Reynolds points to a number of interesting meta-analyses that pool outcomes from past research involving more than a million men and women to show that regular exercise makes us more resistant to depression.
Schuch et al. find:
...men and women with the lowest fitness were about 75 percent more likely to have been given diagnoses of depression than the people with the greatest fitness. The men and women in the middle third were almost 25 percent more likely to develop depression than those who were the most fit.
Some of the same authors pooled results from 25 studies that evaluated exercise as a treatment for depression:
...pooled results persuasively showed that exercise, especially if it is moderately strenuous, such as brisk walking or jogging, and supervised, so that people complete the entire program, has a “large and significant effect” against depression
And, on mechanisms, a clip from another
Schuch et al. piece combining the results of twenty studies involving 1353 people to evaluate the neurobiological effects of exercise among people with major depressive disorder (MDD)...
The results demonstrate that a single bout of exercise increases atrial natriuretic peptide (ANP), brain natriuretic peptide (BNP), copepetin and growth hormone among people with MDD. Exercise also potentially promotes long-term adaptations of copeptin, thiobarbituric acid reactive species (TBARS) and total mean frequency (TMF). However, there is limited evidence that exercise promotes adaptations on neurogenesis, inflammation biomarkers and brain structure. Associations between depressive symptoms improvement and hippocampus volume and IL-1β were found. Nevertheless, the paucity of studies and limitations presented within, precludes a more definitive conclusion of the underlying neurobiological explanation for the antidepressant effect of exercise in people with MDD.