Tree species control and spatial structure of forest floor properties in amixed-species stand

The influence of tree species on forest floor properties (total N, P, K, Ca , and Mg, extractable K, Ca, and Mg, pH, basal respiration, potentially-min eralizable N, horizon thickness, bulk density, and loss on ignition) of mic rosites of a mixed-species stand was examined using a variation-paritioning method and redundancy analysis (RDA). The influence of eastern hemlock (Ts uga canadensis), American beech (Fagus grandifolia), red maple (Acer rubrum ), striped maple (Acer pensylvanicum), red oak (Quercus rubra), paper birch (Betula papyrifera), yellow birch (Betula alleghaniensis), and uncommon sp ecies was expressed with a tree species influence index (TSII) based on the size of the individual trees of each species and their distance from a mic rosite. Maps and Mantel correlograms were used to describe and quantify the spatial structure of forest floor and TSII variables. Trend surface analys is (TSA) and a neighbourhood matrix (NM), based on the mean values of fores t floor properties at neighbouring microsites, were used to incorporate a s patial component in the partitioning of the forest floor variation. TSII ex plained 29.6% of the forest floor variation of which about half was spatial ly structured and jointly expressed by the trend surface polynomial and the NM. The NM proved to be more efficient in capturing small-scale spatial pa tterns than the TSA. The main ecological trends observed for both the local and the total TSII effect were the influence of the beech-hemlock gradient on calcium, and the differential effect of striped maple and red maple on the amount of organic matter and associated nutrients. The relevance of the incorporation of a spatial component in the variation-partitioning of fore st floor data and the potential of this technique to deal with the complexi ty of natural mixed-forests are verified and discussed.