Gómez-Aparicio et al.-1 Response of tree seedlings to the abiotic heterogeneity generated by nurse shrubs: an experimental approach at different scales

Spatial heterogeneity of abiotic factors influences patterns of seedling establishment at different scales. In stress-prone ecosystems such as Mediterranean ones, heterogeneity generated by shrubs has been shown to facilitate the establishment of tree species. However, how this facilitation is affected by spatial scale remains poorly understood. We have experimentally analysed the consequences of the abiotic heterogeneity generated by pioneer shrubs on survival, growth and physiology of seedlings of three important tree species from Mediterranean mountains (Acer opalus subsp. granatense, Quercus pyrenaica and Quercus ilex). Patterns of abiotic heterogeneity and seedling performance were studied at two scales differing in grain: 1) the microhabitat scale, by using open interspaces as controls of the effect of different shrub species, and 2) the microsite scale, analysing the effects of fine-grain heterogeneity (within-microhabitat heterogeneity). Results showed that, at the microhabitat scale, seedling establishment of the three tree species significantly benefited from the modification of the abiotic environment by nurse shrubs. However, we found shrub-seedling interactions to be species-specific, due to differential modification of both aboveground (light availability) and belowground (soil compaction, water content, and fertility) abiotic factors by nurse shrub species. Heterogeneity at the within-microhabitat scale was rather high, although it did not significantly affect seedling performance of any of the tree species. The study demonstrates that the effects of the abiotic heterogeneity generated by shrubs are not consistent across the range of spatial scales considered. The regeneration niche of tree species becomes very complex at fine spatial scales, and thereby estimators of abiotic heterogeneity are valuable descriptors of spatial patterns of seedling establishment only when microsite “noise” is averaged out at greater scales. Gómez-Aparicio et al.-3