Tree species and parent material both influence spatial variation in soil resource availability, and biodiversity and ecosystem functioning theory suggests that species may interact to produce synergistic or antagonistic effects on soil nutrient pools. We investigated these ideas in stem-mapped stands at four sites in northern hardwood forests in NW Connecticut, USA, using tree neighborhood configuration and elemental composition of parent material to predict calcium, magnesium, and potassium availability on the soil exchange complex. The tree species investigated included several with affinities for high-calcium soil (i.e., calcicoles: sugar maple and white ash) and others which are often restricted to low-calcium sites (calcifuges: northern red oak, eastern hemlock, American beech, and red maple). Elemental oxides of the sand-sized fraction of mineral soil were measured with X-ray fluorescence spectroscopy. We used neighborhood methods to ask several questions. First, what is the relative strength of species (or functional group) effect versus local physical factors on nutrient availability once parent material has been taken into account? Second, is there variation among species or functional groups in their effects on nutrient availability? Third, are there synergistic or antagonistic effects on nutrient availability among species or functional groups? Maximum likelihood principles were used to judge among models embodying these questions.
Results/Conclusions
Exchangeable Ca and Mg were correlated with parent material content of these elements (R2 = 0.31 and 0.23), but K was not (R2=0.03). Incorporating site (n=4) in the parent-material-dependent models improved them more than species, but using both site and species resulted in much better models for Ca and Mg. Species effects on exchangeable Ca and Mg (normalized by parent material) diverged slightly from prior classifications as calcicoles or calcifuges: ash, hemlock and the maples had positive effects and oak and beech had negative ones. Using these species effects as the basis for functional groups resulted in more parsimonious models. There was a mild antagonistic effect on Ca, i.e., availability was slightly lower in the presence of both hi- and low-effect functional groups than would be predicted from either group alone. The AIC-best models for Ca (R2 = 0.55) and Mg (R2 = 0.43) incorporated elemental content, site, and functional group. The relative strength of influence of these soil-forming factors on soil resource availability was parent material > local factors > functional group identity > diversity-function effects. Neighborhood methods provide a powerful new tool for resolving long-standing questions about tree-environment relationships.