Although gemma production is common among liverworts, and thought to play a key role in their population dynamics, quantitative data on numbers produced at shoot and colony levels, or on spatial variation in these numbers, are very sparse. Here we investigate gemma output of a boreal and arctic liverwort species, Lophozia ventricosa, at populations in Trøndelag, a boreal region of central Norway, and in arctic tundra on Svalbard. Production parameters investigated include colony size, shoot density, number of gemmae per shoot, proportion of shoots gemmiferous, and gemmiferous shoot density. Colony-level output was calculated as the product of the number of gemmae per shoot, colony size, and gemmiferous shoot density. Nested analysis of variance, used to spatially partition variance in these gemma-production components, reveals that the vast majority of the variation lies at the measurement level, i.e. among shoots within colonies or among colonies within plots. The only significant difference between the two geographic areas is that in boreal sites shoots tend to produce slightly more gemmae, and form larger, denser, colonies. This translates to somewhat higher colony-level output in boreal, compared to arctic, sites. Evaluation of a structural equations model reveals intricate patterns of relationship among the production components and provides strong evidence of negative density dependence of shoot-level production. We argue that explicit evaluation of size-number tradeoffs, allometric scaling of production with shoot size, and detailed studies of genetic versus phenotypic effects on production levels, are necessary to understand how asexual propagule production is shaped by selection on life histories.