem to have started doing) is more difficult.


Lynsey McInnes

Lynsey Bunnefeld

Alex Pigot has a way with null models. He’s already shown that the arc of species’ range size over the course of a species’ lifetime is not necessarily the result of deterministic processes and now he (and Rampal Etienne) have shown that common patterns of community assembly need not be the result of negative biotic interactions if an appropriate null model is used. Wow.

This paper is a great example of a couple of key points that I am most definitely guilty of ignoring. 1. taking time to think whether your null model is biologically as well as statistically null is important. 2. important insights can be made even before your model includes every last contributing factor (see Will’s post above). 3. data examples are important to illustrate your method. Nice.

I’m now going to largely disregard all those things I just said were important and wonder how you might extend this model and wonder what pesky real world effects might topple the null expectation.

I wonder how biotic interactions with non-clade members affect community assembly, i.e., competitors, predators, prey, hosts, etc. I wonder what a null model for this might look like? Should hosts/parasites (for example) evolve in tight coevolution, or not? I wonder what repeat processes of community assembly look like? Always the same, or not (I’m sure this has been treated in microcosms and by Gould). How would phylogenies help with these questions? In the same vein, I wonder how what happens to the new sister species that does not enter the community of his ancestor? I guess I am pondering the effects of space. I have no doubt the authors have too, and would not be surprised if they already have the answers up their sleeve. The authors deal elegantly with variation in a quantitative trait (or traits) meant, I think, the characterise niche. I wonder what happens when you throw in variation in other traits, probably dispersal ability (haha, with all the trauma that goes with defining and measuring that!).

Advertisements

Modelling competition and dispersal in a statistical phylogeographic framework

Ranjard et al. 2014 Modelling Competition and Dispersal in a Statistical Phylogeographic Framework. Systematic Biology. 63: 743-752.

phylo_comp_dispersal

Figure 1 from Ranjard et al. The three scenarios are based on the landscape and colored locations displayed at the bottom left.’ Can you guess which says which?


Will Pearse

The authors have put together an impressive method that detects whether competition and (consequentially?) biased dispersal can be detected using data on species’ present-day distributions and phylogeny. Does the method work? Well, that depends on whether you agree (1) with the model, and (2) that present-day distributions map in a 1:1 fashion onto past processes/distributions.

I’m pretty sold one (1), but then again I have neither the time nor the capacity to go through the maths in much detail! The introduction was interesting to me, since it smelt a little bit like it invokes the community phylogenetics sin everyone loves to hate; I would get mercilessly slaughtered for invoking competition or co-existence on the basis of niche overlap. Inferring about past processes at this scale, however, is a different beast – we’re data and method limited, so I think it’s completely acceptable to start putting testable models out there. Moreover, unlike those sinful community phylogenetics papers, this paper has a testable, verifiable model.

My only real problem comes with (2), and I somehow doubt the authors would disagree with me. The island case-study they use is a great one, and I think the model works perfectly for data like that where we’re dealing with very discrete, very tractable movements between patches of land. However, I think some care would have to be taken when dealing with continental distributions where variation in habitat type will mask the dispersal events the authors are looking at. Again, I think the authors are more than aware of this, and I don’t think they intend this method to be fit to GBIF data or something, but I’m interested to see if this could be tried with some modifications. Combined with this paper (which we’re covering soon), it’s a good time to be a phylogeny nerd – the methods are coming in, so now we need to just use them 😀


Lynsey McInnes

Lynsey Bunnefeld

I enjoyed this paper, even though it made my brain explode. It is a funny mix of idealistic and hopeful. The authors set up a framework to detect the effects of dispersal and competition on genealogies. They argue that population expansion and establishment (eventually speciation) are processes affected by dispersal ability (here characterised by an overall rate of dispersal away from a source population and a dispersal kernel shape parameter), but also by competition (here characterised by how much an already occupied location is refractory to further occupation). The authors argue that researchers have focussed on the effect of dispersal (generally characterised as distance among sites rather than intrinsic organismic traits) while ignoring the effects of competition (despite a large body of ecological literature on the effects of competitive exclusiveness). This is probably true and is likely due to the fact, as the authors note, that it is pretty damn hard to capture the effects of competition adequately.

The authors make a start here by setting up a relatively simple model to find out whether genealogical shapes and geographic patterns of occupancy suggest evidence for biased dispersal among sites (more to closer sites) and competitive exclusion (longer branches to the present as fewer sites are available for colonisation/establishment). They find evidence for both in a bush cricket genus endemic to the Hawaiian archipelago and their sensitivity analyses generally suggest they are able to detect competition when it has had an effect.

So far, so good. I don’t really feel qualified to destroy their methodology, so I’m not quite going to. I imagine there are better way to characterise these processes and that there are probably alternative explanations for the signals they recover, but we won’t dwell on them here.

My biggest problem with this paper and I am not at all sure whether it is not just me being dim is that I’m not really sure at what time-scales this method is expected to perform well. The authors jump back and forth between species and populations. Considering populations, I doubt you would be able to reconstruct bifurcating phylogenies with the resolution needed to detect longer/shorter terminal branches. Indeed, bifurcating trees are not the expectation and only populations geographically very far away from each other are expected to be isolated enough that gene flow from neighbouring populations would not homogenise gene pools. I.e. if a lot of individuals from the source population are dropping on to a second population, they are unlikely to be excluded unless they are so locally adapted to some place else that the environment kicks them out rather than conspecific competitors? Perhaps this is what the authors are trying to characterise anyway, and I am just misunderstanding their definition of competitor/competition.

I think my concern is that any successful model of genetic variation in space should probably take into account dispersal ability, competitors, but also environmental variation (this might be abiotic factors, but also biotic factors such as interspecific competition or intraspecific competition (if populations of conspecifics vary according to other aspects of the environment).

Again, I might be confused, as my head has yet to settle on whether it thinks according to bifurcating trees in macroevolutionary time or networks in population genetics time. My hunch is the authors have tried to operate across both time-scales, and I am not sure it works (or at least I can’t quite grasp it).

I’ll leave it there. Any help, much appreciated!

Mechanisms of maintenance of species diversity

Peter Chesson. Annual Reviews in Ecology, Evolution, and Systematics 31: 343-366. Mechanisms of maintenance of species diversity

Species' relative abundances fluctuating over time under various models of coexistence. If you just want to think of it as a pretty picture, that's probably OK too. Figure 2 from Chesson (2000).

Species’ relative abundances fluctuating over time under various models of coexistence. If you just want to think of it as a pretty picture, that’s probably OK too; we won’t tell anyone. Figure 2 from Chesson (2000).


Will Pearse

Will Pearse

I’m shocked how many people consider this a classic that “must be read” and yet haven’t read it themselves because it contains “too much math”. I don’t care if you haven’t read this (I’ve not read every paper most people consider a classic), and I think focusing on what people haven’t done is pointless, but I think this is a very readable coverage of what is (otherwise) very difficult math. I do find this field very math-heavy, but with only 9 equations (~1 per 3 pages) this is a really insightful review that for novices like me is helpful.

Chesson very nicely re-phrases coexistence around the ratio of species’ competitive differences and their niche differences; if species are sufficiently different, or compete sufficiently little, they will coexist. This came up an awful lot in the community phylogenetic talks I attended at ESA last year, in part because it torpedoes the assumption made by many that closely related species are more similar to one-another and that’s all that matters for co-existence. I manically circled the idea that species’ niches have an effect (e.g., reducing resources) and a response (e.g., ability to grow given certain resources). Chesson seems desperate for us to stop thinking of a niche as something that is solely a function of a species itself: it’s a function of the context within which we see the species, and that is always shifting. Even when we say a species’ niche involves competing with other species, we’re still missing the key component that the species is using up resources, thus warping the effect-response space that all the species around it experience and themselves modify.

Which sounds a lot like I’m saying the paper is about non-linearities and changes through time; it isn’t, and Chesson very artfully points out that a lot of insight can still be had by setting up simple models in well-considered ways. It is pretty dismissive of Neutral dynamics (…although it was written in 2000!) and the take-home from the section/paragraph “nonequilibrium coexistence” could be paraphrased as “stop using this unhelpful phrase”. I was particularly struck by how Chesson viewed the importance of Neutral Theory for exploring biogeographical dynamics; more than a decade later we’re starting to do this, and we’re even having discussions of how species can evolve, under neutral dynamics, to not be neutral today. This opens a whole can of worms as to what we can usefully call a N/neutral model, but more importantly helps us unify questions at a number of different levels of ecology. Which is a good thing!

Much of what’s in this paper is probably uncontroversial to a community ecologist (right?), but I think remarkably little of it has found its way into the mainstream evolutionary biology literature. I find that interesting because I’m often surprised by how much evolutionary biologists keep track of what ecologists are doing. I can’t think of any serious modelling studies where species’ effect and response traits are seriously modelled across a phylogeny (please correct me!), and I wonder what we would find if we looked.


Lynsey McInnes

Lynsey McInnes

I can’t imagine I am the only PEGE person to have cited this paper on the mechanisms of species’ coexistence without having carefully read it? Its been cited 1694 times! So, choosing it for this week’s PEGE was a great excuse to actually sit down and use some train time to read it through and through.

Now, I have always cited this paper when writing about macroevolution and the idea of equilibrium species numbers and turnover in ‘evolutionary’ time. Oops. Chesson sidelines this view of species’ coexistence early on in his paper and instead focuses throughout on ecological/contemporary notions of species’ coexistence in a – relevantly sized, more or less closed – patch. No matter, I am convinced that most if not all of what he talks about is also relevant or at least interesting for people working at longer timescales.

The problem perhaps is that too many people have jumped on the bandwagon of these ideas being relevant to understanding the build up and maintenance of species diversity that we have become blinkered to the possibility that ecological limits might not be constraining diversity at broad (temporal and spatial) scales. It feels so easy and so neat to extrapolate Chesson’s (and others) equations and explanations of how populations of species manage to stably coexist with each other (a delicate balancing act of getting more intraspecific than interspecific competition, with divergence along at least one relevant niche axis) to how ENTIRE species diversify in the presence of one another until their niche space (in physical or ‘hyper’ space) is full. It is also really easy to obtain patterns, for example in phylogenies, that agree with the idea of diversification slowing as niches fill up.

I have a feeling we are on the cusp of entering a new phase of macroevolutionary analyses where we break ranks with trying to match one for one ecological and evolutionary phenomena. I think it is already much more common to think of what units within species evolve (populations/metapopulations depending on gene flow) and also to think of niches as much more labile (e.g., what about niche construction or extent of niche breadth). Similarly, in both macroecology and macroevolution, biotic interactions are moving from postscript to centre stage as more data becomes available to address the effects of biotic interactions on the patterns we can observe and experimental systems are emerging where these effects can be empirically manipulated.

Turns out no matter how hard I try to leave it behind, I am still a macro-scale biologist at heart and it is fun to pull out the macro-scale implications of more or less any paper that I read.

%d bloggers like this: