Numerical Analysis of a Chronosequence, Including the Development of a Chronofunction

Abstract

Multivariate techniques were applied to 11 chemical and physical properties of soil samples collected along a prograded beach chronosequence, located on the west coast of Vancouver Island, British Columbia. The samples were taken from three horizons at each of seven sites. The 21 samples were considered as independent observations. Application of principal component analysis with a normal varimax rotation gave highly interpretable results. The first component was readily identified with podzolic pedogenic processes, whereas the second component appeared to relate closely to sea spray input. A nonlinear optimization procedure was used to fit the first principal component to an empirical equation incorporating a logistic term for time and an exponential term for depth. A correlation coefficient value of 0.99 was obtained. The notion of a pedogenic damping depth analagous to a thermal damping depth is suggested.

Transformation of categorical field soil morphological properties into numerical properties for the study of chronosequences

“…In this case this probably indicates a certain degree of non-linearity in its expression. To support this result it should be noted that non-linear relationships in chronosequences have been demonstrated using a combination of PCA and non-linear regression by Sondheim et al (1981) and Sondheim and Standish (1983). 4.…”

“…Following this transformation, the function obtained with the soil evolution is usually linear and the mutual distances between adjacent categories are usually equivalent throughout the whole range of the variable in question. This probably results in a loss of information with regard to the soil system, which exhibits some non-linear characteristics, as propounded by Schaetzl et al (1994) or Phillips (1993), from a theoretical point of view, or, from a numerical point of view, by Sondheim et al (1981), in the case of the development of the soil as a whole over time.…”

Applying Principal Component Analysis to Soil-Landscape Research-Quantifying the Subjective

“…A survey of the literature in which PCA/FA are used to investigate soil-landscape and soil-site relationships revealed a distressing lack of consideration for important PCA/FA analytical procedures (Arp, 1984;Campbell et al, 1970;Fourt etal, 1971;La Bastide and Van Goor, 1970;Litaor et al, 1989;Nortcliff, 1978;Ovalles and Collins, 1988;Page, 1976;Richardson and Bigler, 1984;Rowe and Sheard, 1981;Sarkar et al, 1966;Severson, 1981;Sondheim et al, 1981;Sondheim and Standish, 1983;Williams and Rayner, 1977). Since clear guidelines for determining the "optimum" final PCA/FA solution currently do not exist, we believe that the burden of proving that reported results are optimum lies with the investigators.…”

Relative Ages of Wetland and Upland Surfaces as Indicated by Pedogenic Development

“…Use of both the Bilzi-Ciolkosz and Harden indices is now well established, as is the more general principle of soil chronofunctions (Birkeland, 1984;Meixner and Singer, 1981;Holliday, 1988;Bockheim, 1980;Sondheim et al, 1981;Johnson and Watson-Stegner, 1987). Previous uses of soil development indices have been in site-specific geomorphic or pedologic studies, or in conjunction with other methods to establish dates or chronologies.…”