Predicting optimal habitats of Haloxylon persicum for ecosystem restoration using ensemble ecological niche modeling under climate change in southeast Iran

Abstract

Ecological restoration plays a vital role in the management of degraded ecosystems; nevertheless, the success of restoration plans depends to a large extent on determining optimal habitats for target species' growth and survival. Ecological niche modeling can be used to predict where climate is presently suitable for a particular species used in restoration, and where suitable climates will be located in the future. Here, we used ensemble ecological modeling to identify areas suitable for restoration of the arid shrub Haloxylon persicum in southeast Iran. We generated environmental suitability predictions for H. persicum restorations from eight different niche modeling algorithms and averaged these predictions to generate an ensemble model prediction. We present projections for the climate suitability of H. persicum for the present climate and the years 2050 and 2070 under the RCP4.5 and RCP8.5 emissions scenarios and for three different global circulation models. Our model projected the optimum mean annual temperature for this species to be between 35.5 and 37 C, the optimum temperature of the coldest quarter to be between 3.5 and 13 C and the optimum precipitation of the coldest quarter to be between 65 to 90 mm. Based on our results, H. persicum is predicted to persist north of 27 N latitude in southeast Iran under current and future conditions; however, the availability of habitat suitable for H. persicum restoration is expected to dramatically diminish under future climate change. This study will help identify and prioritize locations for ecological restoration of this species in the face of climate change.

Predicting the Potential Suitable Distribution of Larix principis-rupprechtii Mayr under Climate Change Scenarios

“…According to the evaluation metrics, the six single species distribution models of FDA, GBM, RF, MAXENT, GLM, and MARS performed better, followed by GAM, ANN, and CTA models, while the SRE model performed the worst. This is consistent with other results using multiple models to predict species distribution, which shows that the simulation effect of RF is better, while the performance of SRE is the worst [2,32,66]. Some studies have shown that there is some uncertainty in using ROC alone to evaluate the model [67].…”

“…Most SDMs require species absence data. In this research, we randomly generated pseudo-existence point data at least 3 km between the pseudo-existence points and the existence point data [27,32]. To reduce the predictive uncertainty caused by the randomly generated point position, we generated three sets of pseudo-absence data, and the number of pseudo-existence points in each group was 1500.…”

“…The distribution areas of L. principis-rupprechtii were divided into four classes [53]: unsuitable area (0.0-0.2), lowly suitable area (0.2-0.4), moderately suitable area (0.4-0.6), and highly suitable area (0.6-1.0). At the same time, we overlaid suitable and unsuitable area classifications under current and future climate scenarios, and finally obtained the unsuitable areas, unchangeable potential suitable areas, degraded potential suitable areas, and increased potential suitable areas under future climate scenarios [32].…”

“…However, because the principles and algorithms of each model are different, it is difficult to select an appropriate model to predict precisely the suitable distribution of species [31]. In order to solve this kind of uncertainty, more and more research has adopted the ensemble modeling approach, which combines predictions across different modeling methods [32]. The ensemble model (EM) is considered to be an effective method to evaluate the potential distribution of species at a large spatial scale [33], and can effectively solve the uncertainty caused by a single model [34].…”

Climate Change May Impact Nile Tilapia, Oreochromis niloticus (Linnaeus, 1758) Distribution in the Southeastern Arabian Peninsula through Range Contraction under Various Climate Scenarios

“…However, choosing the right model for analysis and obtaining an accurate and reasonable distribution of predicted species is difficult because the principles and algorithms of each model are different [33]. To resolve this type of uncertainty, an increasing number of studies employ ensemble modeling approaches that combine predictions from different modeling techniques [34].…”