Research paperLitter and soil-related variation in functional group abundances in cacao agroforests using structural equation modeling

- SEM is a suitable tool for the understanding the relationship complexity among faunal groups within the litter-soil system.
- The interconnection theory between organisms with their niches should be presented in more than one structural model.
- The litter attributes acted as a regulator of soil-fauna functional groups in mature cacao agroforest.

The aim of this study was to apply the method of structural equation modeling using latent variables (constructs) with multi-indicators to test hypothesized interaction models among functional groups of fauna and attributes of litter-soil system in cacao agroforests in the south of Bahia, Brazil. The hypothesized structural model was composed of four constructs: the litter-attributes, litter-fauna, soil-attributes, and soil-fauna. The models tested were able to distinguish between the direct and indirect relationships of the faunal groups and attributes of litter-soil system. The results have shown that the interconnection theory between fauna and attributes of litter-soil system should be presented in more than one structural equation model. The models built from the re-specifications showed a strong theoretical base and data fitted, well representing the fauna as a latent variable (construct) theory within the litter-soil system of cacao agroforests. The overall fit of the hypothesized models highlighted the theoretical plausibility of complex interconnection between soil and litter fauna. Both are influenced by the attributes of their niches. Litter-attributes construct composed only by the indicators lignin, cellulose and polyphenols showed a recalcitrant predominant nature. This low litter nutritional quality would negatively affect the faunal groups therein. The negative relation between litter-attributes and soil-attributes suggest that low-quality litter contributed little to the improvement of soil fertility, particularly related to the level of acidity. The negative relationship between soil-attributes and soil-fauna shows that highly acid and low-fertility soils would promote some restriction on the activity of soil fauna groups. The structural equation modeling with latent variables proved to be a suitable tool for the understanding of the complexity of the relationships between faunal groups within the tested litter-soil system, indicating how the changes in a latent composition would affect other latents. It important to mention that when the approach is about organisms in the litter-soil system, caution is advised when trying, in a mechanistic way, to determine the factors influencing the interconnection of these organisms with their niches, due to the inherent heterogeneity of both litter and soil.