Archive for the ‘species distribution & niche modeling’ Category

The decimation of Madagascar’s rainforest habitat

Monday, December 23rd, 2019

It is honestly with sadness that I announce our new publication on the fate of Madagscar’s rainforest habitat in Nature Climate Change. Modeling deforestation assuming the lowest rate of deforestation across the period 2000-2014, I could only get the rainforest to last to the 2070s… and the highest rate of loss occurred in 2018, outside the time period over which I had data. The slight hope is that protected areas are deforested at a slower rate, and if were to (unrealistically) assume no new deforestation in these areas, then some rainforest habitat would remain.

Morelli*, T.L., Smith*, A.B., Mancini, A.N., Balko, E. A., Borgenson, C., Dolch, R., Farris, Z., Federman, S., Golden, C.D., Holmes, S., Irwin, M., Jacobs, R.L., Johnson, S., King, T., Lehman, S., Louis, E.E. Jr., Murphy, A., Randriahaingo, H.N.T., Lucien, Randriannarimanana, H.L.L., Ratsimbazafy, J., Razafindratsima, O.H., and Baden, A.L. 2019. The fate of Madagascar’s rainforest habitat. Nature Climate Change 10:89-96. * Equal contribution. (article | “behind the paper” | Washington Post | National Geographic | The Conversation | ScienceDaily)

Welcome to our new postdoc, Kelley Erickson!

Monday, September 17th, 2018

Kelley EricksonWe are excited to serve as the new intellectual home of Kelley Erickson, a recent graduate of the University of Miami where she studied the demography of the highly invasive shrub Schinus terebinthifolia (Brazilian peppertree). Kelley is working on incorporating issues related to detectability in species distribution models in a project sponsored by the Institute for Museum and Library Services.  Welcome, Kelley!

NSF Advances in Biological Informatics

Friday, March 2nd, 2018

Awesome news! We were just informed that the National Science Foundation will fund our proposal to use pollen, genetic, and distributional data to estimate the spatial dynamics of how trees migrated poleward after the last glacial maximum.  This is a collaborative project with Sean Hoban (Morton Arboretum), Andria Dawson (Mount Royal University), John Robinson (Michigan State University), and Allan Strand (College of Charleston). We will be hiring two postdocs over the 3 years of the grant. The first position will be based at The Morton Arboretum near Chicago, Illinois and the second at Michigan State University.


Phenotypic distribution modeling

Tuesday, November 14th, 2017

Our latest paper in Global Change Biology on modeling intraspecific phenotypic variation has gotten great press!  Combined, the news outlets covering our research reach ~78 million people and included The San Francisco Chronicle, The Seattle Times, US News and World Report, The Topeka Capital Journal, The Manhattan Mercury, and numerous other regional newspapers, radio stations (e.g., KWMU 90.7), TV stations (e.g., KWCH12), and science news websites (e.g., Science News Online)!

Smith, A.B., Alsdurf, J., Knapp, M. and Johnson, L.C.  2017.  Phenotypic distribution models corroborate species distribution models: A shift in the role and prevalence of a dominant prairie grass in response to climate change.  Global Change Biology 23:4365-4375. doi: 10.1111/gcb.13666

Change in biomass of Andropogon gerardii due to climate change

Change in biomass of Andropogon gerardii due to climate change

Climate paths and climate change communication

Tuesday, February 28th, 2017

Climate path of St. Louis

Climate path of St. Louis, Missouri, USA

How can we communicate global warming to local audiences (= everybody who lives in a place)? Recently I made a poster showing the locations that climatically currently resemble the future climate of St. Louis.

But how did I know where to locate the “future” St. Louis climatically?  By running species distribution models in “reverse”.  First, I created a set of 100 points to represent St. Louis (I actually had them have the exact same coordinates–it’s false sample size inflation, but it doesn’t matter much since at first approximation St. Louis is a point–and it does allow me to use more complex fitting features in Maxent).

Second, I associated these points with the climate layers I have for the 2070s (once for each emissions scenario).

I then trained a Maxent model using this future climate data, then projected it back to the present.

Finally, I calculated the geographic center of gravity of all cells using the predicted suitability as weights.  The center gravity is the average location of the “future” climate of St. Louis!  I found I got slightly better (= intuitive) results by thresholding first, then using suitability values above the threshold as weights. I also found I got better results when using only mean annual temperature and precipitation, rather than all 19 WORLCLIM variables.

This procedure is fairly simple and takes advantage of the fact that 1) “species” distribution models are not just for species, and 2) the output of a SDM (or whatever you want to call them) is really just an index of similarity of a multivariate space (= climate layers) at a set of points (= presences) and another set of points (= all grid cells in the layer to which you’re projecting).

I’ll be trying the poster out at the Missouri Botanical Garden’s upcoming Science Open House–hopefully it will spark some conversation!

Species distribution models not for species

Wednesday, July 20th, 2016

SDMs - Not just for species

Mathematically these are all the same.

Have you ever see the number of people who drowned by falling into a swimming pool–films starring Nicholas Cage model?  You might also know it as “linear regression.”  Have you ever seen a species distribution model?  By calling it thus we make the same limiting semantic complexification as in the first case.

This post is not about the debate over whether we should be calling it species distribution modeling or ecological niche modeling (notice the participle form of each term–adding an ing refers to the act of modeling).  I’m talking about calling them species distribution models.

To put it bluntly, the underlying mathematics of a SDM doesn’t care whether it’s depicting a species or anything else.  In fact there are numerous examples of applications of species-less “SDMs”:

I even once met a person who uses Maxent to locate opportune places for underwater archaeology!

The fundamental commonality that allows all of these phenomena to be modeled by “SDM” algorithms is the nature of the response data–either unary (i.e., just presences) or binary (presences and absences).  Ergo, if you can describe a pattern with unary or binary data, you can also likely apply an “SDM”.

So feel free to refer to “my species distribution model” to reference your particular model of a species’ distribution.  But don’t let the moniker box you into thinking they’re just for species!

The fine limits of climate data

Wednesday, July 13th, 2016

Modeling Geum radiatum was very challenging compared to modeling I’ve done before because we know the species specializes on small habitats with microclimates.  And these particular microhabitats are are not well reflected by the coarse-scale climate data that is available.  I used ClimateNA for the basic climate layers. ClimateNA purports to be “scale free” but actually isn’t (and its creators acknowledge this).  Indeed, I found that the relative humidity measured by Eric’s team was far higher than the values predicted by ClimateNA for the same years and locations.  Ergo, ClimateNA (and other similar climate data sets) cannot really reflect microclimate on the order of of 100s of meters or less.

So how can one capture microclimate when one doesn’t have microclimate rasters? One option is to use microtopography, fine-scale measures of slope and aspect that hopefully help create microclimatic conditions relevant to the species.  This gives us the “topoclimate” of Guitiérrez Illán  et al. (2010).  So, I coupled fine-scale topography data at 90-meter resolution with climate data from ClimateNA at 1-km resolution.

Now the normal way to combine data sets of differing resolutions would be to resample the coarser-scaled rasters so they have the same resolution as the fine-scaled rasters.  But I felt this would create an illusion of precision in how well we know climate at fine scales.

So instead I wrote an R script to extract values from each data set in its native resolution and fed that to the predict() function.  Essentially, the script goes cell-by-cell across the fine-scaled topography raster stack and extracts the matching coarse-scaled data from the climate raster stack.  Then it writes a fine-scaled prediction raster from the ENM.

Geum radiatum (zoomed prediction)

Geum radiatum (zoomed prediction)

The output doesn’t look especially blocky when you see it in the publication (Ulrey et al. 2016), but if you look carefully at the zoom-ins you can indeed see the effect of matching coarse- and fine-scaled data sets. The larger reddish cells reflect appropriate macroclimatic conditions (i.e., at 1000-m scales), while the smaller dark red areas indicate appropriate macro- and topoclimatic conditions.

I admit the map isn’t visually pleasing–it’s obviously an artifact that the predictions are blocky.  But by keeping the data sets at their native resolution I believe the output better reflects the limit of our knowledge about the predictors, and in turn, the geographic realization of Geum‘s microclimatic niche.


Guitiérrez Illán, J., Gutiérrez, D., and Wilson, R.J.  2010.  The contributions of topoclimate and land cover to species distributions and abundance: Fine-resolution tests for a mountain butterfly fauna.  Global Ecology and Biogeography 19:159-173. (article page)

Ulrey, C., Quiantana-Ascencio, P.F., Kauffman, G., Smith, A.B., and Menges, E.S.  2016.  Life at the top: Long-term demography, microclimatic refugia, and responses to climate change for a high-elevation southern Appalachian endemic plant.  Biological Conservation 200:80-92. (article page)


Friday, July 8th, 2016

CliffhangerI think I was on a long-haul flight across the Pacific when I succumbed to jet-lag induced doldrums and watched Sylvester Stallone’s Cliffhanger which stars him (surprise) as a mountaineer who gets himself out of a dastardly plot by climbing around and flexing his muscles. So if there’s a Rocky of the rare plant world, it’s Appalachian avens, or Geum radiatum.

Like Stallone, Geum likes to hang on cliffs in the Southern Appalachians. And talk about hang!  Eric Menges sent me a few photos of their field work–they use ladders to census the populations.  Now there are also a few populations found on so-called grassy balds (I am fond of this name), but the cliffside populations tend to be more common.  And these cliffs occur at high elevations where the distinction between cloud and mist dissipates.

Geum radiatum

Geum radiatum (Wikimedia)

And like Stallone Geum is threatened–so much so it’s listed on the US Endangered Species Act.  In fact Pedro Quintana-Ascencio‘s models predict that the overall growth rate of the populations for which they have census data is currently below replacement level.  But–intriguingly–growth rate is highly positively linked to relative humidity.

Which is really cool in a technical way–because I was able to extract coarse-scale humidity data from the ClimateNA dataset, then relate this to their fine-scale measurements.  And this in turn allowed Pedro to predict population growth rates under future scenarios of climate change.


“Hang in there, Geum!”

Alas, Geum needs a strongman like Stallone… even though we predicted relative humidity will drop by just a few percent, that’s enough to exacerbate the species’ current downward trajectory.  It’s a slow fall, but it’s still a fall.



Ulrey, C., Quiantana-Ascencio, P.F., Kauffman, G., Smith, A.B., and Menges, E.S.  2016.  Life at the top: Long-term demography, microclimatic refugia, and responses to climate change for a high-elevation southern Appalachian endemic plant.  Biological Conservation 200:80-92. (article page)