It does matter, because if it were log base 10, it would imply that the yield of the fields in 1982 varied approximately a factor of 30, which I would have trouble believing. But with the natural logarithm, it is only roughly a factor of four, which is much more reasonable.
Very interesting! You may want to note that in Figure 9, log refers to log base e (ln), not log base 10.
Does it matter if both axes are on the same scale?
It does matter, because if it were log base 10, it would imply that the yield of the fields in 1982 varied approximately a factor of 30, which I would have trouble believing. But with the natural logarithm, it is only roughly a factor of four, which is much more reasonable.