In this work, data of passive scalar plumes in a turbulent boundary layer are investigated. The experiments are performed in a wind tunnel where a passive scalar is injected through an L-shaped tube. Two source configurations are analysed for two different tube diameters. The passive scalar concentration is then measured at different distances from the source and wall-normal locations. By exploiting the recent advances of complex networks theory, the concentration time-series are mapped into networks, through the visibility algorithm. The resulting networks inherit the temporal features of the mapped time-series, revealing non-trivial information about the underlying transport process. This work represents an example of the great potentialities of the complex network approach for the analysis of turbulent transport and mixing.