Modelling Climate Suitability for Ephedra viridis in Space and Time
Faculty Mentor: Richard Gill, Biology
Climate change is expected to decrease soil moisture in the sensitive, water-limited ecosystems of America’s southwestern deserts, leading to shifts in plant distributions and altering ecosystem function. Studies have already documented the loss of desert grasses and the expansion of desert shrubs on the Colorado Plateau near Canyonlands National Park.i The objective of this work is to identify the climatic and biophysical factors influencing the distribution of an evergreen shrub, Ephedra viridis, and its expansion in plot-scale studies over the past 30 years.ii A gymnosperm and member of the plant division Gnetophyta, this shrub is unique because of its photosynthetic stems and scale-like leaves, contributing to drought tolerance.iii These traits, coupled with its year-round activity, may be giving E. viridis an advantage in the increasing warmth and aridity of its habitat. To examine the influence of climate and soil moisture on the past distribution of the shrub, we modelled its climate niche throughout its range as well as the soil moisture and temporal climate trends of a sample site in the Colorado Plateau.
We acquired Ephedra viridis specimen location records from the SEINet biodiversity databaseiv and 30-year normal climate data from the WorldClim gridded climate database.v In addition to monthly precipitation, maximum temperature, and minimum temperature, WorldClim provided bioclimatic variables derived from those data, such as average temperature of the warmest quarter (three-month period) and precipitation of the wettest quarter. We used several subsets of the bioclimatic variables and their values at each E. viridis presence point as input to a climate envelope model, Bioclim. Bioclim uses the percentile distribution and median values of each climate variable across the known locations of a species to determine the ideal combination of variables for the species.vi The farther the value of a given site from the median, the less suitable the site is in relation to the model. Each subset of variables constituted a different model, which could subsequently predict suitability scores for the entire southwest region of the U.S. We then calculated the AUC, a model evaluation statistic, for each model, averaging across five training and testing partitions of the presence points. We compared the AUCs to determine which subset of bioclimatic variables best predicted E. viridis occurrence in space.
In order to examine temporal trends, we used the model with the best AUC to calculate the suitability score over 50 years of a site in the Needles district of Canyonlands National Park, an area known to have increased in E. viridis density over the last 30 years. We also plotted individual climate variables for the site and performed linear regressions to summarize trends.
Finally, we ran another model, SOILWAT, to estimate the seasonal soil moisture values at the site over a 34-year period. SOILWAT is a process-based model which simulates the movement of water through surface and soil layers, representing the soil water available to plants given the climate and soil type of the site.vii Because this model requires daily weather values, we used the PRISM gridded climate database.viii
The spatial climate model which best predicted E. viridis, with an average AUC of 0.79, included annual precipitation, mean annual temperature, mean temperature of the wettest quarter and of the driest quarter, and precipitation of the warmest quarter and of the coldest quarter. The median values for temperature variables were mild, and precipitation was consistent across quarters. The Colorado Plateau contained the highest concentration of high suitability scores, although the model did not perfectly overlap with the known presence points (Fig 1). Our site has had suitable temperatures but less precipitation than the best sites. We found that the suitability of our site, while not consistently high in the past, has increased in recent years based on this model. At our site, annual mean temperature and mean temperature of the coldest quarter have increased, as have precipitation of the coldest quarter and temperature of the driest quarter (Fig 2). The SOILWAT model revealed that temperature was the biggest driver of soil water content across soil layers at our site, with the warmest quarter (typically June, July, and August) at lower levels than the driest quarter (typically April-June or Sep-Nov). The highest moisture levels were in the coldest quarter.
Discussion and Conclusion
The spatial climate model suggests that E. viridis does best with a combination of consistent precipitation and mild temperatures. The improving suitability of our Canyonlands site based on the model, in conjunction with the documented increase in E. viridis, indicates that directional climate change at this site has made it more suitable for this evergreen species. The rising temperature, while increasing potential soil moisture evaporation, may represent an advantage for E. viridis in the winter, when grasses are not active and yet plenty of soil moisture is available. A slight increase in winter precipitation also adds to this advantage. With the driest quarter becoming warmer, increasing the evaporation potential for shallow soil layers, deeply rooted shrubs such as E. viridis may also have a growing advantage over perennial desert grasses in the spring and summer. Future research will incorporate soil moisture availability into the spatial species distribution model to determine the full importance of this factor in the distribution of E. viridis. We will also model future suitability for E. viridis under climate change scenarios.
i Munson, Seth M., Jayne Belnap, Charles D. Schelz, Mary Moran, and Tara W. Carolin. 2011. On the brink of change: plant responses to climate on the Colorado Plateau. Ecosphere 2 (6): art68.
iii “Species: Ephedra Viridis.” U.S. Forest Service. Accessed December, 2016. https://www.fs.fed.us/database/feis/plants/shrub/ephvir/all.html.
iv SEINet, New Mexico-Arizona chapter. Accessed December, 2016. http://swbiodiversity.org/seinet/index.php.
v WorldClim Global Climate Data. worldclim.org.
vi Booth, Trevor H., et al. 2014. “Bioclim: The First Species Distribution Modelling Package, Its Early Applications and Relevance to Most Current MaxEnt Studies.” Diversity and Distributions 20 (1): 1–9.
vii Lauenroth, W. K., and J. B. Bradford. 2006. Ecohydrology and the partitioning AET between transpiration and evaporation in a semiarid steppe. Ecosystems 9 (5): 756–67.
viii PRISM Climate Group. http://www.prism.oregonstate.edu/.