Are northern-edge populations of cork oak more sensitive to drought than those of the southern edge?

Abstract

The changes in climate registered at the planetary scale threaten the persistence of current populations for many plant species, with effects particularly evident at the edges of species distributions. However, intraspecific differences in functional traits could modulate the plant responses to the expected increase in drought. Using a traitbased approach, we evaluated under controlled conditions how Quercus suber seedlings from the latitudinal edges of the distribution range of the species respond to different watering treatments in terms of vegetative growth and biomass allocation. In addition, we simulated an extreme drought by stopping watering until death to determine chemical and physiological traits under drought stress and to identify which morphological traits were more associated to drought resistance (expressed as survival time without watering). Seedlings from the northern provenance presented higher aboveground biomass allocation (i.e. shoot length and biomass allocation to shoot and leaves), while the southern ones were characterised by longer roots and higher biomass allocation to roots. Under extreme drought, seedlings from the southern provenance maintained higher photosynthetic rates than northern seedlings and were able to modulate their water-use efficiency (estimated from δ13C) depending on environmental conditions, which allowed them to survive for a longer period. Finally, drought resistance was partially explained by the plant biomass allocation pattern. Traits related to growth in height and light interception were negatively related with drought resistance, whereas traits involving investment in root biomass were positively related with resistance. These geographical differences evidence a local adaptation to drought at the southern edge of Q. suber distribution. Our results highlight the importance of the conservation of the genetic resources that peripheral populations harbour at distribution edges.