Most growth is exponential (for a while, as a logistics curve (aka S-curve) does start exponentially), how about showing log(y(x)) instead of y(x).

But not always: if zero is not on the bar chart, e.g. log(y(x)) with [x] = Kelvin, such a plot will of course not work. So keep temperature as temperature increment over pre-industrial times or other baseline.

Also, does coastal nitrogen have three data points, or why the edge in the middle? Scatter charts plus trend are more the real thing than a curve if this is two linear relationships, with a second slope after a certain time, and keep y(x) plotting here, of course.