Our study investigated whether problem-oriented designed ecology lessons with phases of direct instruction and of open experimentation foster the development of cross-domain and domain-specific components of "experimental problem-solving ability" better than conventional lessons in science. We used a paper-and-pencil test to assess students' abilities in a quasi-experimental intervention study utilizing a pretest/posttest control-group design (N?=?340; average performing sixth-grade students). The treatment group received lessons on "forest ecosystems" consistent with the principle of education for sustainable development. This learning environment was expected to help students enhance their ecological knowledge and their theoretical and methodological experimental competencies. Two control groups received either the teachers' usual lessons on forest ecosystems or non-specific lessons on other science topics. We found that the treatment promoted specific components of experimental problem-solving ability (generating epistemic questions, planning two-factorial experiments, and identifying correct experimental controls). However, the observed effects were small, and awareness for aspects of higher ecological experimental validity was not promoted by the treatment.