Phylogenetic relatedness and the determinants of competitive outcomes

Godoy et al. 2014 Phylogenetic relatedness and the determinants of competitive outcomes. Ecology Letters 17 (7): 836-844

Figure 3 from Godoy et al. How fitness, demographic, and competitive differences vary with phylogenetic distance.

Figure 3 from Godoy et al. How fitness, demographic, and competitive differences vary with phylogenetic distance.

Will Pearse

In a fantastic follow-up to the many criticisms of the community phylogenetic approach, Godoy et al. fit a form of the Chesson framework to ecological data, and find that while fitness differences are greater among distant relatives, competitive differences are not. Being phylogenetically dissimilar did not mean that species were more likely to co-exist.

This is an excellent demonstration of a point that many have suspected for some time, but few (none?) have been able to conclusively show in a field experiment. This probably has something to do with the work involved in doing it…! Of course, that it’s been found once does not mean it’s a general pattern, but along with other work from the same authors decomposing traits into niche and fitness components, it seems empirical ecology is now matching its theoretical counterpart. Some are going to take papers such as these as the first nails in the coffin of community phylogenetics: personally, I think they open the door to a whole world of new approaches that we’ve been wanting to explore for some time.

Generating hypotheses about the kinds of traits that map onto different kinds of evolutionary processes means we can ask more sophisticated questions about evolutionary ecology. We don’t need to just stop at declaring that a trait shows ‘phylogenetic signal’, we can ask what model of evolution generated these traits, and (more importantly) how the evolution of those traits interacts with how they play out in species’ modern ecology. Indeed, that’s what many community phylogeneticists have been trying to do since the very beginning.

Now we can start asking more nuanced questions about the kinds of evolutionary models we are fitting. Measuring the traits that enable co-existence in one area is fantastic, but it’s unlikely that only the eighteen species in this study evolved in isolation. How did the surrounding flora (and interactions in other environments) affect the evolution of these interaction components? If (as the authors rightly argue) Brownian motion gives us very little predictive power for deeper phylogenetic structure, are there alternative models that might? Is it ever truly possible for competitive interactions and hierarchy to be strongly conserved, if diffuse competition among many competitors is frequent? If competitive hierarchies change over time, does it make sense to ask if a particular snapshot of them, in particular environmental conditions, is evolutionarily stable? Personally, I think it’s a good time to be a community phylogeneticist…

Lynsey McInnes

Lynsey Bunnefeld

Unlike Will, I’m not a community phylogeneticist (still not sure I buy into communities) and haven’t been following the recent developments in community phylogenetics that seem to be making it a much more robust field (see Will’s post above). Instead, I just jumped into this paper without previously ever having thought of the way you could split up species’ differences into stabilising niche- and average fitness- differences. What a good idea and what a shame that distinction wasn’t recognised long ago.

The authors then go on to see if they can untangle how these two features relate to phylogenetic distance using some nifty field experiments with 18 plant species. Again, I got overwhelmed by the fanciness of the experimental design and the work involved in it. And am happy to believe their findings that only average fitness differences show phylogenetic structure (more distant relatives have bigger differences) and that increased variance over longer phylogenetic distances mean that communities as a whole don’t show phylogenetic structure.

Being the macro person I am, I wonder how these results generalise to other communities and how you might go about finding out without having to conduct an epic field experiment every time you want to try. I think these authors have already published theory for these ideas so it is definitely time to get out of the computer and into the community (haha) but just how might you do it? Early community phylogeneticists went to town fitting models to species presence/absence in areas and giant phylogenies, clearly we need to be more nuanced than that. Could we go a roundabout way and find the traits that underlie the average fitness and the stabilising niche differences and use these in a similar framework to Godoy et al. advocate here? Has this been done already?

The authors find that variance increases with increasing phylogenetic distance, does this mean that clear patterns will not be found as we zoom out from narrowly defined communities? Is this OK?

Will sees these developments as a kind of new dawn for community phylogenetics. I just wonder whether the new dawn is not just tearing the field apart in increasingly nuanced ways. I for one am not confident that we can use phylogeny to robustly predict how communities will respond to change or use snapshots of current communities to work out how they got put together. At least not without a lot of knowledge of the system in hand and then who needs these phylogenetic metrics anyway?


Hyperdominance in the Amazonian tree flora

Hyperdominance in the Amazonian Tree Flora. ter Steeg et al. Science 342 (6156). DOI:10.1126/science.1243092. Hyperdominance in the Amazonian tree flora

I can't see the Amazon for all these tree plots - taken from ter Steege et al.

I can’t see the Amazon for all these tree plots – taken from ter Steege et al.

Jun Lim

Jun Lim

Among other things, the authors set out to try to estimate how many tree species are in the Amazon, how they are distributed in space and among habitat types. They did this in part by extrapolating the total Amazonian tree diversity by fitting the mean rank-abundance data for over half a million trees within 1,000 well-studied plots in the Amazon to Fisher’s log-series. They found that there were “hyper-dominant” tree species, which represented about 1.4% of Amazonian tree diversity while representing over half of all trees in the Amazon. On the other hand, the rarest 11,000 species accounted for a paltry 0.1% of all trees! Furthermore, as it turns out, this inequity in abundance among tree species in the Amazon had an interesting spatial pattern. These so-called “hyper-dominants”, had on average larger geographic ranges, but the majority of them were found to be dominant in only one or two out of several distinct forest types, suggesting that dominant species were habitat-specialists.

This paper, however, leaves me with an intense wanting, although these findings are clearly the first of many awesome papers to come. Firstly, it would be interesting to see how these patterns of dominance generalize to other tropical tree systems, such as the ever-wet forests of South-east Asia (where I grew up, albeit in the virtually deforested Singapore). Anybody who knows even a little of the rainforests of Malaysia and Indonesia thinks immediately of the dipterocarp trees (Dipterocarpaceae) that form much if not most of the canopy.

Another thing that really got me thinking was the idea of scalar-dependency in niche specialism. Sure, if you are a species that does well in one particular soil type, all the small-scale heterogeneity in habitat does not matter much to your distribution (kinda like generalism). But in the larger scheme of things, you start to appear much more constrained (specialism). What’s interesting is that this plays out in a consistent and similar way across many species and across many different habitats. From a community assembly standpoint, this brings up the age old question of the relative importance of dispersal limitation (high numbers of species can effectively coexist even if a habitat was homogeneous) and niche-based explanations for high species diversity, especially considering the strong role of habitat heterogeneity across many systems (which filters for different communities in different parts of the landscape). I feel it is still unlikely that the relative role of these two processes will be disentangled any time soon, although this paper was a big step towards that goal.

Will Pearse

Will Pearse

It’s hard to think of an image in ecology that represents much more blood, sweat, and tears than the one at the top of this article. 567 plots would be a lot anywhere in the world, but dense Amazonian rainforest does not make for easy fieldwork. That you can go online, register, and download much of this data right now makes this even more wonderful; thank you everyone involved in RAINFOR!

Like Jun, I sense this is the first of many papers working with this dataset (and is an excellent start at that!), but I sense it may still take us some time yet to get a handle on the Amazon! I’m having some trouble getting my head around the spatial scale of this dataset, and so I’m unsure how I feel about the most common (hyperdominant) species in the Amazon being habitat specialists. In many ways I’d be shocked if species were truly generalists in the sense that they were everywhere across the Amazon, although I suppose I’d need a dataset like this to be sure! I’m particularly interested by how more diverse genera are less likely to contain hyperdominants. I’m tempted to infer that because hyper-diverse genera are more similar to one-another and have similar ranges, they are competing too strongly with one-another to become dominant. Given there’s a taxonomic effect to hyperdominance, perhaps a phylogenetic analysis would help get at these issues (…although I’d rather not be the one making a phylogeny of the Amazon…!)

I think it’s legitimate to ask how many species there are in the Amazon using this dataset, and I’m frankly amazed by how flat the middle of the rank-abundance plot in figure 2 is for the Amazon. While I agree with the authors’ general conclusions here, I am slightly concerned about extrapolating out into such low population sizes. It’s probably fair to say that no one single curve can describe both the hyperdominant and extremely rare species in the Amazon, and the extrapolation is based on assuming that a straight line that follows the medium-richness species will cover everything. I’m sure the authors would agree that this is a simplification; naively I would expect expect this estimate to be too high, yet their estimate of ~15000 species in the Amazon actually struck me as quite low when I first saw it. I guess much of this might fall back to what we’d be happy considering a species; trees are not known for playing by phylogeneticists’ rules, and maybe very rare species can survive for quite a while in the Amazon thanks to outcrossing and hybridisation.

Lynsey McInnes

Lynsey McInnes

I always open a Science paper with a slight sense of foreboding that if I want to understand even a little of what the authors have done, I’ll have to trawl through endless supplementary files. So, first things first, I really appreciate the slightly longer format of this article so that by the end of it, you have a sense of what was done, alongside the pitfalls and the potential implications. A real, whole paper; result!

And what a paper. My mind is still a bit fluffy on the spatial scale and connectedness of each plot (I was surprised there was no map figure of the interpolated richness per 1 degree cell), but that is a minor grumble. This is a massive, impressive collaborative effort to probe the distribution of trees in Amazonia. Naively, I was taken aback by a number of findings: that there are potentially >16,000 tree species (seems like a lot to me), that only 227 of them seem to dominant, that each hyper-dominant was kinda habitat-restricted. Conversely, I wasn’t surprised that species in different families were distributed differently (one or a few hyper-dominants vs. tons of restricted-range endemics) or that two well-chosen traits didn’t predict hyper-dominance.

It’s clear that such a big effort is going to generate tons of follow-on papers, many of which have been primed in this first article. There seemed to be some confusion whether to focus on the hyper-dominants and what made them so vs. the thousands of species with tiny or unknown distributions many of which the authors suggest are close to extinction. An interesting route by which the authors will follow up this paper will likely be to try to find out how important the non-dominant species are for ecosystem functioning. They seem to suggest that perhaps they are not so important. I wonder what the cut-off for usefulness vs. exciting rarity is? Does it vary among plant families? How much could we lose without having any impact at all? How much complementarity is there? How easy (and valid) will be to model ecosystem functioning or resource cycling concentrating only or mostly on the hyper-dominants?

I also found it funny that the authors did not wonder why their Maxent models predicted populations of many species in places where extensive surveys suggest they are not found. What is found in their place? Populations of phylogenetically or functionally related species? Or was some environmental or topographic variable missing? An easy (but perhaps quite dull) follow-on paper perhaps…

A final follow-on that I could think of would be to compare spatial patterns of some of the species’ patterns (a mix of hyper and non-hyper dominants) among regions and forest types (kinda like figure 4) to see if one can tease out any patterns of dispersal limitation. I think the authors conclude that many species are restricted to one or two forest types: are these generally within a region or across all or multiple regions (this might have been covered, but I missed it). If you are happy with the high levels of extrapolation (interpolation?) involved, this dataset is a treasure chest for dispersal ecologists…

So many options! And more or less freely-available online already (see link above). Let’s get going!

Contrasting changes in taxonomic, phylogenetic and functional diversity during a long-term succession: insights into assembly processes

Purschke, O., Schmid, B. C., Sykes, M. T., Poschlod, P., Michalski, S. G., Durka, W., Kühn, I., Winter, M., Prentice, H. C. (2013), Contrasting changes in taxonomic, phylogenetic and functional diversity during a long-term succession: insights into assembly processes. Journal of Ecology, 101: 857–866. doi: 10.1111/1365-2745.12098


Stora Alvaret on southeast of Öland with Eketorp Fortress in background (from Wikipedia)

HarisSL pic

Haris Saslis-Lagoudakis

Why this paper? I just saw it advertised on the facebook page of the International Biogeography Society by one of its authors. It’s not really my field, but the title seemed cool. Message to self (and all of us): be shameless to put your work out there.

What’s the paper about? The idea is pretty straightforward: after an ecosystem is disturbed, its biodiversity changes and goes through several steps before it reaches some sort of “equilibrium” again. This process is called succession. The paper I’m discussing today applies a comparative approach to determine whether succession is ruled by deterministic or stochastic processes, and whether these processes change during succession.

What did they do? Their study site is a 4.5 x 4.5 km landscape on the Baltic Island of Öland, Sweden. They assessed taxonomic, phylogenetic and functional (alpha and beta) diversity at different successional stages, as well as turnover between stages covering a more than 270-year-long succession from arable-to semi-natural grassland. Taxonomic diversity was measured based on presence-absence data from across plots. Functional diversity was measured from 11 functional traits associated with response to and/or tolerance of disturbance, scored for all plant species. Last but not least, phylogenetic diversity was calculated on a phylogenetic tree extracted from a published supertree of Central European vascular plant species. First, the authors quantified taxonomic, phylogenetic and functional diversity within grassland communities (alpha diversity), and turnover of these measures between communities (beta diversity), at four successional time steps. Grassland age (step) was assigned using a GIS overlay analysis of land-use maps from different times in the past, and they sampled 55 units for each successional stage. Then, they assessed whether, for each of the four successional time steps, species co-occurring within sites were phylogenetically or functionally more (or less) similar than expected, given the taxonomic diversity. Finally, they measured phylogenetic and functional turnover between successional stages.

What did they find? In summary, the main finding is that – perhaps unsurprisingly – the three aspects of biodiversity studied show different patterns during succession. During early and early-mid successional stages, species richness increases, in contrast with functional or phylogenetic diversity. Functional similarity between species within plots of early stages was higher than expected, a sign of abiotic filtering, which encourages co-occurrence of species with similar traits. During mid-late and late successional stages: Species richness does not increase further, but functional diversity within sites increases significantly, and closely related species are replaced by phylogenetically more distinct species. Between the earlier and late successional stages, functional turnover was higher than the within-stage turnover, suggesting that different environmental filtering processes govern community assembly at different successional stages. Throughout succession, species co-occurring within sites were functionally more similar than expected by chance, indicating that community assembly is deterministic with respect to species traits. However, functional turnover between stages was higher than predicted, and higher than within-stage turnover, indicating that different assembly processes act at different successional stages, possibly caused by differential environmental filtering, during the course of succession.

What do I think? I actually quite liked reading this paper, although I felt at times there were too many results for my brain to cope with. Nevertheless, I found their findings pretty cool. Nothing shook my world, but I like reading articles that convince me that what I imagined was true is actually true (hey, a scientist has so few chances for an ego stroke). One finding that stands out for me is that “phylogenetic turnover did not differ significantly from random expectations, either within or between successional stages, and provided no insights into the temporal dynamics of the processes underlying community assembly”. Phylogenetic diversity is often used as a proxy of functional diversity in this type of studies, often based on the assumption of trait conservatism: that closely related species are ecologically similar. As the authors suggest, the mismatch between functional and phylogenetic diversity could indicate two things. First, that ecological traits might not always be phylogenetically conserved. Second, that phylogenetic diversity represents a more inclusive measure of ecological similarity than measures of functional diversity based on a limited set of traits. I suppose I am slightly biased towards the phylogenetic approach, so it’s useful to know that other measures can provide better resolved results. If anything, I guess this brings to the surface the need for more interdisciplinary and integrative approaches.

What do you think?

Lynsey McInnes

Lynsey McInnes

I enjoyed this paper a lot. I was entertained by the idea of a chronosequence (sampling in the present day but classifying plots based on, in this case, grassland age). Neat! I’m sure this isn’t as good as having data at different timepoints for a single plot, but it surely must help in uncovering some facets of the temporal trajectory of grassland diversity.

The authors throw a whole suite of analyses at their chronosequences, looking at taxonomic, functional and phylogenetic diversity and turnover within and between timepoints for each facet of diversity. They find that each changes through time, but differently, that phylogeny doesn’t add much to the story, but function does, suggesting that there is a deterministic element to community assembly – you either have the right traits at the right time or you don’t. Seems reasonable.

Again, I’m no community assembly pro (though perhaps I should just succumb given how many community-y papers we cover here), but I appreciated the authors thorough approach here and I thought this was a good study on a limited spatial scale whose methods could get rolled out to a more macro scale with appropriate tweaks for less comprehensive trait data availability.

Some thoughts that occurred to me. Perhaps phylogenetic diversity came out as kinda uninformative in this context as the taxonomic scope was too broad…basically around 200 species widely distributed across the plant phylogeny. At this scale, can many traits show conservatism (there must have been rampant trait convergence, esp. if we are only looking at grassland-relevant traits)? Perhaps phylogeny should have been assessed at narrower taxonomic scales, e.g. within grasses, within eudicots, what have you… Then again, the gains made by doing this, might already be covered by the functional approach.

I didn’t look into this in detail, but I wasn’t sure what the authors were after with all the dispersal related traits? Were they included as a measure of species’ ability to get from “elsewhere” (the elusive regional species pool perhaps?) into the plots? So, nothing to do with crossing plots (now I think about it, probably not).

Was there any consideration of the proximity, location, layout of the plots? I guess this concern stems from using a chronosequence rather than an actual timeseries. The plots classified in different successional stages are limited in the diversity (all three kinds) given what came before. The authors might need to correct for within-stage variation in their interpretation of across-stage variation. I think.

But that was a pedantic grumble, I really do like the idea of chronosequences. I wonder in what other contexts this space for time substitution is useful (for instance, its been used a bunch in climate change related studies) and when does it really breakdown?

In conclusion, a surprisingly (for me) thought-provoking study. Thanks Haris!

Will Pearse

Will Pearse

I liked this paper, and it wins the Will Pearse “thank you for thinking about your methods” award because the methods are wonderfully executed, and give the authors exactly the answer they need. I just want to chime in (briefly!) about the same sentence that Haris picked up on:  “phylogenetic turnover… provided no insights into the temporal dynamics of the processes underlying community assembly”. I too am biased towards phylogenetic approaches, and I’ve actually been finding similar patterns (no change in phylogenetic structure over time, no significant phylogenetic structure, etc.) but I interpret them completely differently.

When I find that, despite changes in species composition, or changes in the functional trait structure of a community, that the phylogenetic structure of that community stays the same, it provides really quite profound insights into the community as a whole. Why is it that the trait that, apparently, determine community memberships are seemingly free from phylogenetic constraint? What does that say about how that pool of species, which have been evolving and interacting (potentially) for millions of years? These are the kinds of questions I want to be asking, because I’m really not interested in whether a particular trait shows phylogenetic signal or not, rather I’m interested in what divergence between traits and phylogeny can tell me.

When I find that phylogenetic turnover is minimal, but species (or taxonomic, if you like to call it that) turnover is quite strong, that suggests to me that close-relatives are replacing one-another. Does that mean, rather than being uninformative about ecological interactions, that focusing on species as our unit of measurement, is not giving us the most information about that system? I’ve been examining phylogenetic homogenisation of species recently, and it’s a very different process to species/taxonomic homogenisation – and figuring out why has helped me better-understand my urban ecosystems.

Spatial scaling of functional structure in bird and mammal assemblages

Belmaker et al. The American Naturalist 181(4): 464-478. DOI:10.1086/669906. Spatial scaling of functional structure in bird and mammal assemblages.

This is a guest post with Chris Trisos. 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!

Chris Trisos

Chris Trisos

Belmaker and Jetz tackle two major problems in community ecology. First, community assembly processes such as environmental filtering and competitive exclusion can leave either the same pattern or can act in opposing directions to leave a random pattern in community trait structure, neither of which can be interpreted as clear evidence for either assembly process. Second, how the relative importance of community assembly processes changes across spatial scales is still unclear.

The first problem is dealt with by delimiting an ecological species pool. This consists of the subset of species within the regional species pool that also fall within the trait volume of the local community, and thus can survive under the local environmental conditions. Now, the functional difference between the regional pool and the ecological pool can be used to test for environmental filtering and, because the signal of environmental filtering has been factored out, it should be easier to detect the influence of species interactions on community assembly using the functional difference between the ecological pool and the local community.

Belmaker and Jetz apply this approach to regions and local communities over a range of nested spatial scales from 400 to 3 million km2. The general expectation is that environmental filtering dominates at relatively large spatial scales and biotic interactions at smaller spatial scales. Interestingly, their results, at least for mammals, suggest that both environmental filtering and competitive exclusion operate across the entire range of spatial scales.

Now for some picking at the details. This paper is not the first to delimit ecological species pools, but it is novel in quantifying the functional difference between regional, ecological, and local species pools, as opposed to just the trait structure within each pool. The functional difference used here is the average nearest-neighbour distance in trait space between the species absent from the ecological or local species pool and those present in it (fig. 1). This is a pretty cool approach and certainly forces you to consider more explicitly what is different about the species that are absent when compared with focusing only within communities.

However, I am not as confident as the authors that high and low functional differences between species pools map as neatly as they suggest onto clustering and overdispersion in trait space within species pools. This is especially true for the link between low functional difference, trait overdispersion, and competitive exclusion in community assembly. One can easily imagine a case where the functional difference measure is very similar for two separate ecological pool to local community pairings, but where the pattern of trait overdispersion within the two local communities is very different. Fig. 2c and d gives an example of this. Because of this I would like to have seen a comparison of functional difference with a measure of trait overdispersion or regular spacing within local communities (e.g. standard deviation in neighbour distances). This is an important link to clarify given that a lot of theory on species interactions structuring local communities makes predictions about the overdispersion or clustering of trait structures within local communities as opposed to between communities and species pools. It would also be neat (if the dataset is available – maybe for plants?) to link functional difference measures to measures of species interaction strength. It might be that, for competitive exclusion, a given functional difference is associated with a higher strength or more asymmetrical interaction between pairings of present and absent species than between pairings where both species are present in the local community.

I’m also unsure about whether the standardized effect size of functional differences can be used to infer the importance of a given community assembly process, as is done in the paper to test for the importance of biotic interactions in the tropics (any thoughts?). A process that results in weak effect sizes could still have a very important role in structuring a community. For example, the potential for environmental change aside, a species that is only just unable to survive at a site due to local environmental conditions is perhaps no more or less environmentally filtered than a species with trait values far from the survival set for the local environmental conditions.

Will Pearse

Will Pearse

This is a cool paper that examines some fundamental aspects of ecology: using functional trait data to examine shifts in community structure. I don’t think it’s immediately obvious on a first-reading just how massive the bird trait database they’ve collected is – this is an incredibly useful resource, and this alone probably makes the paper worth reading.

I’m particularly interested in definitions of source pool, and so I like the authors’ attempt to understand what an ecological source pool actually is. My only concern is that they define their ecological pool by drawing a convex hull drawn around observed communities’ trait distances (top of p. 470); I think it’s going to be harder to detect deviation from the ecological pool on the basis of functional traits if you’ve used those traits to construct the ecological pool. However, the authors have explicitly defined their ecological pools in an intuitive way that’s readily applicable to other study systems, and they do detect some pattern, so perhaps I’m being unduly harsh. In passing, we’ve discussed the methods the authors use in a previous post.

It’s beyond the scope of this paper, but here’s something I’m always interested in: can we use these functional measures to detect certain kinds of community? Are there some examples of communities in this dataset where there’s no overlap in species between communities, but their functional composition is the same? In other words, can we find communities that are functional analogues of one another? Can we use these convex hull methods to partition communities into different sub-components, either to examine the functional dispersion within  them, or to see if these functional groups exist in other communities? These are big questions (I’m missing obvious references about them, right?) but they’re exactly the kinds of questions that papers like this make me want to explore.

Lynsey McInnes

Lynsey McInnes

Congrats for making it down to the third comment! Thanks Chris for choosing a great paper and contributing loads of interesting discussion points. Now, as a non-community ecologist, I feel a bit out on a limb here, but let’s see if I can contribute something useful too!

First, I did really enjoy this paper although I sometimes struggled to keep the various scales and metrics straight in my head (probably a function of me not using the associated figures to their full potential). I appreciated that they set up the reasoning behind the paper well and I bought in to most of their arguments for why this is an interesting question and why their methods are good for answering it. I second Will in being impressed with their trait dataset too!

Although not an innovation of these authors, I appreciated the simple distinction of ecological vs. regional pools – quickly and effectively removing one element of the question – environmental filtering. Makes you wonder why people didn’t do this (or some permutation of it) straight from the birth of community phylogenetics? I wonder how much richness of the local community matters to what breadth of things get caught in the ecological pool and whether this causes any problems with extracting functional difference between the two scales. I guess repeating the analyses at a whole host of spatial scales helped address this potential issue and looking at the effect of latitude helped untangle whether things are different in packed out tropical areas as opposed to spacious temperate ones.

If I understand correctly, the authors find that functional difference is generally lower than random between the ecological pool and the local community indicating that the local community is a kind of matching subset of species able to cope with the local environmental conditions, suggesting some kind of assembly rules/competitive exclusion is going on.  I swing between thinking this a really cool result and wondering what all the fuss is about, but I recognize that such a reaction is simply due to not being well versed in the decades of controversy surrounding these ideas.

My, more real, concern is that the authors purport to have gotten at process/mechanism in their analyses and it feels to me more like they have carried out an impressively robust and thorough analysis to show a set of results that they already kind of knew. That sounds unduly harsh – what I really want to say is – now comes the fun part! There are still some differences among communities and regions, and it would be exciting to know what leads to these differences: for example when are simple physical problems like dispersal limitation the cause of certain community compositions, when is it competitive exclusion? Is it intrinsic traits, extrinsic environment or stochastic events that underline such differences? It strikes me that the next step might be to incorporate a temporal perspective on the arrival of species (or rather traits) into a community and the effects this has, to incorporate finer-scale characterization of the landscape and environment (paleo conditions too, why not?) and, as Chris hinted at, get plant/producer information in there as well. That all sounds massively difficult and like a step away from finding generalities to finding specifics; I would counter that the authors have done a great job at clearing up longstanding issues in community assembly and their setup stands them in the best position to take the next step. I’d be interested to see how this goes.

*As a sidenote, one day prior to reading this paper I read a recent paper by Alex Pigot & Jo Tobias in Ecology Letters – Species interactions constrain geographic range expansion over evolutionary time – the authors take a temporal perspective on this question and it strikes me that some modification of their setup could be helpful in tackling more head on the temporal element of community assembly. Just a thought.

%d bloggers like this: