Predicting plant species richness in forested landslide zones using geostatistical methods

Ecological Indicators

Wiktor Tracz¹  Yisa Ginath Yuh²  Elvis Tangwa²  Vilém Pechanec³ 

[1] Corresponding author.;Department of Geoinformatics, Palacky University Olomouc (UP), 17. listopadu 50, 77146 Olomouc, Czech Republic;Department of Geomatics and Land Management, Faculty of Forestry, Warsaw University of Life Sciences, Nowoursynowska 159, 02-7767 Warsaw, Poland;

关键词: Terrain attributes;    Regression kriging;    Cokriging;    Species richness;    Convergence points density;    Forested landslides;   

DOI  :  

来源: DOAJ

【 摘 要 】

Landslides, like most natural disturbances, facilitate the evolution of new plant species. Hence, a detail characterization of topographic conditions can improve the prediction and mapping of species in such complex terrains. Within the Outer (Flysch) Upper Carpathian region, south Poland, we analyze the role of convergence points prepared in a previous study from slope and slope exposition (aspect) data, derived from a 1 m digital elevation model. Convergence points reflected microscale variability in topographic conditions and were analyzed in this study as convergence point density (CPD). Our objective was to use CPD to predict species richness on forested landslides using three geostatistical methods; Ordinary kriging (OK), Ordinary cokriging (OCK), and regression kriging (RK). Our results showed a relatively high correlation (r ∼ 0.65) between species richness and CPD compared to slope or slope exposition or with both. OCK and RK generally improved prediction. However, the application of cokriging in such terrains remains challenging and will not be appropriate, particularly if species richness has a small sample size. RK outperformed OK and OCK, decreasing the root mean square error (RMSE) by 33% and 10%, respectively. RK was also more robust to topographic heterogeneity and the limited number of observations than OCK. We conclude that a denser sampling of species composition or a more robust indicator is needed to improve these results. Notwithstanding these limitations, our results can be used as the first step to support short-term conservation efforts, especially when time-dependent changes in species composition are unimportant.

【 授权许可】

Unknown