Aboveground vegetation, four belowground fauna groups and humus composition have been analyzed in order to investigate the links between autotrophic and heterotrophic communities in a Norway-spruce mountain forest in Tours-en-Savoie (France). The aboveground plant community was recorded in small patches corresponding to contrasting microhabitats. Animal communities and humus layers were sampled within the same patches. The relationships between humus profile, faunistic and floristic compositional gradients were investigated by Multiple Factor Analysis (MFA) and, for the first time in ecology, a Hierarchical Multiple Factor Analysis (HMFA) was used to interpret differences among humus layers. The analysis revealed a pattern with three main groups of microhabitats. The thorough study of separate humus layers could explain this result. The interplay of plant-animal-soil interactions is likely to drive the ecosystem toward three alternative states supporting humus traditional classification between mull-mor-moder. HMFA revealed the importance of depth to explain this contrast among humus forms, using humus layers as diagnostic tools in both inert and living components. HMFA also showed contrast between unexploited and exploited parts of the forest, but the study of soil and vegetation indicate that this contrast does not only hold in forest management but also in geomorphology. RV-coefficients among the six groups of variables showed significant fauna-fauna relationships in almost all humus layers except Actinedida. Plant-soil interactions are not as strong as expected and are even weaker when the soil in question is deep. In addition, HMFA failed to show direct interactions between plant and soil fauna but, paradoxically, HMFA does suggest that indirect plant-fauna interactions are at the focus of the ecosystem strategy that leads to the differentiation of ecological niches within the forest mosaic. (c) 2012 Elsevier GmbH. All rights reserved.