Canonical Correspondence Analysis

34

Site

unmodified

recently

dredged

1

0

0

2

1

0

3

1

0

4

0

1

5

0

1

Site 1 was dredged more than 10 years ago, Sites 2 & 3 are unmodified and sites 4 &

5 have been recently dredged. Note that we do not need to explicitly include the

category  dredged more than 10 years ago . Indeed, you must not include it if you

are to avoid the environmental variables becoming linear combinations of each other.

See also:

Selecting Environmental variables

Linear combinations of environmental variables

Transforming environmental variables

Circular environmental variables

7.5

Transforming environmental variables

Palmer (1993)

 recommends logarithmic transformations of soil nutrient

concentrations when undertaking vegetational analysis "because a 1 unit difference in

nutrient concentration is probably much more important at low concentrations than it

is at high concentrations".

Linear transformations of variables will have no effect on the results of a CCA.

Therefore, changing the units of a variable (e.g. centimeters to millimeters) or the

base point (e.g. elevation above a fixed point to elevation above sea level) will have

no effect.

See also

Data transformations

Selecting Environmental variables

Linear combinations of environmental variables

Dummy Environmental variables

7.6

Multicollinearity

If the environmental variables are highly correlated then results obtained by any

multiple regression method can be highly unreliable. ECOM tests for appreciable

collinearity between the environmental variables and will give a warning message if

appreciable levels of collinearity are present.

Copyright 2004  PISCES Conservation Ltd