Phylogenies support out-of-equilibrium models of biodiversity

Manceau et al. 2015 Phylogenies support out-of-equilibrium models of biodiversity. Ecology Letters 18 (4): 347-356

Overview of speciation underManceau et al.'s model.

Overview of the new Speciation by Genetic Differentation model of Manceau et al.

Will Pearse

It’s been a good few months for Neutral Theory in Ecology Letters (…and so, by extension, PEGE!). In this paper, Manceau et al. put forward an extension of Neutral Theory, including a new model of speciation (Speciation by Genetic Differentation), and relax the dependence upon a static metacommunity. Both are exciting extensions to the theory.

The concept of the meta-community is something that’s always troubled me about Neutral Theory, because it seems a bit much to appeal to something outside the system to keep the system stable. Yet at the same time the only thing I can think of that’s less realistic than a meta-community is not having a meta-community, since clearly no community evolves in isolation. In this model the meta-community can change through time (i.e., it’s not at equilibrium); it’s no longer a deus ex machina, and is instead a part of the ecological theatre (see what I did there? :p).

Equally, the speciation mechanism the authors put forward is a useful development. I’m a fan of protracted speciation models, but my general problem comes from fitting them onto a specific evolutionary process. I don’t doubt they describe pattern and process well, but they don’t seem to be linked to one process in particular. Thus the genetic differentation model the authors suggest is, to me, extremely exciting. As with all these exciting new models, it’s almost a shame that the cleverest bit – the maths – is too complex to present in the body of the paper (I say almost a shame, because I’m certain I wouldn’t understand it!).

To me, the most important concept in this paper is that telling phrase ‘out of equilibrium’. Arguments about whether diversification is or isn’t density-dependent are never going to go away, but there are some who are calling for the debate to happen in the context of recent ecological theory about what carrying capacities in systems look like. Personally, I think that a discussion on density-dependence has to happen with an understanding of species’ abundances, and that means individual-based models. Work like this is an important step towards this.

Lynsey McInnes

Lynsey Bunnefeld

It has indeed been a good couple of months for neutral theory at PEGE. Here, we have another tweak to Hubbell’s original theory to bring phylogenetic tree topologies more in line with empirically observed trees. Specifically, the authors tweak the speciation mechanism from point or random fission speciation (i.e., instantaneous) into ‘speciation with genetic differentiation’ such that new species form only when they have accumulated enough mutations to be distinct genetic ‘types.’ Furthermore, the authors relax the assumption of constant metacommunity size instead allowing size to vary stochastically according to the growth (birth – death) rate of the clade.

I must admit I read this paper far quicker than it merited, so my thoughts are a bit hazy and any qualms might be unfounded. On that note, here goes…

I did appreciate the positivity bouncing around in this paper. The authors were resolutely positive about the capacity of phylogenies and macroevolution in general to inform us on diversity patterns. This was nice to see as many people, myself included, often despair on the ability of phylogenies to tell us anything.

Their two tweaks to neutral theory also sound, on the whole, sensible tweaks that make sense given what we know about species and given that we agree we want to retain the simplicity of the neutral theory while identifying the key assumptions that make it fall down.

First, speciation by genetic differentiation. Indeed, closely related species generally do differ genetically. Whether this difference is just an accumulation of neutral mutations or some kind of adaptive divergence (and which of the two kinds is more common) is another question. I felt like the authors could have discussed this issue more deeply because as a naive reader I was left wondering about the biological reality of such an abstract speciation mechanism. Sure, the model does not claim to be 100% realistic, but a discussion of the different signatures expected depending on the speciation mode would have been nice. The authors talk a lot about future directions and models they would like to compare theirs too (e.g., Pigot’s biogeographic model) and I look forward to hearing about these extensions. They somewhat cryptically refer to some lineages acting like speciation hubs that presumably shoot out new species willy-nilly. What kind of lineages would these be? Large ranged? Sexually selected? Weird mating system? Host shifter?

Second, growing/shrinking metacommunity. I agree with the authors that a constant metacommunity seems unreasonable. But I would have liked to hear more about how a growing/shrinking metacommunity might come about. Are we talking about finer partitioning of a finite area or colonisation of new habitats or competitive exclusion of crappy species, or what? Is a metacommunity the right term to use when we are thinking about the interactions of ENTIRE species. Could populations of the same species occupy different metacommunities? (Meta-metacommunities!! :$).

My hunch is the authors have also thought of all of the above and this is just a first pass attempt, albeit an impressive one that (again) shows that with just a few small tweaks the overall premise of the neutral theory is really useful in understanding general diversity patterns. I remain on the fence whether all this tweaking is destroying the original premise of the neutral theory (as I see it, to provide a conceptually simple null with which we can work out which non-neutral processes really do matter).



Extinction risk and tradeoffs in reserve site selection for species of different body sizes

Justin Kitzes and Adina Merenlender. Conservation Letters 6(5): 341-349. DOI:10.1111/conl.12015. Extinction risk and tradeoffs in reserve site selection for species of different body sizes

Single Large cat, Or Several Small Kittens? From minneybiscuit

Single Large cat, Or Several Small kittens? From minneybiscuit

Martina Di Fonzo

Martina Di Fonzo

I really enjoyed this paper because it falls exactly within three areas of conservation biology I am most interested in! Firstly, this paper addresses the age-old question of how much is enough to protect in order to achieve long-term species’ persistence. Specifically, it attempts to resolve the “SLOSS debate” which has been going on since the 1970s, with respect to whether Single Large or Several Small fragments are more effective for conserving biodiversity in a fragmented landscape. Rather than thinking about this question in a dichotomous manner, the authors conclude that the optimum variety of patch size and clustering depends very much on the individual species. This general finding is not novel, and it supports previous work on this topic, for example a paper by Nicol and Possingham (2010), who show that the optimum patch design for metapopulation restoration is heavily reliant on the metapopulation parameters themselves.

The authors do come to two very interesting, new conclusions. They use both mammalian metapopulation simulations and a real-world example to show that: a) having several small(er) patches is only beneficial when these are within 0.5-1.25 times the species’ maximum dispersal distance, and b) intermediate-sized species (~1kg) gain the most from reserve network clustering. These are very useful findings for conservation planning, and the authors discuss ways in which one could design reserve networks, while considering the optimum level of patch clustering for different sized-species. Their findings also touch on a second key point of interest to me, which has to do with evaluating the effectiveness of setting common persistence targets for conservation. This paper is a clear example of how important it is to evaluate whether species-specific goals could be more appropriate than applying blanket “rules of thumb”, which may not always be beneficial.

One point which I do not agree with in this paper is the authors’ conclusion that “network decisions should be made for the largest bodied species, as they will have the highest absolute extinction risk”. By basing network design on the largest species, wouldn’t this result in patches being too far away for many smaller (intermediate-sized) species to disperse to? This seems a little counterintuitive, especially since the intermediate-sized species were found to benefit the most from clustering. It would seem more appropriate to determine the patch distance and sizes which could result in the greatest overall reduction in extinction risk, across all the different species’ within the system. The costs of protection, as well as the likelihood of successful reserve network establishment are further factors that should be considered when drawing up such designs, which the authors do not mention.

The third area of conservation biology that this paper touches on, which interests me a lot, is their sophisticated metapopulation modelling to infer species’ extinction risk! The paper’s supplementary information has a very detailed description of the models, which could be a useful reference for anyone attempting a similar set-up. Although computationally complex, the authors candidly recognize the biological simplicity of their metapoulation models and landscape matrix, which could be perceived as a potential failing of this analysis. But I have to agree with them: there is always room for adding complexity, and it would be very interesting to test their method under further real-world situations, however, I do believe that the patterns they have found can provide a useful framework for conservation planning decisions.

So, overall, I found this paper very interesting! I would also like to give a heads-up to anyone who is interested in this topic that a colleague of mine at La Sapienza University in Rome (Luca Santini) is currently working on an analysis which reaches very similar conclusions, and his upcoming paper should be kept in mind as further support for these novel ideas.

Will Pearse

Will Pearse

This paper dredged up all sorts of undergrad SLOSS (Single Large or Several Small) debate memories, which was fun! I enjoyed the paper, and think it makes the point that there’s probably no such thing as an optimal reserve for all species in an ecosystem very well.

The next two points are more suggestions for a follow-up study, they aren’t intended to be critical. I think any discussion of SLOSS needs to incorporate edge effects, whereby patches are less good at the edge and so smaller patches can be less suitable than we naively expect. However, edge effects might interact with species body size, and I wonder what effect that would have on this demonstration of interspecific variation based on species body mass. Moreover, while some of these analyses are conducted over an ’empirical’ landscape, my feeling is that variation in habitat quality and type among the patches is going to have a really big impact on reserve design.

I view the SLOSS debate as having come to an amicable truce now, in part because there is no one single answer. I don’t really think the authors are trying to give a single answer to this debate, rather they’re trying to give advice driven by this particular case study/modelling exercise. We need to shift what kinds of reserves we’re designing for depending on what we want to conserve, and I think allometric scaling seems like a pretty good way of doing that when we haven’t got the data on a particular species, although I’d want to ground-truth any proxies before relying on them too heavily. Maybe the best answer is a modelling exercise tailor-fitted to your own system!

Lynsey McInnes

Lynsey McInnes

I liked this paper a lot. It was an extremely well-written, balanced and useful advance within the reserve design literature that concluded there is no optimal design that will cover all, e.g., mammal species. This is because they have a ‘characteristic range’ related to their maximal dispersal distance which means that, effectively, small things can’t overcome interpatch distances beyond on a certain size. Sounds reasonable, but has rarely been shown in such a comprehensive, quantitative way.

The authors are very open about limitations, including the structure of their model, the homogeneity assumed between patches, the allometric relationship used to relate body size and dispersal distance, and so on. Their conclusions, though, should be more or less robust to these assumptions because the effect is overwhelming lydriven by dispersal ability.

I was really happy to read this quantitative assessment of an issue that seems to be typically approached in a very qualitative way, focussed on the largest species. I might have gotten confused,  but I think the authors caution against this traditional approach as this biases design to large patches big enough for the largest of species and forgets about the right inter-patch distance for dispersal of a range of species.

Their conclusion that there is no optimal design is worrying. What additional factors could be included to get, at least a subjective, optimum? Perhaps some measure of function or phylogenetic ‘coverage’ or some other measure of health and/or endemicity of the species involved (ignore species that are screwed for other reasons, or found in good numbers elsewhere?). Like with the Joppa paper two weeks ago, including the economics of conservation might be useful too. Ooops, some of those might be construed as controversial, but at least might result in designs thought optimal, at least for some ends. Local optima?

I’ll end with a plug for a friend’s paper on mammalian dispersal distance. Whitmee and Orme conducted a thorough study to find robust predictors of dispersal ability for a painstakingly compiled database of mammalian dispersal distances. Merging it with the modelling framework set up in this paper might help us get one step further in designing good (lets avoid terms like best or optimal) networks for mammals.

%d bloggers like this: