7
General Data Collection and
Sampling Design Considerations for
Integrated Regional Ecological Assessments
Patrick S. Bourgeron, Hope C. Humphries, and Mark E. Jensen
7.1 Introduction
Large-scale data collection is the initial observational phase of integrated regional ecological assessments (IREAs). The data collection methodology determines to a large extent the accuracy and
precision of all subsequent analyses, such as pattern recognition (Bourgeron and Jensen, 1994;
Bourgeron et aI., 1994a; Dale and O'Neill, 1999).
The data analysis and interpretative phases of
IREAs often tend to be emphasized by scientists
over formal data collection procedures (see the case
studies in Chapters 30 through 34; for a very formal approach to survey design, see the US-EPA
Ecological Monitoring and Assessment Program,
O'Neill et al., 1994; Kapner et al., 1995). The lack
of formal, consistent data collection protocols for
IREAs is unfortunate, because the quality of the
characterization of a region and the quality of subsequent interpretations depend on the quality of the
data, both in terms of the thoroughness of coverage and the type of information collected.
Several characteristics of IREAs make the use of
classical sampling and survey design methodologies (see Chapter 6) difficult to implement at best
(see discussion on aspects of the problem in Legendre and Legendre, 1998, p. 16): (1) IREAs usually take place in large areas (e.g., the 58 million
ha of the interior Columbia River basin, ICRB),
which may contain inaccessible locations; (2) they
aim at characterizing the range of variability in the
Patrick S. Bourgeron and Hope C. Humphries wish to
acknowledge partial funding provided by a Science to
Achieve Results grant from the U.S. Environmental Protection Agency ("Multi-scaled Assessment methods:
Prototype Development within the Interior Columbia
Basin").
92
biotic and abiotic components of ecosystems, as
well as biotic-abiotic interactions (e.g., plant communities in relation to environmental gradients)
(see Chapter 22), regardless of their relative abundance (e.g., rare or common); (3) they aim at describing multiple components of ecosystems at
multiple spatial scales.; (4) the data collected are
generally used for multiple purposes; and (5)
IREAs are often conducted under strict deadlines
and budgetary and political constraints, which seriously limit the resources and time allocated to de
novo sampling design and field survey (Treweek,
1999).
Therefore, the data collection effort of IREAs
should address the following considerations: (1) the
analysis of spatial (or temporal) structures is of primary interest; (2) spatial (or temporal) scale is a
key concept; (3) as a result of the hierarchical structure of ecosystems (see Chapers 2 and 3), spatial
heterogeneity is functional and not the result of
some random, noise-generating process; (4) the recovery of all patterns, common and rare, and of
their intrinsic range of variability is often prohibitively costly under the requirements of classical
sampling designs; and (5) existing data usually
have to be utilized because of time and resource
limitations. The statistical methods and models
used in IREAs also must include realistic assumptions about the spatial structuring of ecosystems
(Legendre and Legendre, 1998).
In collecting data for IREAs, the emphasis on
the multiscaled spatial heterogeneity of multiple
ecosystem components contrasts with the purpose
of classical sampling design, which is to determine
unbiased estimates of the means of variables over
entire popUlations (Austin and Heyligers, 1991; Legendre and Legendre, 1998; also see Chapter 6). A
requirement of classical sampling design is the
General Data Collection and
Sampling Design Considerations for
Integrated Regional Ecological Assessments
Patrick S. Bourgeron, Hope C. Humphries, and Mark E. Jensen
7.1 Introduction
Large-scale data collection is the initial observational phase of integrated regional ecological assessments (IREAs). The data collection methodology determines to a large extent the accuracy and
precision of all subsequent analyses, such as pattern recognition (Bourgeron and Jensen, 1994;
Bourgeron et aI., 1994a; Dale and O'Neill, 1999).
The data analysis and interpretative phases of
IREAs often tend to be emphasized by scientists
over formal data collection procedures (see the case
studies in Chapters 30 through 34; for a very formal approach to survey design, see the US-EPA
Ecological Monitoring and Assessment Program,
O'Neill et al., 1994; Kapner et al., 1995). The lack
of formal, consistent data collection protocols for
IREAs is unfortunate, because the quality of the
characterization of a region and the quality of subsequent interpretations depend on the quality of the
data, both in terms of the thoroughness of coverage and the type of information collected.
Several characteristics of IREAs make the use of
classical sampling and survey design methodologies (see Chapter 6) difficult to implement at best
(see discussion on aspects of the problem in Legendre and Legendre, 1998, p. 16): (1) IREAs usually take place in large areas (e.g., the 58 million
ha of the interior Columbia River basin, ICRB),
which may contain inaccessible locations; (2) they
aim at characterizing the range of variability in the
Patrick S. Bourgeron and Hope C. Humphries wish to
acknowledge partial funding provided by a Science to
Achieve Results grant from the U.S. Environmental Protection Agency ("Multi-scaled Assessment methods:
Prototype Development within the Interior Columbia
Basin").
92
biotic and abiotic components of ecosystems, as
well as biotic-abiotic interactions (e.g., plant communities in relation to environmental gradients)
(see Chapter 22), regardless of their relative abundance (e.g., rare or common); (3) they aim at describing multiple components of ecosystems at
multiple spatial scales.; (4) the data collected are
generally used for multiple purposes; and (5)
IREAs are often conducted under strict deadlines
and budgetary and political constraints, which seriously limit the resources and time allocated to de
novo sampling design and field survey (Treweek,
1999).
Therefore, the data collection effort of IREAs
should address the following considerations: (1) the
analysis of spatial (or temporal) structures is of primary interest; (2) spatial (or temporal) scale is a
key concept; (3) as a result of the hierarchical structure of ecosystems (see Chapers 2 and 3), spatial
heterogeneity is functional and not the result of
some random, noise-generating process; (4) the recovery of all patterns, common and rare, and of
their intrinsic range of variability is often prohibitively costly under the requirements of classical
sampling designs; and (5) existing data usually
have to be utilized because of time and resource
limitations. The statistical methods and models
used in IREAs also must include realistic assumptions about the spatial structuring of ecosystems
(Legendre and Legendre, 1998).
In collecting data for IREAs, the emphasis on
the multiscaled spatial heterogeneity of multiple
ecosystem components contrasts with the purpose
of classical sampling design, which is to determine
unbiased estimates of the means of variables over
entire popUlations (Austin and Heyligers, 1991; Legendre and Legendre, 1998; also see Chapter 6). A
requirement of classical sampling design is the