Identification of degradation factors in mountain semiarid rangelands using spatial distribution modelling and ecological niche theory

Abstract

This paper presents an approach to studying rangeland degradation by using remote sensing data, geographic information systems, and ecological niche theory. The role of environmental factors and land use in the spatial distribution of degraded rangelands in the Central Caucasus was assessed. Degradation stages were modelled in R using ENVIREM predictors, VIF test to select noncorrelated variables, ENMeval package to select the optimal model parameters, and the Maxent method to develop distribution models in the dismo package. Selected models had AUCtest, AUCtrain and CBI values close to 1, deltaAICc and AUCdiff values close to 0, and quite low AICc values. Environmental predictors of climate type (Thornthwaite aridity index and Emberger’s pluviothermic quotient) were explanatory variables for least disturbed rangelands, while topography (Terrain roughness index) largely explained the distribution of most disturbed grasslands. Quantitative (Schoener’s D) and graphical (Kernel density estimation, analysis of predictive maps) assessment revealed a significant overlap of ‘ecological niches’ and potential ranges of grasslands at different degradation stages, which indirectly supports the hypothesis of the important role of overgrazing in their degradation. Livestock management is likely to help restore disturbed mountain meadow steppes to steppe grasslands. Restoration of the arid shrub ecosystems to steppe grasslands or meadow steppes probably requires additional agricultural practices.

Keywords:

• Spatial modelling
• remote sensing
• ecological niches
• rangeland degradation

Acknowledgements

This research was carried out under the State Assignment, project 075-00347-19-00 (Patterns of the spatiotemporal dynamics of meadow and forest ecosystems in mountainous areas (Russian Western and Central Caucasus)).

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

The data that support the findings of this study are openly available in protocols.io at dx.doi.org/10.17504/protocols.io.b5iuq4ew