Seasonal boundary dynamics of a groundwater/surface-water ecotone

Ecology, Sept, 1998 by Brian G. Fraser, D. Dudley Williams

Classification of sites was made by two-way indicator species analysis (TWINSPAN, Hill 1979) following conversion of quantitative taxon abundance data into a qualitative format. Conversion was accomplished by categorizing each taxon and its abundance level into individual units termed pseudospecies. In the analysis, taxa with high abundance, for example, are included not only at the appropriate pseudospecies abundance level, but also at all lower levels. As a result any given taxon may be present in a data set numerous times (Furse et al. 1984). Further details pertaining to this and advantages associated with TWINSPAN are given by Furse et al. (1984) and Wright et al. (1984). based on cursory examination of our invertebrate density data, we applied three logarithmic abundance categories (each with a corresponding pseudospecies) to family-level identifications. Abundance categories 1 and 2 correspond to pseudospecies for which 1-9 and 10-99 individuals, respectively, were collected. Abundance category 3 corresponds to pseudospecies for which [greater than]100 individuals were collected. We elected to use three pseudospecies per taxon as we thought this adequately reflected the difference between rare and abundant families. The analysis was arrested following three dichotomies as beyond this level groupings were likely to be too small to be meaningful.

For TWINSPAN site groups (first TWINSPAN dichotomy only), mean values of individual chemical parameters that were significantly correlated with CA axes 1 or 2 (and therefore presumably related to differences in macroinvertebrate assemblage) were calculated. Means were compared using t tests (SPSS 6.1.3; Norusis 1995) following appropriate transformations to fulfil the assumptions of normality (Wilk-Shapiro) and homoscedasticity ([F.sub.max]).

TABLE 1. Mean hydraulic gradient (i, cm/cm) between the surface of
the river bed and a depth of 1 m below the bed surface in the Speed
River, February 1993 through October 1993. River cross section is
divided into two sections - midriver to Station 1 and midriver to
Station 9.

Sampling       Midriver to       Midriver to
period          Station 1         Station 9

Spring            0.08              0.04
Summer            0.16              0.07
Fall              0.10              0.06
Winter            0.12              0.08

For the multiple regression, an [Alpha] level of P [less than] 0.05 was used as the critical value for statistical significance. For the comparisons of TWINSPAN site group means, [Alpha] was corrected for multiple tests using the Bonferroni method (P [less than] [Alpha]/n, where n is the number of null hypotheses tested; Sokal and Rohlf 1981). We deemed it appropriate to use P [less than] 0.05 in the regression analysis, as this was employed purely as an exploratory tool to identify potentially important environmental variables. A more rigorous level of significance was adopted for the subsequent comparison of TWINSPAN site group means, which validated only those variables, identified by the regression analysis, likely to be of biological significance.