Design-based mapping of plant species presence, association, and richness by nearest-neighbour interpolation

The difference between potential and actual distribution of species is emphasized, pointing out the ecological importance of providing maps that depict the actual species presence on the study region. Owing to the impossibility of performing complete surveys over large areas, the presence/absence of species at a pre-fixed spatial grain is estimated for any location of the study region from the presences/absences recorded within plots centred at sample locations and having the same grain. Estimation is performed in a design-based framework by means of the well-known nearest-neighbour interpolator. Association and richness maps are obtained as product and sum of the presence maps of single species. The design-based asymptotic unbiasedness and consistency of these maps are theoretically proven and pseudo-population bootstrap estimators of their precision are proposed and discussed. Coverage of presence, association maps and related overlapping indexes are estimated, as usual, by counting the sample locations within the study region. A simulation study is performed on a real community of 302 tree species settled in a 50-ha rectangle in the lowland tropical moist forest of Barro Colorado Island (BCI), central Panama, to check the finite-sample performance of the proposal. A case study for estimating the presence map and the association of holly oak and white violet in the Montagnola Senese (Central Italy) is reported. Technical details are contained in the appendices.