To achieve deeper understanding when learning from multiple representations, learners should actively select, organize and integrate the relevant information from text and graphics within a coherent mental representation. However, as learners often fail to select and integrate all relevant information, especially from graphics, they need specific instructional support. The current study investigated the effects of instructional support in the form of adjunct aids (i.e. fill-in-the blank tasks) with references to the graphics (e.g. "see Figure X") on retention and understanding. In our study, 106 learners (N) received multimedia instructional materials about the formation of auroras either with or without adjunct aids -- the former with references to graphics (signals), or with none. In line with our hypotheses, adjunct aids with signals led to deeper understanding, as reflected by higher scores in the comprehension test. In contrast, adjunct aids with signals did not lead to higher scores in the retention test. Thus, our results are in line with previous research, showing that instructional support for integrating text and graphics specifically fosters deeper understanding. Possible boundary conditions and implications for future research are discussed.