Do communities exist? Complex patterns of overlapping marine species distributions

Leaper et al. 2014 Do communities exist? Complex patterns of overlapping marine species distributions. Ecology 95:2016–2025


Benthic flatfish and benthopelagic cod on a shore – Jan van Kessel senior, 1626–1679 (Or: what you get when you google ‘demersal fish australia’)

Will Pearse

Will Pearse

I picked this paper because I felt it was a more stats-y/empirical-y way at getting at some old debates in ecology. The methods seem both very complex and very simple at the same time; if you’re interested, skim the Dunstan paper(s) they reference to get a handle on what’s going on. In essence, they’re simultaneously binning species intoarchetypes (groups) and figuring out what predicts where those archetypes are. It’s a binned species distribution model. Speaking as someone who tried (and failed) to do this, I can assure you it’s a hard thing they’ve done a good job of 😀

The authors find archetypes, but do a good job of stressing that this doesn’t mean they’re present all together in the same place all the time. For me, this variation is almost the most interesting thing, because it could reveal competition. Under the archetype approach, things become simpler because you’ve restricted yourself to the species you most expect to be closely-interacting, and so hard-to-find processes should be easier to detect. I’ve spent a lot of time bleating on about how competition can take place within the context of habitat filtering, and how I think phylogeny is a great way to get at this – I think this kind of approach is even better. There should be different traits that determine species ‘ distances from the centroid of their archetype than those that determine how distant each archetype is, because these two things should reflect fundamentally different processes. Goodness only knows what space you’d calculate these distances in, though!

There may be a further temporal and spatial component to these communities. Fish don’t exactly stay still, and a diver conducting a transect is likely to encounter a number of ‘swim-by’ species that are hovering around the area. For species to interact they have to spatially and temporally overlap, and I wonder to what extent all of these species were doing that. I don’t mean this as a criticism: by treating these species as an archetype, we can drill down and explore what each species is doing. How many predators are in a particular archetype, and how many prey species? How frequently were two predators within an archetype seen at the same time (–> whether they compete for prey). Could the method be extended to deal with fuzzy archetypes, such that commonly-encountered predators could affect multiple archetypes? Why is it that, for the whole of this article, I kept wanting to type “habitat type” instead of archetype? Is there a difference? I’m quite excited to find out!

Lynsey McInnes

Lynsey McInnes

This was a funny paper on the back of a suite of PEGE posts commenting on rather grand visions of the future of various subdisciplines of ecology. I think it might have done us good to remember that applying swanky new concepts and frameworks to actual datasets in the first instance involves a lot of work collecting a lot of data and then a degree of frustration interpreting results that don’t fall neatly in one neat quarter of some kind of hypervolume.

So, kudos to Leaper et al. for a well-executed study apparently conducted by people that know their study system well alongside people developing new methodologies. The authors are interested in the nature of communities, whether they are discrete and easily circumscribed or more fluid, with species composition merging into one another across neighbouring communities.

They use the idea of a species ‘archetype’ to get at these questions, this being the quantification of similarity in species’ responses to environment. I.e. rather than seeing which species co-occur in space, the method clusters species based on results of species distribution modelling (i.e. this set of species likes it hot and wet, and so on).

I am on the fence as to whether this idea is a good one. On the one hand I like to cluster and to generalise, on the other I wonder about circularity and the proliferation of too elaborate data transformations. Let alone all the dodginess that comes with SDMs.

The authors find that although they can identify different clusters of species, these species don’t correspond to distinct spatial entities (‘communities’) suggesting that communities, at least of demersal fish or macroinvertebrates off the southern coast of Australia and northern coast of Tasmania, are not highly structured units. Are we surprised?

I think that I, ever the macroecologist, would have liked to have gone one step further and investigated whether there are traits associated with archetype strength (probably things like dispersal ability, philopatry, niche breadth, species richness, the usual) and by extension how ‘closed’ communities generally are. To do this properly, you would of course need a broader dataset spanning a wider range of communities and species. And you get back to the mismatch between burgeoning numbers of frameworks and a limited amount of appropriate data.

The authors note that communities appear to be made up of a selection of species from multiple archetypes that all match bits of the environment found at a site, but that environmentally similar sites are likely to have a different set of species (albeit chosen from the same archetypes). I think this is a neat, but also intuitive, observation that there are assembly rules, that all bases need to be covered, but exact species can be subbed in and out. I like the idea of substitutable species and wonder how true it is in terms of functional redundancy. My bet is species are not really substitutable, but wonder how much variation in this ‘trait’ there might be.

In short, I think we still have a long way before we have a good grasp of how communities are structured and how that structuring changes through time and across space. My hunch is that careful studies such as this one will be important in building enough real world examples of how things work, whether or not the idea of archetypes is crucial or superfluous to the debate.


Niche syndromes, species extinction risks, and management under climate change

DF Sax, R Early, and J Bellamare. Trends in Ecology and Evolution 28(9): 517-523. doi:10.1016/j.tree.2013.05.010. Niche syndromes, species extinction risks, and management under climate change

Do you have a niche syndrome? You can get a cream for that you know... From Sax et al. (joke from Sarah!)

Do you have a niche syndrome? You can get a cream for that you know… From Sax et al. (joke from Sarah!)

Sarah Whitmee

Sarah Whitmee

I chose this paper before I knew that Lynsey and Will were to review a very similar one just two weeks earlier, coincidence or something more sinister? Actually, I don’t think either of those things, but is in fact due to a new phase in the field of species distribution modelling (SDM), one in which concepts are clarified and the assumptions of models are questioned and tested. Well that’s my hope anyway…. Nevertheless, it is becoming increasingly clear that a big assumption made by those practicing the dark art of SDM, namely a linear relationship between the realised niche (the area of environmental space actually occupied by a species) and the true environmental tolerance limits of a species does not exist, at least not for most species. This was illustrated in the Araujo et al study discussed here, where it was neatly demonstrated that lower thermal tolerances varied widely across species while upper tolerances were much more conserved. Happily, I think this study complements the earlier paper, phew!

The authors start in true TREE style with a nice overview of the current state of the field and the definitions of key terms. While often this is just restating the obvious it’s actually a pretty useful exercise for anyone working in SDM to go through, varying definitions of the fundamental, realized and potential niche plague the discipline and formally stating them for yourself can help later down the line when trying to interpret and understand models.

They then get down to the business end of the paper, the introduction of a new concept: the ‘tolerance niche’. The tolerance niche is defined as “the set of physical conditions and resources that allow individuals to live and grow, but preclude a species from establishing self-sustaining populations”. While I’m not convinced that the field of species distribution modelling needs more jargon I can see the usefulness of this concept in theory, specifically in relation to predicting the impacts of climate change on species persistence. The idea is that while a species cannot thrive in these tolerance zones they can persist temporarily in them, for example during dispersal to reach new suitable climates or to outlast temporary climate fluctuations. The introduced concept is then used to illustrate a number of niche syndromes, or ways we might want to think about species responses under climate change. They choose horticultural plants as a key group for illustrating how you could formulate hypotheses for these niche syndromes and also how you might work out a species tolerance niche, from evidence of specimens in botanic gardens outside a species native distribution.

Overall I like this paper, its clear and well thought through. I buy into the idea of the tolerance niche and think it’s a good step towards more dynamic species distribution models, rather than the time-slice approach employed in earlier analyses. Being a macroecologist I have a problem with this paper that I often encounter with the more fine scale SDM work. While the concept or model works well for a single species or a particularly well studied ecosystem (its no coincidence that a large proportion of SDM studies are all about the South African fynbos you know) these concepts fall down due to a lack of data when you try to scale up for multi-species predictions. The example species given in the paper works beautifully for the concept they are proposing but I wonder how many other species show such clear relationships. So I would have like to have seen more real world examples to convince me that the tolerance niche can truly be estimated. I know very little about plants (outside the ones in my garden) but I’m nervous about the idea of inferring tolerance simply from presence in a botanic garden outside the native distribution with a better understanding of how the plant is managed in situ for example it might be protected against winter frost or supplemented with food or key nutrients. I guess for species of key conservation concern such an approach might pay dividends.

I did like the idea of managed relocation for slow growing species, putting them in an area that they can tolerate in the short-term but which will eventually become climatically suitable for growth and reproduction. It’s a risky business though and I would want a much higher confidence in my climate models before attempting such a bold step.

To sum up the tolerance niche is a neat concept but how applicable it will be in the long term, I’m not sure. I’m a fan of these authors though so perhaps other will have a different view.

Will Pearse

Will Pearse

Like Sarah, I like these authors (and I know Lynsey does too!), so perhaps this is going to be a bit of a biased PEGE. I view the essential idea of this paper as defining the idea of a tolerance niche: conditions where a species can just-about survive, but can’t form a self-sustaining population. I buy it, and think it’s a nice idea and a great paper.

Indeed, I think similar ideas have been floating around in population biology for some time. We have source populations, that fire off propagules into the meta-community, and sink populations, where propagules arrive, individuals persist, but there’s a net loss of individuals and so that population can’t sustain itself. These concepts have really helped population biologists think about meta-community dynamics, and as we try to link macroecology more intimately with local-scale abundance changes and interactions, it seems sensible to conceptually link these concepts.

I wonder if we can go a step further, and stop treating different parts of the niche (fundamental, realised, tolerance, etc.) as discrete boundaries, and instead be up-front in acknowledging that we know species do better in certain parts of their niche than others, and maybe try to quantify that. Essentially, just have some value (why not fitness) that changes throughout niche space, and acknowledge that, in some parts of niche space, fitness will be so low that a population won’t be sustaining. Indeed, such an approach (borrowing heavily from Chesson) would let us handle realised vs. fundamental in a much more intuitive way, because we could distinguish between niche differences and fitness differences when trying to understand whether species can coexist. Because, as I’m sure we can all agree, parameterising a fitness function across niche space would be really, really tractable and easy to do :p

Hopefully the above makes sense; I have a very serious case of man flu!

Lynsey McInnes

Lynsey McInnes

This paper touches on plenty of topics that we have covered in different guises here at PEGE, with a perhaps more applied angle than most. In essence, the authors seem to be drawing attention to the potential importance of a kind of buffer zone of conditions that species could survive in beyond their current distributional limits (the tolerance niche) and how this zone might be extremely important in mitigating climate change induced species’ disasters.

The idea seems pretty reasonable and meshes well with experimental studies that find that translated populations can make it in conditions not found anywhere within their range. The authors also showcase a brilliant dataset of naturalised and garden centre/botanical garden populations of a plant species in the eastern US. The inclusion of this data lifts the paper from one based on loose concepts to one with real results; that was nice!

However, I do still worry how ‘useful’ the concept of a tolerance niche is going to be beyond these plant examples. In effect, the authors are creating an additional category of niche that in most cases is going to be quite difficult to identify? I would have liked to have seen one additional step to go the figure one and that would be to investigate traits that predict the extent and/or location of this tolerance niche. For instance, the authors draw attention to the distinction between short- and long-lived species, but are there any other distinguishing features? These could be along the lines perhaps of large range (probably also has a bit of tolerance niche), restricted range endemic (probably doesn’t), is part of a complicated food web (probably doesn’t have much of a tolerance niche), most kinds of generalist (probably do have tolerance niches). The next step, as the authors emphasise, is where is the tolerance niche located in relation to the realised niche and how easy is it to get to?

The above just gave me the nagging feeling that the tolerance niche concept might be quite difficult to implement as, like everything macro- it seems (I’m having a down on general patterns week), these things depend on so many other things: landscape structure, biotic interactions, the usual suspects. So, while the concept of the tolerance niche could provide US with a kind of buffer so that we can worry less about species’ survival (they can probably tolerate a bit more heat, drought, what have you) than they currently do, it seems like a difficult concept to draw strong or helpful conclusions from across broad taxonomic or spatial scales.

In conclusion, this was a well-written, thoughtful paper, but I am not convinced that the new concept and piece of jargon are robust or flexible (can something actually ever be robust and flexible?) enough to be rolled out very widely. As always, its a data problem…

%d bloggers like this: