What are the consequences of gender quotas on women's descriptive and substantive representation? Using a novel dataset of over 20,000 rural local governance bodies spanning 20 years, we find that randomly implemented gender quotas for women do not substantially increase the chances of women winning elections to the same seat once the seat is unreserved. Troublingly, even fifteen years of exposure to local women leaders has limited effects. We probe the reasons for this finding using an original phone survey. When we try to reach elected representatives in reserved seats using their official numbers, male relatives frequently intercept the call, suggesting ``gatekeeping.'' Data suggest that in poor, rural areas, quotas may fail to transform who is de facto in power. In all, unlike more hopeful findings from urban areas, our findings suggest that quotas mechanically improve the representation of women but do not create conditions for women to be in power after the quotas are withdrawn.