Do hypotheses from short-term studies hold in the long-term? An empirical test

1. A sequence of population estimates for two now-extinct populations of Euphydryas editha bayensis is presented. After removing biased sampling days, estimates of demographic parameters from the long-term data were used to test five hypotheses built from studies of shorter duration. Such tests of short-term conclusions are rare. 2. The long-term demographic parameters include sex ratio, mortality, dispersal, mean flight date, and duration of flight season. The two populations differed with respect to sex ratio and mean flight date, and sexes differed with respect to mortality and dispersal. 3. Three of the five hypotheses were supported directly or indirectly by patterns in the parameters. These hypotheses predict that dynamics are asynchronous over the long term, that larval mortality, not adult abundance and mortality, is the primary determinant of changes in population size, and that topography mediates larval mortality. 4. Two hypotheses were not supported or supported only in part. Flight phenology differed between the study populations as predicted, but flight order was opposite that expected from the topographic composition of each habitat. Variability in sex ratio and the occurrence of female-biased ratios in the habitat of one of the populations also suggest that previous observations of sex ratio are not generalisable. 5. Populations were extremely volatile over the study period. Removal of biased sampling days did not change basic trends or fluctuations in the data. This volatility suggests that E. editha populations residing in similar habitats may risk immediate extinction.