Chemical fertility of forest ecosystems. Part 1: Common soil chemical analyses were poor predictors of stand productivity across a wide range of acidic forest soils

Forest soil fertility can be defined as a combination of physical, chemical and biological factors characterising the biomass production capacity of the soil. However, numerous ecological variables affect tree growth and the aim of the present study was to investigate the specific influence of soil chemical properties on tree productivity at 49 acidic forest sites. A standardized tree productivity index based on tree height expressed as dominant height of the studied stand divided by maximum tree height observed at the same age for the same species in the same climatic region was firstly computed at each site. This index is assumed to limit the influence of species, ages and climate. A soil database was also compiled with data on soil properties from 47 temperate (France) and two tropical (Congo, Brazil) sites. Data included seven tree species, varying in age from 1 to 175 years. Commonly used indicators such as C:N ratio, soil pH, as well as available and total pools of soil nutrients were compared to the standardized tree productivity index, to find the most reliable indicator(s). Nutrient pools at fixed mineral soil depths (down to 100 cm) were used, as well as (for 11 stands) the depth comprising 95% of fine roots. Our results show that none of the common soil chemical parameters tested in this paper could individually explain stand productivity. Combinations of different parameters were also tested using PCA and they could better explain the variability of the data set but without being able to separate the sites according to their standardized tree productivity index. Moreover, random Forests performed on our dataset were unable to properly predict the standardized tree productivity index. Our results reinforce the idea that the influence of the soil chemical fertility on stand productivity is complex and the soil chemical parameters alone (individually or combined) are poor predictors of tree productivity as assessed by the H0:Hmax index. In this paper we focused on static soil chemical indicator and more dynamic indictors, such as nutrient fluxes involved in the biogeochemical cycles, could better explain stand productivity. A companion paper (Legout et al., 2020) focuses on the connection between productivity and different components of the biogeochemical cycle, using data from 11 of the stands presented in this paper.