Ecological Engineering 71 (2014) 335–345 Contents lists available at ScienceDirect Ecological Engineering jou rn al hom ep age: www.elsevier.com/locate/ecoleng Comparison of habitat suitability models using different habitat suitability evaluation methods Yujun Yi a,b,∗ , Xi Cheng a , Silke Wieprecht b , Caihong Tang a a Ministry of Education Key Laboratory of Water and Sediment Science, School of Environment, Beijing Normal University, Beijing 100875, China b Institute for Modelling Hydraulic and Environmental Systems, University of Stuttgart, D-70569 Stuttgart, Germany a r t i c l e i n f o Article history: Received 21 January 2014 Received in revised form 8 May 2014 Accepted 11 July 2014 Keywords: Preference curve Fuzzy logic Data-driven method Habitat suitability Chinese sturgeon a b s t r a c t When simulating the habitat suitability for a species or a community, the selection of the habitat eval- uation index is still a matter of debate. This study examined Chinese sturgeon spawning grounds in the Yangtze River as an example. The water level and velocity were simulated by a two-dimensional hydraulic model. Five methods, including a two-dimensional preference curve method, one- and two-dimensional expert knowledge-based fuzzy logic methods, and one- and two-dimensional data-driven fuzzy logic methods, were used to calculate the habitat suitability of Chinese sturgeon spawning grounds below the Gezhouba Dam. The data-driven fuzzy logic method used a training dataset and the nearest climbing algorithm to optimize fuzzy sets and fuzzy rules. The weighted usable area (WUA), hydraulic habitat suit- ability index (HHS) and the spatial distribution of suitable Chinese sturgeon spawning grounds at different discharges (4000 m 3 /s, 9000 m 3 /s, 12,000 m 3 /s, 16,000 m 3 /s, 20,000 m 3 /s, 30,000 m 3 /s, and 40,000 m 3 /s) were calculated. The performances of the habitat suitability models based on different methods were compared. There were few differences in the results regardless of the use of a preference curve method or a fuzzy logic method. For the data-driven fuzzy logic method, the quality of the training dataset is very important. Therefore, the data-driven fuzzy logic method is a good method when the amount and coverage (e.g., including very high and very low discharges) of the measured dataset are sufficient. However, condi- tions at different discharges can be taken into account comprehensively with the expert knowledge-based method when the available data are not sufficient. © 2014 Elsevier B.V. All rights reserved. 1. Introduction The influences of dams and reservoirs on stream ecology remain hot issues. Especially in China, with its abundant water resources and large number of dams, ecologists and hydrologists are still searching for a useful way to evaluate the ecosystem response to dam construction and to protect and repair damaged ecosys- tems. Since the late 1970s, aquatic habitat simulation models have been used to analyze fish habitats in water resource management (Bovee, 1982). The models are used to evaluate habitat suitabil- ity for aquatic organisms, based on physical variables, such as water depth, flow velocity, and substrate (Bovee, 1986; Jowett, 1997). Physical habitat models are particularly useful for assessing the impact of hydropower projects, analyzing the effects of water ∗ Corresponding author at: Corresponding author at: Beijing Normal University, School of Environment, No.19, Xinjiekouwai Street, Beijing 100875, China. Tel.: +86 10 58801757; fax: +86 10 58801757. E-mail address: yiyujun@bnu.edu.cn (Y. Yi). abstraction on river ecology, and determining the minimum flow requirements of aquatic populations (Bockelmann et al., 2004). These models may also be used to evaluate the impact of restora- tion projects on surrounding environments (Shields et al., 1997; Lee and An, 2014). The Physical Habitat Simulation (PHABSIM) model, which is based on preference curves, was the first fish habitat model and is now being used worldwide (Bovee, 1986; Spence and Hickley, 2000). PHABSIM is based on a one-dimensional hydraulic model and uses preference curves to provide habitat quality evalua- tion rules. PHABSIM is suitable where the physical habitat limits populations, and, although it provides a quantitative output, it is perhaps most useful in providing a qualitative comparison between management options. Other models based on PHAB- SIM include RHYHABSIM (Jowett, 1996) and FISU (Yrjänä et al., 1999). Parasiewicz (2001) developed Meso-HABSIM for meso- habitats. Other models based on the preference curve method are WHYSWESS (Yi et al., 2010a,b), RIVER2D (Im et al., 2011), and RCHARC (Thoms and Sheldon, 2002). All these models link physical variables to habitat suitability by means of uni- or http://dx.doi.org/10.1016/j.ecoleng.2014.07.034 0925-8574/© 2014 Elsevier B.V. All rights reserved.