Mammal predator and prey species richness are strongly linked at macroscales

Sandom et al.. Ecology 94: 1112-1122. DOI:10.1890/12-1342.1. Mammal predator and prey species richness are strongly linked at macroscales


Mammals, mammals, everywhere, where there’s plenty of prey to eat. Taken from Sandom et al

Lynsey McInnes

Lynsey McInnes

Will picked this paper out of a choice of five I gave him as he wanted to diversify away from community phylogenetics. For me, I liked the safe return to pure macroecology and I liked the similarity in approach to a recent paper I published looking at whether global diversity patterns of vertebrates reflect those of monocots? (Blatant self-promotion). I also used structural equation modelling in that paper (requested by a reviewer, I will admit) but remain dubious over their power to extract relative strengths of direct and indirect effects (more on this in a second).So, Sandom et al. look to see whether there is evidence of top down or bottom up effects of predator richness on prey richness (and vice versa) using mammals as their study taxon. They find a significant effect of prey richness on predators but little evidence for an effect in the other direction. This is a nice result and I was quite surprised by the strength of the signal. Like Jetz et al.’s, my own paper and a few others, at these macro-scales there has only been limited evidence for biotic effects on diversity patterns – collinear diversity apparently being mostly the result of similar responses to environmental gradients.  And therein lies the problem – when can we ever know we have included the right variables in our model to cover the myriad environmental effects such that the variable ‘prey richness’ is not just filling into for some omitted environmental variable? This is macrocological madness strike 2 (strike 1 was when ‘latitude’ always came out as a significant predictor of richness gradients as it stood in for some immeasurable combination of abiotic variables). OK, while there was no mechanistic explanation for ‘latitude’ explaining richness patterns and there are strong mechanistic explanations for an expected association among trophic levels, I am still dubious…I have another couple of methodological worries, although I totally admit I don’t see any easy fix for them. My first worry is that there are so few predator species (125) vs. prey (3966) meaning that grid cell richness has a much lower range and maximum for predators (max: 19) than prey (186). I feel uncomfortable treating these groupings as equivalent, presumably the spatial autocorrelation among grid cells in terms of species present must extend over longer distances for predators than prey? Maybe this is not strictly an issue with the questions being asked here and may only be a helpful explanatory reason why there was not much evidence for a top down control of predator richness on prey richness? I also had a similar problem with my own data on monocots vs. vertebrates (there are a magnitude more monocot species than vertebrates…).

Two more queries: why such big grid cells? Why not 100km x 100km like most other macroecological studies on mammals so far? Why PCA climatic variables rather than make informed choices on the variables you want to include (I don’t think PCA is strictly a verb, sorry). This is, admittedly, a pet-hate of mine, but it just strikes me as an unnecessary extra step to remove the reader from the data. In this instance, I also wonder how each axis changed from region to region in the region-specific SEMs?

Finally, will the diet database be made openly accessible? I really hope so, it sounds like a great resource for a bunch of further questions. For instance I would really have just liked to have seen more ‘basic’ analyses of the data: average range sizes for the two groups? Body sizes?

That was a really grumbley post and very unfair given each and every point could have been applied to my own work. I suppose one is always harshest on what one knows best…I found this paper interesting to read, thoughtful on the different relationships possible and a great attempt to incorporate biotic effects into macroecological studies. More please!

Will Pearse

Will Pearse

This is an extremely far-reaching paper; the authors link predator and prey diversity at the macro-scale, and show that there are more predator species where there are more prey species.

My initial nit-picking niggle was wondering what exactly we can infer about such an inherently local-scale process as predation at the macro-scale – what relevance do predators and prey hundreds of kilometers away from one-another have? However, I think this link has got to be real – the authors control for spatial auto-correlation, they control for habitat (in the best way we can), and they still find this relationship. In fact, they find differences among predators and preys in their dependence on environmental conditions, which makes me think they’re picking up something real.

Maybe predators specialise on particular prey species, and so prey diversity begets predator diversity. We kind of know that to be the case, although we also know some species are generalists. An alternative, that I find more interesting, is that maybe (at a coarse scale) having more prey species makes a system more robust, and so able to support a wider variety of predators. Indeed, if prey species are more dependent on environmental factors than predators (as this paper finds), maybe having a wider variety of prey species gives the predators a more reliable food supply, and so a greater diversity of species can be supported. This is essentially a species-level re-working of the insurance hypothesis in the ecosystem function literature.

Which, essentially, brings me right back to the nit-picking question I had to begin with. Why is it that every explanation I think of for these patterns involves local-scale patterns, when this is a global-scale pattern? Can my thinking scale up to two degree grid cells? I’d now like to see how we can relate these coarser grid cell patterns to what’s going on inside the grid cells. What happens inside a grid cell that has an unusually high diversity of predators or prey?


About will.pearse
Ecology / evolutionary biologist

4 Responses to Mammal predator and prey species richness are strongly linked at macroscales

  1. @bunnefeld says:

    Interesting….but I want to know now…can we use this for conservation science. Does this mean we can use the map and decide where we should reintroduce prey and predators? Obviously, I don’t know anything about macroecology, so under what assumptions could we use this for decision making in conservation?

  2. will.pearse says:

    Yeah… I don’t know. I guess you could say a cell with more predators than we would expect would be ‘saturated’, but I’m not quite sure how much you can infer about the likelihood of success of an introduction, given that we’re already operating at a spatial scale that’s too high to incorporate community-level dynamics. So, in short, I’d like to know too!

  3. Camilla Fløjgaard says:

    Hi, our data is now published in Ecology and Evolution:
    Ecological trait data are of often lacking for species-rich clades. We have compiled and evaluated diet data for all terrestrial mammals in a unique dataset and introduce a new method for dealing with missing trait data. The approach used here is transferable to other trait data sets and taxa. Since trait data are essential for understanding the ecology, evolution, large-scale diversity patterns , the data is naturally made freely available to enable macroecological analyses.
    Hope it will be useful to all macroecologists out there:-)
    Best regards, Camilla Fløjgaard

Leave a Reply

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

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

Google photo

You are commenting using your Google 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 )

Connecting to %s

%d bloggers like this: