Abstract
This study aims to develop a product a deep learning-based ecosystem diorama designed to improve environmental literacy among fifth-grade elementary school students. The development of this deep learning-based ecosystem diorama is intended to foster meaningful learning through students’ active engagement in observing, studying, and reflecting on ecosystem concepts in a contextual manner. The subjects of this study were fifth-grade students at SDN 46 Mataram. The total number of research subjects was 21 students. Data collection was conducted through interviews, tests, questionnaires, and documentation. The data analysis techniques used were qualitative and quantitative analysis. The results of the study indicate that the diorama media is feasible and effective in improving students’ environmental literacy. Validation by media and content experts yielded an average V value of 0.942, while content validation by content experts yielded an average V value of 0.93. The student response questionnaire in the small-group pilot test yielded a percentage of 86% in the “very practical” category. Meanwhile, the student response questionnaire in the large-group pilot test yielded a percentage of 84% in the “very practical” category, and the teacher response questionnaire yielded a percentage of 94% in the “very practical” category. Regarding student learning outcomes, the small group pilot test yielded an average N-gain score of 0.497, while the large-group pilot test yielded an average N-gain score of 0.482. Furthermore, for the student environmental literacy questionnaire, the average N-gain was 0.580 in the small group trial and 0.450 in the large-group trial. Based on these results, it can be concluded that the deep learning-based ecosystem diorama medium has proven effective in improving students’ environmental literacy.