Intra-generic species richness and dispersal ability interact to determine geographic ranges of birds

Laube et al. 2012. Global Ecology and Biogeography 22(2): 223-232. DOI:10.1111/j.1466-8238.2012.00796.x.

Below, we give our first impressions of this article. Please comment below (or tweet, either at Will or with #pegejc, if you’re so-inclined) and then, after a week, we’ll put a brief summary of any discussions underneath the article. Think of this as a journal club discussion group!


Will Pearse

Will Pearse

I was simultaneously elated and annoyed when this paper came out, because I’ve been wanting to do something like this for some time, and these guys have done it much more nicely than I could have done. I’m going to focus on what I see as two major issues with species distribution modelling (SDMing) that they’re tackled: dispersal ability and biotic interactions. I should point out that I’m not a species distribution modeller!

Biotic interactions are frequently ignored because they’re hard to quantify, and I think measures of phylogenetic similarity could help with that. The authors have good a priori reasons to think that close relatives compete more strongly, and I think they find pretty strong evidence that close relatives (congeners) are forcing each other out of areas. While we have to be careful when using phylogenetic structure (I read that paper before it was cool, honest), I think it’s helpful to start asking whether phylogenetic structure can improve the fit of SDMs, and go from there. Maybe we can use the phylogenetic dispersion of the species within a region to help predict whether another a particular species will colonise that region. Yes, we could start mechanistic right away (what traits do species have, and do they bring them into competition), but it might be simpler to start with a phylogenetic approach that can be applied to any kind of species.

Dispersal ability is another key problems for SDMs, and I particularly like the way the authors seem to be trying to understand differences between poor and strong dispersing species. I’ve been to way too many talks where modellers mention that SDMs need to account for dispersal ability, then just plough on and ignore it regardless. We might expect strong dispersers to be brought into competition more readily, and the authors have actually used that to their advantage here.

I have to be balanced, and so I would caution that projecting back in time thousands of years, using projected environmental data, is a pretty risky business with a lot of hidden assumptions. However, it should be harder to pick up the biotic signal given all the climate changes, so I think the bias is in the right direction (i.e., it should be biasing them away from detection these patterns).


Lynsey McInnes

Lynsey McInnes

I was instantly drawn to this paper after remembering lead author’s talk on the same subject at the International Biogeography Society’s student conference in Oxford last year. Then, as now, I found the analysis thoughtful and well-balanced and a really strong attempt to actually address two elements of biogeographical analyses that typically get brushed under the carpet. Countless authors (and I count myself among them), call up dispersal ability and biotic interactions as potential factors influencing species’ distributions only to ignore them due to lack of data or methods to handle these factors. Laube and co-authors grab the bull by the horns and set up an elegant way to see whether they can untangle the influences of dispersal ability (using a clever (at least to a non-bird person) morphological proxy) and biotic interactions (quantified by congeneric richness) on a species’ ability to fill its potential range using an amazing dataset on the Sylvia warblers. The results are also intuitively satisfying: ‘good’ dispersers fill more of their potential range but only if it’s not full of congeners (interpreted as competitors). Neat.

Pessimistic me always wants to jump at a bunch of what ifs to negate results like these that are so intuitively appealing, but, in this instance, I’m pretty convinced the authors have thought of most of the eventualities and have covered their backs. The consistency observed from the various analytical techniques helps in this regard.

Although the authors found no consistent effect of taxon age on their results, I found myself wondering if the dataset and/or results could be twisted around to reveal insights on the spatial and temporal trajectories of diversification of the genus. For instance, why is the congener able to occupy chunks of the focal species’ potential range? Was that lineage there first? Is it just a better disperser? Does it have slightly different niche requirements so that that area is more suitable for it? That intra-generic richness had the greatest effect in the least suitable habitat suggests this might be case.

One could also ramp up the scale of analysis to look at clade filling? What parts of the genus’ range remains unfilled? Why? What about the effects of competitors from other genera or trophic levels?

Also, what was the effect/patterns of congeners within the realised range of each species?

The authors end with a note of caution: if competition is important in setting range limits, predicting species’ range responses to climate change has an added dimension in that we really should account for how related species in neighbouring areas might hinder establishment of range-shifted populations (but these related species might themselves be moving out). Wah – head mangled by thoughts of a jigsaw of ranges being pushed and pulled across the continents. Needless to say, this paper emphasizes that SDMs and GCMs are not all that is needed if we want to figure out how species are going to redistribute. On a more cynical note, how different are these warbler species, would it matter if we lost a couple!?!

I liked this paper a lot; it was a thoughtful treatment of some issues that are so often acknowledged and promptly ignored. It’s thrown up (for me) more questions on the patterning of closely-related species’ ranges in space and I’m looking forward to seeing more papers that are as elegant as this.

Advertisements

About will.pearse
Ecology / evolutionary biologist

9 Responses to Intra-generic species richness and dispersal ability interact to determine geographic ranges of birds

  1. I pretty much agree with Will and Lynsey’s views on this article. I like it, and my criticisms are really suggestions for further research, unavoidable problems with a study of this nature, or issues that bias AGAINST detection of these patterns. So while what I’ve written looks rather critical, remember that I do actually really like the paper!

    It would be interesting to have more discussion on what we mean by dispersal ability. Do we want to know whether a species can fly a long way, whether they have a strong tendency to return to the location where they were raised, whether they can cross unsuitable matrix habitats, whether the habitats they use are well connected so they don’t have to cross unsuitable matrix habitats, etc? The authors look at one facet of this. It would be interesting to see what would happen if the connectivity of suitable areas was also included in the analysis.

    The authors incorporate bill depths into their proxy for dispersal ability, arguing that shallow bills are related to migratory tendency. Is this really the case within Sylvia warblers (e.g. shallow billed resident Mamora’s warblers and thick billed trans-Saharan migratory barred warblers)? Migratory tendency is known for Sylvia warblers, so why use a proxy? Anyway, this is a small issue, and I think resultant errors in the estimation of dispersal ability would bias against them finding the patterns they did (??)

    A coarse scale analysis will always be plagued by the problem of habitat data availability. The authors do the best job with what they’ve got; most Sylvia warblers like scrub, so incorporating two different categories of scrub into their SDMs is a good thing. However, two categories does not do justice to the complicated beast that is scrub, so much of intra-generic differences in habitat preferences will not be captured in the SDMs.

    The authors did not attempt to account for dispersal limitation or species interactions when fitting SDMs. I would have liked the implications of this to have been discussed more, but as Will said, this would make it harder to detect the patterns that they observed, so in the context of their paper is not a problem.

    Putting my field ornithologist hat on, I’m intrigued to know more about the mechanism of competition between Sylvia warblers. I’ve seen three species holding territories in an area of a few 100m squared, but in different microhabitats. Do fine scale niche differences allow coexistence? The paper alludes to this in the discussion, and this is supported by a study looking at habitat associations of the Sardinian warbler as it colonised Cyprus (http://onlinelibrary.wiley.com/doi/10.1111/j.1474-919X.2011.01211.x/abstract); it used different scrub types to the endemic Cyprus warbler, so the authors concluded they were not competing. But how stable is this? Would they compete in simpler/ poorer quality habitats?

    Regarding using congeneric species richness as a proxy for biotic interactions, a picky ornithologist could start asking whether a blackcap (genus Sylvia) competes mores strongly with a chiffchaff (genus Phyloscopus) than a whitethroat (genus Sylvia), but I won’t! Remember, I really like this paper!

    • willpearse says:

      Nice to have an ornthiologist’s take on the paper! I admit I missed the bill size aspect – not being a birder, I can’t comment – I think dispersal ability traits like this are likely to be driven by knowledge of the particular study system. I find it hard to believe that bill length is correlated with dispersal in all birds, but perhaps there is something in this particular genus. I don’t know!

      I think it’s quite likely there would be more competition in poorer quality/less complex habitat. That said, I wonder if you’d get more interspecific flocks in difficult conditions as everyone’s trying to forage in the same place. Again, I know nothing of birds, so I can’t tell!

    • lynsey83 says:

      Hey Martin. Thanks for your comments! I second Will that it’s great to get comments from someone that owns a field ornithologist hat. The restriction of all competitors also being Sylvias had been bugging me (I can totally see the pragmatism behind the author’s choice) and it was good to hear from you that there might be non-congeners that are more important. Sadly though, that this is most probably the case makes generating robust SDMs & range projection models even more problematic. Oh dear. One step at a time, I guess.

    • yaelsnail says:

      Apart from agreeing with previous comments, I really REALLY agree that it would be great to discuss more what is meant by “dispersal ability” and to consider that different aspects/types of “dispersal” might have different effects. Well, this is unsurprising since my current research is all about comparing different measures of dispersal and then seeing which is most important in structuring macro-diversity patterns (in plants). Anyway I am very happy to hear this coming from someone else as well! Especially in birds, I think the distinctions between different components/aspects of dispersal are too often glossed over. Is migratoriness (is that a word?) actually related to the likelihood that a bird species will colonize a new region? Are the distances that birds tend to travel over one day/from birth to reproduction/during migration correlated with each other? And how often do birds get transported accidentally by wind and storms?

  2. yaelsnail says:

    First of all, thanks Lynsey and Will for starting this online journal club! I am really excited (genuinely, not just because I’m American ;-)) about joining in – I’m confident you’ll pick good papers and I think the practice of writing about them will be great.

    I think this paper was an excellent start and I’m glad to have had it brought to my attention. As a beginner macroecologist very interested in dispersal and range size, it gave me some very good new ideas and perspectives. And I appreciated the relative simplicity of the author’s clever approach.

    The main points I took away:

    1. Competition can be important at large, i.e. continental scales too, not just locally.
    I love this kind of discussion about scales, especially thinking about how processes that work directly at one scale can affect patterns at other scales. This also reminded me of a recent talk here by Brody Sandel (working in Aarhus) about the importance of spatial scale for ecological patterns (especially for studying them). There was a question afterward about which scales ecological processes actually function at (i.e. is competition really only a local-scale process?) … I think this paper is a strong indication that we need to think about scales carefully and not jump to conclusions re: at which scales do different processes have an effect.

    2. Is it best to think of range size as the product of (total area of suitable habitat) and (range filling)?
    I suppose this is probably a common idea in literature about species’ distribution ranges, but as an evolutionary biologist just pretending to be a macroecologist, I’m not sure. Anyway, it’s not how I’ve thought about range size in the past and I think it will be a useful framework to hang my thoughts on in future.

    3. These results suggest to me that a good way to think about range size is that species have large ranges when “all the stars align” and all the right conditions exist at once, while small range sizes happen when any of the needed conditions are missing.
    Again, maybe this is already a common idea, but I had been thinking about it in rather the opposite way, i.e. small ranged species are special and big ranged species are more normal (why? dunno. just one of those unexamined unconscious assumptions I think). But from this paper it seems to me that species will only have large ranges if: they have a big potential range, competition is low in the potential range, they are good enough dispersers to get everywhere in their potential range (relative to the barriers in the range), and they’ve had enough time to get everywhere.

    4. It was interested by the authors’ finding of no effect of species age or habitat shifts since the Last Glacial Maximum, and their discussion of the importance of dispersal vs. historical constraints/barriers, especially comparing birds to Proteas, mammals, and trees.
    Indeed! Are barriers only important to “rubbish dispersers” that don’t get around as well as birds? Are there qualitative differences between organisms in different classes of “dispersal ability” (however you define that)?? i.e. Is competition primarily important in setting ranges of great dispersers, but environmental barriers primarily important for setting ranges of poor dispersers?

    Some questions:

    1. Should I switch from studying plants to studying groups with better data, like birds or mammals? Just kidding. Kind of.

    2. Is there an alternative to using SDMs for their questions? I have to ask because I’m generally quite skeptical of SDMs, although I understand their appeal.
    I started thinking, what if they had just compared the areas where each species occurs, the areas where it doesn’t occur and no other Sylvia species occurs, and the area where it doesn’t occur but other Sylvia species do. Could evaluate differences in environment, number of congeners, etc. I guess this kind of analysis wouldn’t have anything to do with range filling anymore, might not be as direct, but wouldn’t depend on the same assumptions …

    3. This is a small point, but the authors mentioned in the results that there was no phylogenetic signal in the residuals. Can someone tell me what this means (in this case)? Currently blanking.

    4. Is there any alternative explanation (other than competition) for the link between intra-generic species richness and range filling? If not, I don’t see why there should be any doubt that intra-generic species richness is a good way of measuring potential competition.

    One last comment:
    I checked on Dawideit et al. 2009, the paper that originally suggested estimating bird dispersal ability using a combination of Kipp’s distance and bill length. In their study, these two factors could only explain 40% of the variation in natal dispersal distance of British birds.
    Is this good enough?
    I mean, 40% is awesome if you don’t have any other data, if that’s the best you can do. Especially for such relatively easily measurable traits! But, because I see dispersal as a super-key trait for understanding patterns in both ecology and evolution, and especially as birds are so well studied and well known, I would like to see more work on finding proxies to predict bird dispersal ability, and on understanding bird dispersal in more detail!

    Thanks again Lynsey and Will. Next time I will try to write less at length =)
    Yael

    • willpearse says:

      Thanks for this Yael! I’m particularly struck by your comment about competition being something that happens across continental scales – it had never occurred to me to think of this as odd, but you’re right, it’s kind of cool we can detect these processes. Kind of makes me wonder if we could detect facilitation as well, but perhaps that’s for another day!

      I’m not quite sure what an alternative to SDMs would even look like. SDM to me is a peculiar term, because it relates to anything we use to understand species’ ranges, right? So anything we tried would inevitable become an SDM. Maybe that’s too semantic an answer… Anyway, I have no idea what an alternative to SDMs might be – maybe someone else does?

      Regarding your question as to what phylogenetic signal in the residuals means in this case – I’m ashamed to admit I don’t know. They use a measure (‘A’) that I hadn’t actually heard of, so I’m reading about it now. The paper is by Pavoine et al. (2008) in Theoretical Population Biology, entitled “Testing for phylogenetic signal in phenotypic traits: New matrices of phylogenetic proximities”. I’m sure I shouldn’t admit my ignorance in print, but there we go!

      • yaelsnail says:

        Oo, look Will, we are having a science conversation!

        It seems to me like facilitation should be detectable in the same way – if you have an idea for which species might have a positive interaction with your species of interest. Then the question to test would be, is your species more likely to occur in areas where the likely facilitator occurs too? For these birds, the easiest facilitative relationship to pick up might be with the tree species that make up their preferred habitat. But this would not be very ground-breaking!

        As far as alternatives to SDMs, I guess I just meant: What if we try to address this question without making any calculated predictions about what the Sylvia species’ “potential ranges” are?

        And regarding admitting ignorance in print … I personally think science would be a lot more welcoming and less stressful if people did it more! So thank you for admitting you didn’t know about the phylogenetic signal in the residuals either =)

  3. willpearse says:

    I’m always in favour of reducing stress 😀 I agree with you about facilitation – I can’t think of anyone who’s done that though. Interesting.

    I’m not quite sure what an alternative to predicting where things’ potential ranges might be. I guess something exploring relative abundance within a range would be a useful approach, but I guess that’s just another area of SDMs. Admittedly it’s hard to deal with abundance data because of sampling issues, but then the same applies to presence/absence data!

  4. Pingback: Niche incumbency, dispersal limitation and climate shape geographical distributions in a species-rich island adaptive radiation | PEGE Journal Club

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: