• Peatland ecology;
  • Principal coordinates of neighbour matrices (PCNM);
  • Spatial autocorrelation;
  • Spatial dependence processes;
  • Variation partitioning;
  • Vegetation structures


Question: What are the relative influences of environment and space in structuring the plant composition in a peatland complex?

Location: Lakkasuo, southern boreal zone, Finland.

Method: We used principal coordinates of neighbour matrices (PCNM) to model spatial structures in the plant composition of a peatland complex comprising ombrotrophic and minerotrophic, open and forested areas. We used redundancy analyses (RDA) and variation partitioning to assess the relative influences of chemical variables (peat and water characteristics), physical variables (hydrology, soil properties, shade), as well as broad-scale (>350 m) and medium-scale (100–350 m) spatial structures on vegetation assemblages.

Results: We identified five different significant spatial patterns circumscribing (1) the minerotrophic–ombrotrophic gradient; (2) dry ombrotrophic and wet minerotrophic areas; (3) open and shaded areas; (4) dry open/shaded and wet patches within the ombrotrophic areas; and (5) dry open patches and dry forested patches. With spatial structures and environmental variables, we were able to model 30% of the variability in plant composition in the peatland complex, 13% of which was attributable to spatial structures alone.

Conclusions: We demonstrated that in the peatland complex, the spatial dependence processes were more important at the broadest scale, and found that patterns at a medium scale might reflect finer-scale patterns that were not investigated here. Spatial autocorrelation in vegetation composition in the peatland complex appeared to be driven by Sphagnum species. Our results emphasize that spatial modelling should be routinely implemented in studies looking at species composition, since they significantly increase the explained proportion of variance.