Abstract

BackgroundShrubs play a key role in biogeochemical cycles, prevent soil and water erosion, provide forage for livestock, and are a source of food, wood and non-wood products. However, despite their ecological and societal importance, the influence of different environmental variables on shrub distributions remains unclear. We evaluated the influence of climate and soil characteristics, and whether including soil variables improved the performance of a species distribution model (SDM), Maxent.MethodsThis study assessed variation in predictions of environmental suitability for 29 Australian shrub species (representing dominant members of six shrubland classes) due to the use of alternative sets of predictor variables. Models were calibrated with (1) climate variables only, (2) climate and soil variables, and (3) soil variables only.ResultsThe predictive power of SDMs differed substantially across species, but generally models calibrated with both climate and soil data performed better than those calibrated only with climate variables. Models calibrated solely with soil variables were the least accurate. We found regional differences in potential shrub species richness across Australia due to the use of different sets of variables.ConclusionsOur study provides evidence that predicted patterns of species richness may be sensitive to the choice of predictor set when multiple, plausible alternatives exist, and demonstrates the importance of considering soil properties when modeling availability of habitat for plants.

Highlights

  • Species distribution models (SDMs) are tools used to assess the spatial distribution of potentially suitable habitat for species, and to hypothesise how suitability is affected by environmental change (Guisan & Thuiller, 2005)

  • Visual inspection of maps generated by Maxent indicated that VC+S and VC models resulted in more realistic projections of habitat suitability than those calibrated with only soil variables (Fig. 2)

  • Species distribution models are frequently calibrated only with climate variables, but for plant species, does the addition of soil properties as predictors improve model performance? For 29 Australian shrub species, we found that: (a) on average, models calibrated with both climate and soil variables (VC+S) performed better than those calibrated solely with climate variables (VC ) (Fig. 1); (b) maximum temperature of the warmest month and pH were the most important contributors to VC+S models for ten and eight species, respectively (Table 3); and (c) models calibrated with only soil variables (VS) had lower area under the receiver-operating characteristic curve (AUC) and True Skill Statistic (TSS) scores, indicating lower classification accuracy than VC models (Fig. 1), and frequently generated unrealistic predictions (Figs. 2 and 3)

Read more

Summary

Introduction

Species distribution models (SDMs) are tools used to assess the spatial distribution of potentially suitable habitat for species, and to hypothesise how suitability is affected by environmental change (Guisan & Thuiller, 2005). Some studies incorporated other factors in SDMs such as irradiance (see Franklin, 1998; Summers et al, 2012) as a light source for the plants, topography (Franklin, 1998; Hosseini et al, 2013), and landuse (Meier et al, 2012; Stanton et al, 2012; Titeux et al, 2016) in which degradation in plant habitats and loss of plant biodiversity is strongly influenced by changes in landuse and increase of urbanization is considered (Lawler et al, 2014) This poorly integration of non-climatic factors in modelling studies may partly reflect difficulties with obtaining appropriate data sets at relevant spatial scales, with regards to soil variables that are related to plant functionality. Our study provides evidence that predicted patterns of species richness may be sensitive to the choice of predictor set when multiple, plausible alternatives exist, and demonstrates the importance of considering soil properties when modeling availability of habitat for plants

Methods
Results
Discussion
Conclusion
Full Text
Paper version not known

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call

Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.