Food webs have received global attention as next-generation biomonitoring tools; however, it remains challenging because revealing trophic links between species is costly and laborious. Although a link-extrapolation method utilizing published trophic link data can address this difficulty, it has limitations when applied to construct food webs in domestic streams due to the lack of information on endemic species in global literature. Therefore, this study aimed to develop a link extrapolation-based food web model adapted to Korean stream ecosystems. We considered taxonomic similarity of predation and dominance of generalists in aquatic ecosystems, designing taxonomically higher-level matching methods: family matching for all fish (Family), endemic fish (Family-E), endemic fish playing the role of consumers (Family-EC), and resources (Family-ER). By adding the commonly used genus matching method (Genus) to these four matching methods, a total of five matching methods were used to construct 103 domestic food webs. Predictive power of both individual links and food web indices were evaluated by comparing constructed food webs with corresponding empirical food webs. Results showed that, in both evaluations, proposed methods tended to perform better than Genus in a data-poor environment. In particular, Family-E and Family-EC were the most effective matching methods. Our model addressed domestic data scarcity problems when using a link-extrapolation method. It offers opportunities to understand stream ecosystem food webs and may provide novel insights into biomonitoring.

1. INTRODUCTION
2. MATERIALS AND METHODS
    2.1. Study site selection and data acquisition
    2.2. Food web construction andmodel development
    2.3. Evaluation of model performance
3. RESULTS AND DISCUSSION
    3.1. Model performance in predicting foodweb individual links
    3.2. Model performance in predictingfood web indices
    3.3. Limitations, ecological implications,and conclusion
CRediT authorship contribution statement
Declaration of Competing Interest
ACKNOWLEDGEMENTS
SUPPORTING INFORMATION
REFERENCES

• Minyoung Lee(Department of Biological Sciences, Ulsan National Institutes of Science and Technology (UNIST), Ulsan 44919, Republic of Korea)
• Yongeun Kim(Ojeong Resilience Institute, Korea University, Seoul 02841, Republic of Korea)
• Kijong Cho(Department of Environmental Science and Ecological Engineering, Korea University, Seoul 02841, Republic of Korea) Corresponding author