J Plant Ecol ›› 2018, Vol. 11 ›› Issue (2): 169-179.

• Research Articles •

### Relative importance of hydrological variables in predicting the habitat suitability of Euryale ferox Salisb.

Ji Yoon Kim1, Gu-Yeon Kim1, Yuno Do1, Hee-Sun Park2 and Gea-Jae Joo1,*

1. 1 Department of Integrated Biological Science, College of Natural Sciences, Pusan National University, Busan 46241, Republic of Korea; 2 Nakdong Estuary Eco Center, Busan 49435, Republic of Korea
• Received:2015-12-21 Accepted:2016-09-27 Published:2018-02-06
• Contact: Joo, Gea-Jae

Abstract: Aims Aquatic ecosystems are a priority for conservation as they have become rapidly degraded with land-use changes. Predicting the habitat range of an endangered species provides crucial information for biodiversity conservation in such rapidly changing environments. However, the complex network structure of aquatic ecosystems restricts spatial prediction variables and has hitherto limited the use of habitat models to predict species occurrence in aquatic ecosystems. We used the maximum entropy model to evaluate the potential distribution of an endangered aquatic species, Euryale ferox Salisb. We tested the relative influence of (i) climatic variables, (ii) topographic variables, and (iii) hydrological variables derived from remote sensing data to improve the prediction of occurrence of aquatic plant species.
Methods We considered the southern part of the Korean Peninsula as the modeling extent for the potential distribution of E. ferox. Occurrence records for E. ferox were collected from the literature and field surveys. We applied maximum entropy modeling using remotely sensed environmental variables and evaluated their relative importance as prediction variables with variation partitioning.
Important findings The species distribution model predicted potential habitats of E. ferox that matched the actual distribution well. Floodplain wetlands and shallow reservoirs were the favored habitats of E. ferox. Quantitative loss and fragmentation of wetland habitats appeared to be a major reason for the decrease of E. ferox populations. Our results also imply that hydrological variables (i.e. normalized difference water index) derived from remote sensing data greatly increased model prediction (relative contribution: 10.5–37.0%) in the aquatic ecosystem. However, interspecific competition within a similar niche environment should be considered to increase the accuracy of the distribution model.

Aims Aquatic ecosystems are a priority for conservation as they have become rapidly degraded with land-use changes. Predicting the habitat range of an endangered species provides crucial information for biodiversity conservation in such rapidly changing environments. However, the complex network structure of aquatic ecosystems restricts spatial prediction variables and has hitherto limited the use of habitat models to predict species occurrence in aquatic ecosystems. We used the maximum entropy model to evaluate the potential distribution of an endangered aquatic species, Euryale ferox Salisb. We tested the relative influence of (i) climatic variables, (ii) topographic variables, and (iii) hydrological variables derived from remote sensing data to improve the prediction of occurrence of aquatic plant species.
Methods We considered the southern part of the Korean Peninsula as the modeling extent for the potential distribution of E. ferox. Occurrence records for E. ferox were collected from the literature and field surveys. We applied maximum entropy modeling using remotely sensed environmental variables and evaluated their relative importance as prediction variables with variation partitioning.
Important findings The species distribution model predicted potential habitats of E. ferox that matched the actual distribution well. Floodplain wetlands and shallow reservoirs were the favored habitats of E. ferox. Quantitative loss and fragmentation of wetland habitats appeared to be a major reason for the decrease of E. ferox populations. Our results also imply that hydrological variables (i.e. normalized difference water index) derived from remote sensing data greatly increased model prediction (relative contribution: 10.5–37.0%) in the aquatic ecosystem. However, interspecific competition within a similar niche environment should be considered to increase the accuracy of the distribution model.