Using Regional Climate Projections to Guide Grassland Community Restoration in the Face of Climate Change

Abstract

Grassland loss has been extensive worldwide, endangering the associated biodiversity and human well-being that are both dependent on these ecosystems. Ecologists have developed approaches to restore grassland communities and many have been successful, particularly where soils are rich, precipitation is abundant, and seeds of native plant species can be obtained. However, climate change adds a new filter needed in planning grassland restoration efforts. Potential responses of species to future climate conditions must also be considered in planning for long-term resilience. We demonstrate this methodology using a site-specific model and a maximum entropy approach to predict changes in habitat suitability for 33 grassland plant species in the tallgrass prairie region of the U.S. using the Intergovernmental Panel on Climate Change scenarios A1B and A2. The A1B scenario predicts an increase in temperature from 1.4 to 6.4°C, whereas the A2 scenario predicts temperature increases from 2 to 5.4°C and much greater CO2 emissions than the A1B scenario. Both scenarios predict these changes to occur by the year 2100. Model projections for 2040 under the A1B scenario predict that all but three modeled species will lose ~90% of their suitable habitat. Then by 2080, all species except for one will lose ~90% of their suitable habitat. Models run using the A2 scenario predict declines in habitat for just four species by 2040, but models predict that by 2080, habitat suitability will decline for all species. The A2 scenario appears based on our results to be the less severe climate change scenario for our species. Our results demonstrate that many common species, including grasses, forbs, and shrubs, are sensitive to climate change. Thus, grassland restoration alternatives should be evaluated based upon the long-term viability in the context of climate change projections and risk of plant species loss.

Elevated precipitation alters the community structure of spring ephemerals by changing dominant species density in Central Asia

“…Research regarding the change mechanisms of plant community compositions and ecosystem functions in natural ecosystems is always a major research challenge in ecology (Kane et al, 2017;Xu et al, 2018), especially in response to climate changes in recent decades (Eskelinen & Harrison, 2015). Desert ecosystems are more dependent on water availability than on any other factor (e.g., nutrients) in arid and semiarid regions (Weltzin et al, 2003).…”

Toward reliable habitat suitability and accessibility models in an era of multiple environmental stressors

“…The majority of SDM studies still relies on a single environmental stressor (or group of related stressors), with bioclimatic stressors (e.g., warmer and dryer conditions) representing, by far, the most popular environmental predictor used to forecast species’ distribution changes (Titeux et al., 2016). These studies observed drastic reductions in the modeled climatic suitability of currently occupied habitat for macroinvertebrates (up to 65%, Domisch et al., 2013; Parmesan et al., 1999), vertebrates (up to 80%, Warren, Wright, Seifert, & Shaffer, 2014), and plants (up to 90%, Aguirre‐Gutiérrez, van Treuren, Hoekstra, & van Hintum, 2017; Kane et al., 2017). In addition, poleward and upward shifts of species distributions are widely observed (Perry, Reid, Ibanez, Lindley, & Edwards, 2005, Kelly & Goulden, 2008, Chen, Hill, Ohlemuller, Roy, & Thomas, 2011, Lenoir & Svenning 2013) and predicted (Aguirre‐Gutiérrez et al., 2017; Barton, Irwin, Finkel, & Stock, 2016; Inoue & Berg, 2017), yet the velocity of species range shifts is generally thought to be inferior to the velocity of climate change (Bertrand et al, 2011; Chivers, Walne, & Hays, 2017; Corlett & Westcott, 2013; Devictor et al., 2012; Liang, Duveneck, Gustafson, Serra‐Diaz, & Thompson, 2017; Schloss, Nunez, & Lawler, 2012; Zhu, Woodall, & Clark, 2012).…”

Evaluating the Utility of Species Distribution Models in Informing Climate Change-Resilient Grassland Restoration Strategy

“…All models were evaluated by the area under the receiveroperator characteristic curve (AUC) (Merow et al, 2013;Kane et al, 2017). Cross-validation by resampling k-folds (k = 5) was performed to obtain a more discriminant AUC value, and for all models mean AUC was >0.82, indicating robust model performance for all species distribution models (Table S1).…”

“…Initially, all models are informed by eight bioclimatic (BIOCLIM) variables: mean temperature for the wettest (BIO8), driest (BIO9), warmest (BIO10), and coldest (BIO11) quarters averaged from 1960 to 1990 as well as the mean precipitation for the same quarters (BIO16 through 19 respectively). These variables were selected because they are the finest temporal BIOCLIM variables and are thus most likely to represent with the other BIOCLIM variables at occurrence locations were included in training each model; this value was reduced from the 0.75 threshold used by Kane et al (2017) due to the use of a greater number of predictor variables. Correlation was assessed on a per-species basis (Pearson and Dawson, 2003;Elith et al, 2010).…”

“…In particular, species distribution models (SDMs) can be used relatively easily (Kane et al, 2017) to assess how the effectiveness of species in a restoration seed-mix will change under different future climate predictions, and seed-mix composition can then be modified in the present without the need to wait years for experimental results. SDMs are used to predict the environmental suitability of a continent or landscape for individual species both in present conditions and under a variety of future climate change scenarios (Soberón and Nakamura, 2009;Elith et al, 2011).…”