9528 ‎(2)‎

• UCREFRP Catalog Number: 9528
• Author: Martinez, P. and N. P. Nibbelink.
• Title: Colorado Nonnative Fish Stocking Regulation Evaluation.
• USFW Year: 2004.
• USFW - Doc Type: Grand Junction, Laramie.
• Copyright Material: NO

at those scales. Specifically, we may expect a strong relationship at 200 m, but a
declining or non-existent relationship at 1000 m (Figure 4).
2. The cross-correlogram
Across-correlogram is a plot of the inter-site correlations as a function of
increasing separation distance between locations (Goovaerts 1998). Cross-correlograms
can be used in two ways. The most common application is to consider the relationship
between different attribute values (e.g. fish density and temperature) within a sample
dataset. An alternative, and the way in which we applied the cross-correlogram here,
involved focusing on correlations among a similar attribute (fish density, or index of
density) across two datasets. The correlation coefficient that is used here was Moran's I.
Moran's I is an extension of the cross-product correlation coefficient. Atypical pattern
in a correlogram is shown and described in Figure 5. We developed cross correlograms
1 between fish density in backwater locations and fish numbers sampled in floodplain
ponds. Theoretically, we would expect to see positive correlations at short distances, just
as with the scattergram if there were positive associations between pond fish densities
and backwater fish densities. The correlogram can also suggest the distance at which this
relationship no longer exists. When the correlation is not significantly different from
zero, we have reached the distance beyond which there is no longer any "influence" of
the source. Therefore, correlograms can be used to suggest spatial scales at which certain
patterns exist (Nibbelink 2002). The term "influence" is in quotations because the
correlogram does not necessarily imply causation. It is merely a suggestion of pattern.
3. The semivariogram
Semivariograms are often used in the geostatistical literature to describe spatial
patterns in terms of dissimilarity instead of similarity (as with the correlogram). A
semivariogram is the average dissimilarity between data separated by a distance (h). A
semivariogram is computed as half the mean squared difference between the attributes
(fish density) of every data pair (Goovaerts 1998). The primary advantage of the
semivariogram is that it is relatively easy to fit one of several types of models to the data
it generates. The fitting of a model can yield useful parameters called the range and the
sill, which describe components of the spatial relationship observed (Figure 6}.
RESULTS
Pond, Lake and Reservoir Locations, and Hydrography
Figure 7 shows the distribution of ponds, lakes and reservoirs above and below the 6,500-
foot contour in Colorado west of the Continental Divide. Out of 3,616 standing waters in the
database we compiled (Appendix II), 31 percent (1,104 waters) are located on or below the
6,500-foot contour (Table 2).
11