This study analyzes the effect of different instructions on threshold concepts within material covering natural selection on students' use of concepts about evolution. Moreover, it examines students' use of concepts as interconnected networks when reasoning about natural selection and analyzes how these concepts relate to each other regarding their connections and centrality within the network. We recruited a sample of N = 128 German 10th-grade students to address these aspects and conducted an experimental intervention study with three groups. We assessed the effect of instruction on the threshold concepts of randomness and probability by teaching them either in contexts of biology, the original discipline of mathematics, or not at all, along with input on natural selection. We analyzed the number of students' key concepts, misconceptions, and threshold concepts by applying an inferential statistical and concept network approach. The results reveal that teaching threshold concepts in biological or mathematical contexts leads to significantly higher use of key concepts than in the control group. Furthermore, only teaching them in a biological context resulted in a significantly higher use of threshold concepts than in the control group. The network analyses reveal descriptive differences in the structures, significant correlations of concepts, and most central and influential concepts (i.e., key concepts: selection factors or limited resources, change in the population; misconceptions: teleology/need, anthropomorphic conception; threshold concept: probability). We discuss the valuable potential of explicitly teaching threshold concepts and utilizing network theory to enhance a nuanced understanding of how students organize and apply evolutionary concepts.