Background/Question/Methods   Population projections are often used to assess life history evolution and population viability, but assume that past and future environmental variation stays within levels observed in the study and that the genetic composition of a population is either unchanging or inconsequential to population dynamics. Here, we assess the validity of these assumptions in a longitudinal, experimental field study of the common ribwort plantain, Plantago lanceolata. A large population was established that included four cohorts with equivalent genetic structure, each planted at a different time between Nov 2000 and Oct 2003. This study uses annual censuses conducted for mortality and reproduction through Dec 2007. We hypothesized that different planting times would cause different early-life environments and different selection regimes across cohorts, and that both factors would result in differing cohort demographies.

Results/Conclusions   Overall mortality through age 5 varied with cohort, block, and sire, suggesting a strong influence of early-life environment, genetic variation, and spatial variability. Net reproductive rate and projected lambda at stationarity varied significantly with cohort and sire, suggesting strong impacts of early-life environment and genetic variation, and with a cohort × sire interaction, suggesting differing norms of reaction across cohorts. Variable early-life environments and genetic variation can cause substantial deviations from inferences of population dynamics. We stress caution in forecasting the results of population analyses without a thorough understanding of the effects of this variability.