Evolutionary responses to environmental change: trophic interactions affect adaptation and persistence

Mellard et al. 2015 Evolutionary responses to environmental change: trophic interactions affect adaptation and persistence. Proc Roy Soc B 282: 20141351.

Effect of niche width on herbivore (solid red line), plant with herbivore (dotted green line), and plant alone (dashed blue line); under smaller (top) and greater (lower) temperature change. Moral of the story: bifurcations matter, people.

Effect of niche width on herbivore (solid red line), plant with herbivore (dotted green line), and plant alone (dashed blue line); under smaller (top) and greater (lower) temperature change. Moral of the story: bifurcations matter, people.


Lynsey McInnes

Lynsey Bunnefeld

‘We have an urgent need to understand and predict the response of individual species as well as whole communities and ecosystems to global change. However, the most commonly employed methods for predicting the response of species to climate change do not explicitly incorporate all fundamental ecological and evolutionary processes that may be major determinants of species responses to climate change.’

Haven’t we all read (and written) statements like that before?! Seems like we are missing a lot: dispersal ability, intraspecific variation, adaptive capacity, species interactions. These are all important issues to incorporate into predicting species’ responses to climate change. Different people take on different aspects and you get the crazies that try to model everything in one go and there is not a single study that does not have to resort to a suite of simplifying assumptions. And each study inevitably finds that their chosen issue IS important. What to do? It all sometimes feels completely overwhelming, as if the only worthwhile reaction is to shrug our shoulders and just wait and see what elements will be important and what not.

Meh. In the spirit of curiosity though and taking a step back from all the grimness, studies such as this one – a modelling study considering the different ways interactions between a plant and a herbivore could impact their responses’ to a changing environment – are, quite simply, interesting problems to tackle.

No doubt the authors could have set things up differently, included different assumptions or different parameters, but, from my limited knowledge of such studies, they appear to cover a broad parameter space and uncover a bunch of interesting responses. E.g., if plant and herbivore traits are correlated, their impacts on each other depend on their relative niche widths and initial conditions of the environment. To a big degree. This is cool stuff. Sobering, but cool. I love the line where they let slip that adding in a carnivore could change things further still!

Do you know what I would like to see though? I would like these authors, or other modellers, to make me a program or a package where I can play around with my own parameter combinations, perhaps from my chosen species or set of species, and make my own plots and response curves. Because I am not an expert in modelling like these guys are, I find myself having to believe what they say and the setup that they have chosen. I have faith that its probably fine, but it would be fun and informative to play around and recreate these plots and make my own. I think the more people that are comfortable (biologists and non biologists alike) with these kinds of analyses, the easier it will be to convince people that tipping points exist, that interactions matter, that responses are going to be quirky, but might be predictable sometimes. I think we need to get our hands much dirtier still. Do such tools exist?


Will Pearse

I always say a modelling exercise is only worthwhile if it tells you something you couldn’t have predicted at the outset. I find it hard to believe I could have guessed that the presence of a herbivore could so profoundly alter the evolution of a plant species, and so, while I would be lying if I said I followed all of it, I enjoyed this paper.

We say it very casually all the time, but predicting changes in species under climate change is all the more terrifying because we have no idea how those species will interact with each other in the future. Studies like this are very sobering to me, since things can get incredibly complex with only two actors. By complex, I mean switch-points that lead to different optima, because anything non-linear with some sort of tipping-point scares the goodness out of me. I have no idea how we could hope to empirically (and analytically) determine what would happen if there were another trophic level or additional herbivore/plant in this system. Perhaps arguments can be made for motifs (regularly repeating units of interactions) making things easier to model, but to be honest I’m sceptical. Grouping 300 interacting species into 10-30 different groups and then modelling those might be easier, but I’m not sure it would be easy. I’m happy to be corrected!

The authors discuss how pushing the plant towards a colder niche would badly-prepare it for climate change. I wonder if the number species we find doing ‘the wrong thing’ (leaving aside model slips!), which we sometimes attribute to species interactions, could be due to shifts like this. If a population is pushed into a particular region of parameter space, then I suppose the only thing to do is make the best of it. What if the ‘wrong move’ is only sub-optimal when we view things solely through the lens of climate? Maybe climate’s only part of the picture…

Advertisements

Convergent structure of multitrophic communities over three continents

Segar et al. Ecology Letters 16(12): 1436-1445. Convergent structure of multitrophic communities over three continents

Figs and fig wasps. Taken from the excellent figweb site (c) Simon van Noort (Iziko Museums)

Figs and fig wasps. Taken from the figweb site. (c) Simon van Noort (Iziko Museums)


Will Pearse

Will Pearse

Put simply, this paper is excellent. The authors have amassed an impressive dataset, performed a thoughtful and sophisticated analysis, and then explained the whole thing so clearly that it almost sounds easy. I look forward to trying to play around with some of these ideas in other systems!

It seems like there really has been convergence here: distantly related species are doing the same thing as each other in different places. So how the hell did this happen? While many evolutionary biologists I speak to seem to have a pretty good idea what they think convergence is, I think we’re still lacking a formal mechanistic model that can be tested. Yes, we can isolate parts of a phylogeny that looks convergent, but I don’t think we have a model of trait evolution we can use to model this and I’m not sure what it would even look like (what is the opposite of a Brownian walk?). Perhaps convergence happens when there’s insufficient dispersal for pre-adapted species to move in and occupy a particular niche. Perhaps convergence can only happen when there’s sufficient flexibility in a particular trait, thus labile behavioural traits should show more convergence and things like the Baldwin Effect will become important. Maybe there’s something special about fig wasps, and their emergence and mating on the surface of figs (they do that, right?) that makes them more susceptible to all this. Maybe it’s none of these things.

Perhaps the most important limiting factor would be the evolution of the figs themselves; I wonder if the most important methodological advance would be simultaneous evolution of fig and wasp traits, and simultaneous diversification/extinction of both taxa. Obviously work has been done on this already, but I’m talking about a more explicit derivation where, instead of individuals in a population interacting, there are individuals from two separate populations (figs and wasps) interacting according to some fixed set of rules. Thus a particular trait shift in one population would have to be matched by a complimentary shift in the other. I sense the maths would get quite intractable quite quickly (well, it would for me…), but simulation shouldn’t be impossible.


Lynsey McInnes

Lynsey McInnes

To maintain full disclosure, I am about to start collaborating with senior author, James Cook, so it is in my interests here to be constructive and probably err on the side of positivity. That said, I enjoyed this paper a lot! The fig wasp system is inherently cool and I thought the analyses here were exceedingly ambitious.The authors set out to test the relationships among fig wasp communities across three continents. According to measures of phylogenetic and ecological distance, do they follow the ‘inheritance’ (long term co-diversification, same ecological and phylogenetic diversities), ‘convergence’ (same ecological diversities got through different phylogenetic routes) or ‘constraint’ (ecological roles divergent because of constraints on colonisation and/or niche shifts by resident species (meaning phylogenetic diversity also different among communities) hypotheses. They find most support for fig wasp communities being similarly structured through ecological convergence.There are two sides of the fence on which one could sit with regard to this paper. On the one hand, the authors have built up a perhaps overly-complicated methodology in order to demonstrate ecological convergence when one has the feeling they already knew this result would emerge. These are fig wasp experts after all. For instance, the authors could have put the wasps into their guilds without any analyses at all. Similarly, I still don’t fully understand the ins and outs of the PVR and how that setup is able to decompose the variance into ecological, phylogenetic and joint components. I also worry about the low sample sizes and the power of a 35 species family spanning a ton of other wasp species (these qualms might be unfounded, I imagine Will would know).

BUT…my interest lies in rolling out such a methodology more broadly, perhaps to sets of communities with which one has little expertise. Then, for example, an objective way to delimit guilds is vital. And a step by step framework for analysis (the authors’ figure one) is a great tool. My mind is already ticking over to the time when one could stack various cross-continental analyses of community structure across groups into a big metaanalysis. Is convergence the norm? My feeling is such a meta-analysis is a long way off though.

One can imagine also developing the methodology within the fig wasp system (imagine having the data to do this for each of the 750 fig/fig wasp systems) or, as the authors suggest, looking at the structure in different parts of a single fig tree species’ range. I wonder if there are environmental correlates of the different signals?

I also liked a lot that the authors quantified both richness and relative abundance. I liked a lot that they had explanations for the reasons behind the signal of convergence (weird fig traits, niche shifts). I also liked that the authors distinguish constraint vs. convergence and wonder whether convergence ever follows constraint (and whether you could tell?).

I wonder if you could ever roll out these studies in some kind of experimental mesocosm? It would be cool to see the genetic underpinning of the various routes to similarity in community structure and how many replicates would get stuck in some setup due to constraint vs. reach the same ‘end’ due to convergence? You could add in the effect of various historical events (climate change! meteorites!), the possibilities are endless.

One final idea, it would also be interesting to look at multitrophic communities much closer together in space and see how movement across communities affects the patterns observed. Although the authors do suggest that their setup would work best for bounded communities. Hm.

So, yes, a very cool project. Thanks James and co 😉

Phylogenetic trait-based analyses of ecological networks

Rafferty, N.E., and Ives, A.I. Ecology (in press). DOI:10.1890/12-1948.1. Phylogenetic trait-based analyses of ecological networks

By Alvesgaspar (Own work) [GFDL or CC-BY-SA-3.0], via Wikimedia Commons

An animal and a plant, or a plant and an animal. They’re all in a phylogeny, somewhere, right? By Alvesgaspar (Own work) [GFDL or CC-BY-SA-3.0], via Wikimedia Commons


Will Pearse

Will Pearse

There are relatively few studies of ecological interaction networks that use phylogenetic information, and so this study, which no only does so but also suggests a new method for doing it, is great.

Anthony Ives is constantly developing new methods that are useful to eco-phylogeneticists and ecologists in general, and this paper is an excellent example. Phylogenetic linear mixed models are a way of simultaneously examining trait and phylogenetic information, and strike me as an extension of Ives’ earlier work with Matt Helmus. I love these methods; they allow us to answer questions we’re actually interested in (what determines species interactions) using phylogenetic and trait data. They don’t just look at a property of our data (like phylogenetic dispersion) and then force us to infer from that other properties of the system. My only criticism (which is better communicated by Nate Swenson) is we should go further – why collect phylogenetic information if we only use it to validate trait data? I, probably unfairly, want to shift the question to ‘what determines how species interactions evolve‘.

I think the interaction network literature is filled with cases of people trying to find unifying, abstract aspects of structure – motifs. I wonder if this is because the phrase ‘interaction network’ can cover so many kinds of interactions (pollinators, herbivores, competitors, etc.) in so many different taxa. Maybe eco-phylogeneticists can help, and there are over-arching phylogenetic patterns that can unify all these different systems and approaches. Every organisms on Earth fits into the Tree of Life somewhere, and that makes every single interaction network study, be it of bats or bell-flowers, comparable in some way. Which can only be a good thing!


Lynsey McInnes

Lynsey McInnes

I was really excited about this paper when I chose it. Traits, phylogenies, my pet interest ecological networks! I have the feeling the paper is really good, but I have to say I struggled with it. I think I dove too quickly into the deep end. Maybe I should have gotten more comfortable with the network literature or the traits on phylogenies literature, but I still haven’t learnt to do things like that…

Across biological sub-fields, everybody knows that species’ interactions matter, but, at least in the fields I have most experience of, namely macroecology and macroevolution, explicitly incorporating the effect of species’ interactions (or making them the point of an analysis) is only a very recent development. I think this means that any advance is a good one (and I feel like we’ve mentioned this a number of times before on PEGE).  So, in that respect, this paper is neat. It takes a well-studied, tractable set of plant-pollinator interactions and attempts to parse out the reasons underlying the different communities: do traits or relatedness underlie them? The authors make the case that methods such as theirs can help predict how guilds of species will fare under climate change as  it allows them to anticipate whether phenological mismatch will take hold or whether, in the pool of the other guild, species exist that can match advancing phenologies. Admirable.

I guess the missing link as I see it can be surmised from the title of the paper – ecological networks. What about evolutionary responses? Is it necessary to consider that mismatches might be ameliorated by co-evolutionary responses to environmental change? Is the timescale too short? Will there be less pressure for evolutionary change if, for the plants for example within the pollinator pool there are species ready to pollinate them following their flowering advance. Perhaps teasing apart this additional potential response will be a feature of a subsequent paper from these guys.

I also wonder how space affects these patterns. How do patterns and expectations change as one investigates multiple populations across a larger region? How does local adaptation affect responses? What about variation in the available plants/pollinators across space? What about gene flow among populations? How might one go about incorporating traits, phylogeny and interactions with intraspecific variation and space? Is that too much to consider at once?

The problem of pattern and scale in ecology: what have we learned in 20 years?

Jérôme Chave, early view. Ecology Letters. DOI:10.1111/ele.12048. The problem of pattern and scale in ecology: what have we learned in 20 years?

Below, we give our first impressions of this article. Please comment below, or tweet Will or Lynsey (maybe use #pegejc). Think of this as a journal club discussion group!


Will Pearse

Will Pearse

I am not a field ecologist, and I downed quadrats and upped my laptop because of the two papers Chave repeatedly cites in this paper: Lawton (1999) and Levin (1992). These papers made me see there was a place for biologists who looked across taxa, space, and time, to find general rules in ecology. This paper demonstrates just how far ecology has come in a little over two decades, and left me with an awful lot of hope for the next two. There’s a lot of material in this paper, and (judging by the sections I think I know something about) it’s all been very deeply considered.

Chave’s review almost operates in reverse of Levin’s paper. He starts with the eco-evolutionary problems that Levin left until the end, presumably because what was an exciting possible avenue in 1992 has now become an exciting reality in 2013. I encourage anyone who feels Chave overplays the importance of computing power to go back to Levin’s review; figure 1 is a 10 x 10 degree grid cell model of the world, and the only colour figure (13) pales in comparison to Chave’s (perhaps bafflingly!) complex reprinted figure 3. Whatever you think of the current vogue for ‘big data’ and complex modelling techniques, I don’t think anyone in 1992 would have predicted the changes that increased computer power, in combination with the open source R ecosystem, would bring. We are beginning to tame the complexity Levin hoped we could explore.

Which is why I would have enjoyed a discussion of agent-based models, and (perhaps more explicitly, if I missed the subtext) the importance of emergent properties of systems. Agent-based models are an excellent way of using information on individual-level actions to predict higher-level properties of systems, and I feel (as an outsider) are changing the way we look at species movement and dispersal. I’ve been thinking a lot about abstract properties of different kinds of systems (after re-reading this recently) – to what extent can we find general properties of ecological systems, at higher scales (be they temporal, spatial, or something else) that allow us to better predict changes in those systems? What general rules and patterns can we pull out of analyses conducted across multiple scales?

It is this idea of abstraction that made me particularly enjoy Chave’s section on using compartmental analysis to define the ecological niche. I’m not a bacterial ecologist, but I wish I were, because I think it’s almost impossible to quantify any other kind of ecosystem in this level of detail. Finding repeating units (like these network motifs) and using them to infer process falls right into the category of the emergent property driven analysis I mentioned above, and gives hope for definitions of niche that aren’t dependent on traits, environmental tolerances, or phylogeny. Grouping species along niche axes that derive from properties of their organisation, which are more intangible but arguably more real than things we can measure with a ruler, could be incredibly powerful. Despite the risks and potential pitfalls, measuring the niche according to dimensions of species’ own organisation, not what we decide is important a priori and can measure, sounds exciting to me.


Lynsey McInnes

Lynsey McInnes

This paper has been floating around online, hailed as a forthcoming classic. Perhaps. I have to say I’ve read it at least four times now and it doesn’t get any easier. There is absolutely tons of great and thoughtful stuff in here, but sometimes the typos, the poor sentence structure and some logical leaps made extracting that thoughtful stuff quite difficult (at least for me). Anyway, moving on from these critiques and focusing on what I was able to extract from the paper…

Scale matters! Looking back over the four papers Will & I have chosen so far for PEGE, this paper does indeed seem right up our street – turns out we are both a bit taken by processing operating across a range of temporal and spatial scales and seem drawn to studies addressing scale issues head on. So, at the very least, PEGE is helping us identify our research interests.

It is indeed amazing the progress ‘ecology’ has made in the past twenty years and this does seem largely due to access to more data, more computing power and the development of swankier and swankier methods and models. My fear is that progress on these fronts has been too rapid and that data and methods are thrown at sometimes quite vague questions, ‘just because we can.’ I think I might be biased here as this issue might plague my own field – macroecology – more than others (this certainly seems the case from some of the cool study examples outlined here). But, to sound a general note of caution, aimed mostly at myself, take things slow, think about the interesting questions and pursue their answers with appropriate methods and data (not necessarily the most elaborate methods and data).

One of the main messages I did extract from the paper is that biotic interactions matter in understanding species’ ecologies. Whether this in the form of interaction networks, community dynamics or broad-scale diversity patterns, how species’ interact with conspecifics, congeners, competitors, mutualists or across trophic levels matters. Recognising this and quantifying these interactions is going to be crucial in predicting how species and in turn communities will respond to global change. Sprinkle in the potential for rapid evolutionary responses (mostly likely different speeds for different populations/species) and these predictions sound hellishly difficult, but might still be possible.

The second main message was perhaps that a crucial species’ trait to consider is dispersal ability. Debate still rages on what dispersal is, but in terms of scale-dependent processes, it’s central. It can influence population connectivity, species’ range size, species’ ability to track good abiotic environmental conditions and/or biotic interactions and on longer temporal scales can influence rates of speciation and extinction. In short, getting a better handle on dispersal (ability, evolution, variation) will help unify our understanding of species’ distributions (and more).

Lastly, I really appreciated Chave’s call for more crosstalk among disciplines from geneticists to ecologists to earth system scientists, an understanding of emergent patterns will only be gotten from fusing information, techniques and expertise from disparate disciplines. Easy to say, hard to do, no doubt.

%d bloggers like this: