The graph in that page is completely different (e.g. the sharp drop in 2015). This seems to be just linearly interpolating between start and end populations.