The Molecular Ecologist: Using R to model the spatial distributions of species

Environmental variation across the range of Joshua tree. Image via The Molecular Ecologist.

This week at The Molecular Ecologist, I’m showing how to use the popular open-source statistical programming language R to estimate species distribution models.

Species distribution models (SDMs) are handy any time you want to extrapolate where a species might be based on where you know it actually is. Maybe you’re trying to figure out where would be fruitful to do more sampling; maybe you want to know where your favorite critters probably lived back during the last ice age; maybe you want to know what regions will be suitable for your favorite critters after another century of global climate change.

Given how widely useful SDMs are, it’s very nice to be able to estimate them using multiple methods implemented within a single open-source framework. To get a taste of the capabilities provided by R and a select set of add-on packages, go read the whole thing.◼

Here you see it, there you don’t: What determines species distributions?

ResearchBlogging.orgFor a year between undergrad and graduate school, I interned with the ecologists at the Western Pennsylvania Conservancy. A fair bit of the work revolved around documenting the locations of rare plants or animals. But “rare” can be relative: common ravens (for instance) were hard to find in Pennsylvania, but much more abundant just over the border in New York. In Pennsylvania, striped maples are so abundant that they can interfere with the regrowth of logged woodland, but in Ohio they’re rare enough to be considered endangered.


Striped maple: rare in Ohio, a pest in
Pennsylvania.

Photo by homeredwardprice.

This, of course, is because Pennsylvania’s human-drawn political boundaries straddle important ecological boundaries, like the transition from raven-friendly Circumboreal forests of New England to the raven-free forests to the south. And you might imagine that, even if there weren’t major topographic features within a particular set of human-created boundaries, there would still be enough climatic changes from one side to the other so that the species present at the southernmost edge of a flat, boring state like Kansas wouldn’t be entirely the same species present at the northernmost edge. That is, if you travel south to north across Kansas, you’ll probably pass a point where the lowest winter temperature becomes too low for some species.

Still, there are probably fewer species whose ranges end within the borders of Kansas than within Pennsylvania. As a paper in the latest Proceedings of the Royal Society attempts to show, this is for the intuitive reason I’ve tried to describe above: while Kansas is comparatively uniform, Pennsylvania is located at a transition between ecological regions. At such transitions, landscapes get complicated — and that complication, the paper’s authors say, is what helps create the boundaries of species distributions [$-a].

To make their point McInnes et al. examine the distribution of bird species across Africa. The broke the continent up into a grid, and scored each grid cell by the proportion of birds present in the square whose ranges had an edge within the cell, a measure they call “impermeability.” The impermeability of a given cell was strongly related to the number of habitat types represented in the cell — and these heterogeneous cells tended to be distributed along the boundaries of major ecological regions, like the Sahara Desert.

That is, species ranges tend to end in areas where desert is shading into grassland, or savanna into forest. Or, to put it another way, one group of species (birds) tend to have range edges that coincide with the boundaries of whole ecological communities. That’s an important observation — but it’s not exactly new. The entire concept of ecological communities arose because it’s obvious there are groups of living things that tend to occur together. McInnes et al. have quantified that observation, but they haven’t really explained it.

References

McInnes, L., Purvis, A., & Orme, C. (2009). Where do species’ geographic ranges stop and why? Landscape impermeability and the Afrotropical avifauna Proceedings of the Royal Society B: Biological Sciences, 276 (1670), 3063-70 DOI: 10.1098/rspb.2009.0656