I'm not a good data scientist, but does Figure 5 look suspect to anyone else? That trend seems to be the central conclusion of the paper, but the data seems to

I'm not a good data scientist, but does Figure 5 look suspect to anyone else? That trend seems to be the central conclusion of the paper, but the data seems to indicate that there are other, more important factors at play. For example, the column of points directly above 0% change in land use. They vary by a large percentage. Another example is the universal 20% drop regardless of land use changes.