Predicting changes in the abundance of species over time is a central goal of conservation ecology. Typically, demographic data are used to parameterise models which describe population dynamics. Vital rates of plants vary across a species' distribution, yet most demographic estimates are spatially constrained. Banksia ornata (Proteaceae), a widespread serotinous obligate‐seeding shrub in southern Australia, has previously had a fire management model developed based on demographic data derived from a chronosequence in one part of the species' range. We asked: are demographic data collected from a limited geographic range robust for developing population dynamic models that are applied across the broader species' range? How much variation exists in the demography of species across their range and does this matter for estimating fire return intervals necessary for species persistence? We examine site‐to‐site variation in the pyrodemography of six populations of B. ornata encompassing its geographic range. We documented changes in population density, time to first reproduction and canopy seed bank accumulation in stands varying in age from 0.5 to 55 years. We model year of maximum seed production on demographic parameters such as population starting density and survival rate. Density, primary juvenile period, rates of mortality and seed bank accumulation varied across populations. Primary juvenile period varied across populations (from 6 to <20 years), while time to maximum seed production was estimated to take 34–39 years after fire. Hence, the fire intervals that correspond to minimising extinction risk (due to variation in primary juvenile periods) diverge moreso among populations than the fire intervals that produce (on average) the highest maximum seed production with respect to time since fire. Demographic data collected at different locations in a species' range are likely valuable for building realistic species response models to disturbance, providing underlying assumptions about key transitions, such as mortality rate, are made explicit. [ABSTRACT FROM AUTHOR]