If you have access to the data points themselves, you can compute the slope with least-squares regression.
If you don't, the best you can do is approximate the points through which the best-fit line goes. Use any two points; for example, it looks like the line goes through (2, 2.3) and (4, 2.2), so the slope is approximately
[tex]\dfrac{2.2-2.3}{4-2}=\dfrac{-0.1}2=-0.05[/tex]
(min/year)
