8172

• UCREFRP Catalog Number: 8172
• Author: Jacobi, G. Z., J. E. Sublette, S. J. Herrmann, D. E. Cowley and M. D. Hatch.
• Title: Final Report
• USFW Year: 1998.
• USFW - Doc Type: Investigations of an Index of Biotic Integrity in New Mexico.
• Copyright Material: NO

use designation, and substrate composition _ Statistical AnalySeS
are given in Table 1.
Chironomid data, analyzed as
presence/absence for this report, were
available for 20 upper Rio Grande sites, 18
of which had data for fish and 17 of which
had data for environmental variables. Data
for benthic macroinvertebrate taxa, analyzed
as- presence/absence-, *were- -available-for 30
upper Grande sites, all of which had
environmental data available. Fish data
were obtained for 32 sites in the upper Rio
Grande; however, data were analyzed for
only 28 of the sites for which environmental
data were available. Analyses for fish were
made on presence/absence data and on
relative abundance data. Presence/absence
data has great utility in assessing faunal
regions (e.g., Hawkes et al., 1986;
Malmqvist and Maki, 1994). Data for
environmental factors, chironomid species
distribution, benthic macroinvembrate
distribution, and fish distribution and
abundance are given in Appendix 2.
Data were manipulated with SAS for the
Windows operating environment (SAS
Institute, Inc., 1990, 1993). Multiple
samples for a site were combined to obtain
single composite samples per site
(Malmgvist and Maki, 1994). To create
data files having sites as rows and species as
columns, data were manipulated with a
combination of SAS and FORTRAN
programs (Appendix 3). To create data
files for use in CANOCO (ter Braak,
1987-1992), SAS programs were used to
reformat species data into Cornell
condensed format and environmental data
into CANOCO full format.
Defining Aquadc Ecoregions with
Environmental Data
To reduce the dimensionality and to
circumvent problems of colinearity of some
of the 22 environmental variables, principal
components analysis (Johnson and Wichern,
1982) was performed. Principal
components analysis is a multivariate
ordination technigne that is useful for
reducing the variation in a large set of
. individual variables to a smaller set of
multivariate combinations of variables
(Corkwn and Ciiborowski, 1988; Stewart
and L(an, 1994). principal components are
uncorrelated axes (orthogonal, or, at 90
degree angles) of variation in multivariate
space. The principal components are
obtained in such a way that the first axis
represents the direction of greatest variation
in multivariate space, the second axis
accounts for the neat most variation, and so
forth. When the number of sample sites is
greater than or equal to the number of
variables, the number of principal
components is equal to the number of sites.
Principal components with eigenvalues
(amount of variance explained) greater than
one were extracted from the environmental
correlation matrix for sites using the
FACTOR procedure of SAS (SAS Institute,
Inc., 1989x). Use of the correlation matrix
was equivalent to standardization of
variables and it eliminated scaling problems
that may arise with a covariance matrix
when all of the variables are not on the
same scale of measurement. The set of
important (eigenvalues greater than one)
principal components were then subjected to
ai
d