Risky future for Mediterranean forests unless they undergo extreme carbon fertilization.

Forest performance is challenged by climate change but higher atmospheric [CO ] (c ) could help trees mitigate the negative effect of enhanced water stress. Forest projections using data assimilation with mechanistic models are a valuable tool to assess forest performance. Firstly, we used dendrochronological data from 12 Mediterranean tree species (six conifers and six broadleaves) to calibrate a process-based vegetation model at 77 sites. Secondly, we conducted simulations of gross primary production (GPP) and radial growth using an ensemble of climate projections for the period 2010-2100 for the high-emission RCP8.5 and low-emission RCP2.6 scenarios. GPP and growth projections were simulated using climatic data from the two RCPs combined with (i) expected c ; (ii) constant c  = 390 ppm, to test a purely climate-driven performance excluding compensation from carbon fertilization. The model accurately mimicked the growth trends since the 1950s when, despite increasing c , enhanced evaporative demands precluded a global net positive effect on growth. Modeled annual growth and GPP showed similar long-term trends. Under RCP2.6 (i.e., temperatures below +2 °C with respect to preindustrial values), the forests showed resistance to future climate (as expressed by non-negative trends in growth and GPP) except for some coniferous sites. Using exponentially growing c and climate as from RCP8.5, carbon fertilization overrode the negative effect of the highly constraining climatic conditions under that scenario. This effect was particularly evident above 500 ppm (which is already over +2 °C), which seems unrealistic and likely reflects model miss-performance at high c above the calibration range. Thus, forest projections under RCP8.5 preventing carbon fertilization displayed very negative forest performance at the regional scale. This suggests that most of western Mediterranean forests would successfully acclimate to the coldest climate change scenario but be vulnerable to a climate warmer than +2 °C unless the trees developed an exaggerated fertilization response to [CO ].