hide
Free keywords:
Establishment conditions
Spatial autocorrelation
Plant traits
Size
Ecology
Soil
Flora
Angiosperms
Gradients
Models
Abstract:
With environmental factors being spatially structured, plant traits that are related to these factors should exhibit a corresponding spatial pattern. We analyzed the distribution pattern of seed mass in Germany. We calculated the median seed mass for 10' longitude by 6' latitude grid cells across Germany using the trait databases BIOLFLOR and the plant distribution database FLORKART. To explain these distribution patterns of median seed mass, we applied multiple regression analyses on twelve selected environmental variables, accounting for spatial autocorrelation. To deal with collinearities of the predictors, we used hierarchical partitioning to analyze the independent and joint explanatory power of the environmental variables. To test whether statistical relationships are due to hidden correlations between seed mass and plant growth form, we conducted seperate analyses for annual, perennial herbs, shrubs and trees. Low median seed mass was found in the lowlands and river valleys whereas high median seed mass was typical for the rich loess regions and the calcareous mountain ranges. Seed mass exhibited a strong positive correlation with soil pH (62% of variance explained) and a negative correlation with soil moisture (25%). Light was less important as a predictor of seed mass. Within the growth forms we observed similar distribution and correlation patterns pointing to a direct link between seed mass and the environmental variables soil pH and moisture. We argue that this striking relationship with soil pH is caused by the high stress from competition on fertile calcareous sites. The negative moisture effect may be due to drought stress. Both relationships are particulary interesting for the prediction of ecosystem responses to climate and land use changes. [References: 65]
