Locating Discontinuities along Ecological Gradients

摘要: Cost-efficient ecological and natural resource surveys need: (1) flexible, logistically simple, statistically sound sampling methods, (2) sensitive, computationally ecologically robust data analysis methods. "Gradsects" (gradient-oriented transects) have recently been shown to be a more efficient method than random using nongradient-oriented transects, especially when surveying large, biologically diverse areas (Gillison Brewer 1985). If biotic (e.g., vegetation) is oriented along defined environmental gradient, the purpose classification, then techniques are needed. This note briefly reviews appropriate in order identify ecotones or boundaries gradsects. vegetation survey inventory made by at fixed intervals steep units plots, quadrats, lines) may traverse different zones reflecting underlying topographic, soil) discontinuities. And, if set obtained each unit multivariate many species observed), these can used locate separating gradsect. A simple but for locating between communities sampled gradsects computation of moving split-window distances, procedure described Whittaker (1960) who was working on "... quantitative methods which relative discontinuities . .. might objectively revealed from transect tables" his Siskiyou Mountains gradsect data. The basic is: obtain gradient; bracket block positions into window preassigned width (i.e., including two adjacent positions, as calculating averages; Legendre 1983:344); (3) split this samples equal groups; (4) average variate within group; (5) compute distance dissimilarity groups (Legendre 1983:Chapter 6); (6) move one position further another distance; (7) after end other, with computed midpoint position, plot distances (ordinate) against (abscissa). Sharp, high peaks location community zones. For continuous gradations expected graph would ". points generally same level, some zig-zag up down due chance variations stand composition" (Whittaker 1960). As an example, Wierenga et al. (1987) both squared Euclidean (SED) Hotelling-Lawley trace F values (HLF) examine coincidence soil northern Chihuahuan Desert. Seven were (Fig. 1), strongly coincident eight series These results substantiated earlier vegetationsoil studies area (Stein Ludwig 1979). 10 smoother pattern peaks, whereas 2 (the SED positions) had greater sample-tosample noise, expected. Although used, several other coefficients examined gave similar results. Any association, similarity, distance, dependence 6) could used. When defining HLF, reduced original smaller principal components analysis, Webster (1973). Relative SED, HLF calculation powerful, complex limited fewer variables components) wider widths (Wierenga 1987). Many SEDs, larger number likely produce accurate distance. large must also obscure because include (Webster 1973). In our varying 6 did not appreciably affect interpretation boundary locations, only emphasis certain peaks. Abrupt shifts types evident high, narrow 5 Fig. gradual ecotone fuzzy lower 3 4 1). technique above similar, differs inP detail from, literature. Beals (1969) coefficient percentage difference 1983:201) compare successive segments five (a 10) altitudinal