21Species range limits are thought to result from a decline in demographic performance because 22 of unsuitable climate at the edges. However, recent studies reporting contradictory patterns in 23 tree species demographic performance at their edges cast doubt on our ability to predict climate 24 change impacts on species ranges. Here we parameterised integral projection models with climate 25 and competition effects for 27 tree species using forest inventory data from over 90,000 plots across 26 1 Europe. Then, we predicted growth, survival, lifespan, and passage time -the time to grow to a 27 large size -at the hot and cold edges and compared them to the range centre. We found that while 28 growth and passage time of European tree species are constrained at their cold edge, survival 29 and lifespan are constrained at their hot edge. Our study shows a more complicated picture than 30 previously thought with demographic responses that differ between hot and cold edges. 31 Keywords: "demography, IPM, passage time, vitale rate, range edge" 32 33Increasing concerns have emerged regarding potential major redistributions of plant species ranges 35 in the coming decades due to climate change (Zimmermann et al. 2013). Indeed, some studies 36 have confirmed that these range shifts are already underway (Chen et al. 2011). Range shifts are 37 directly related to changes in demographic rates and population dynamics. Demographic theories 38 propose that if populations are at equilibrium, mean population growth rate (λ) will drop at the 39 species' range edge (λ < 1) due to alterations in one or more vital rates contributing to λ (Case et 40 al. 2005; Holt & Keitt 2005). This prediction is grounded in a long-standing hypothesis in biogeog-41 raphy, known as the 'abundance-centre hypothesis' (hereafter ACH, Brown 1984), which proposes 42 that demographic performance decline at the range edge results in a decrease in abundance. This 43 decline in demographic performance can arise directly because of abiotic constraints (e.g. frost) 44 or indirectly because of changes in biotic constraints (e.g. competition) (Hargreaves et al. 2014; 45 Pironon et al. 2017).
46The number of studies that have directly tested the ACH with field data on population growth 47 rates, and the vital rates that contribute to them, at the centre and the edge of species range is sur-48 prisingly limited. Nonetheless, these studies showed only weak or contradictory support for the 49 ACH. Transplant experiments have shown that population growth rate or some vital rates tend 50 to decline beyond the edge but not necessarily right at the edge (Hargreaves et al. 2014; Lee-Yaw 51 et al. 2016). For long-lived organisms such as trees, their generation time rules out transplant ex-52periments that would cover their full life cycle. Rather, researchers have to rely on models based 53 2 on natural population monitoring data (see Purves 2009). Only a few studies have used this ap-54 proach over the full species range and they found no clear evidence of a de...