82
Basic biogeography: estimating biodiversity and mapping nature
purposes such as land management and planning,
which stresses functional properties of ecosystems in
preference to composition.
The use of ecoregional schemes in conservation has
been growing and it follows that tests of ecoregional
schemes that explore the extent to which the boundaries between ecoregional units mark meaningful
ecological boundaries are thus of increasing interest
(Magnusson, 2004 ). Magnusson (2004) argued that if
distance affects similarity in ecological attributes, any
arbitrary division of the landscape will result in objects
within a unit being signifi cantly different from objects
in other divisions. It follows that rigorous tests of ecoregions have to be undertaken with due attention to artefacts arising from spatial autocorrelation. He suggested
testing for natural breaks in species ’ distributions
across ecoregional boundaries to test how well they
refl ect divisions in ecological composition.
This form of analysis is likely to be informative from
a conservation perspective even if it is, in essence,
using a compositionalist criterion to judge a scheme
focused on functional properties. Just such an analysis
was undertaken by Karanth et al . (2006) for birds in
North American ecoregions. Supporting the validity of
ecoregions, they reported lower species richness,
higher local turnover and higher extinction probabilities at the edges of ecoregions, but not for all
regions.
A less encouraging test was provided by Riitters et al .
(2006) who compared ecoregional boundaries (derived
both from Omernik ’ s and Bailey ’ s work) with a detailed
land - cover map of the USA. They report that whereas
ecoregions accounted for 65 per cent to 75 per cent of
the total variance of per cent agriculture and per cent
forest, this is unsurprising because dominant land -
cover is included in practice in ecoregional defi nitions
within these schemes. By contrast, ecoregions
explained only 13 to 34 per cent of the variance of
the other seven landscape - level pattern indices they
examined.
They concluded that the ecoregional stratifi cations
tested were not effective mapping units for land - cover
pattern, because within - unit variance of land - cover
pattern was typically two to four times larger than
between - unit variance. Notwithstanding the equivocal
nature of these recent tests, ecoregional schemes have
found increasing application in conservation science.
The most important such scheme, the WWF - ecoregions
approach, is discussed in Chapter 5 .
4.5.2 Ecoregions
If biome schemes are important foundations for several
strategic conservation schemes discussed in Chapter 5 ,
they are rather coarse units for landscape level application, while fl oristic (compositionalist) approaches for
identifying natural communities at fi ner scales of
analysis (above) are subject to criticism for artifi cially
imposing a rigid structure on what are typically rather
variable assemblages in space and through time
(Chapter 3 ). Moreover, these approaches are rather
dependent on prevailing views in phytosociology
(vegetation science) (Carri ó n & Fern á ndez, 2009 ).
The problem, then, is how to classify ecosystems and
habitats independently of composition, at fi ne scales of
analysis and for purposes such as land management. It
was to this end that ecoregional schemes were developed by R.G. Bailey (e.g. 1996 , 1998 ) and J.M. Omernik
(1987) , following the coining of the term by Crowley
(1967) . Omernik ( 1987 , p. 123) defi nes ecoregions as
‘ regions of relative homogeneity in ecological systems
or in relationships between organisms and their
environments ’ .
The ecoregion approach, as exemplifi ed by Bailey ’ s
cited works, essentially follows a controlling factors
methodology, starting at the coarsest (global and continental) scales with climatic parameters (much like
Figure 4.11 ), by which the major Ecoregional Divisions
(units of the order of 10
5 km
2
) are recognized. It
then recognizes major azonal topographical features
(mountain chains), before proceeding to successively
fi ner landscape mosaics (10
3 down to 10 km
2
) in which
edaphics and human transformation infl uence ecosystem form and function.
The approach of focusing on controlling factors is
underpinned by the concept that ecosystems do not
function independently from one another, but that
smaller systems (habitat patches) are connected by
ecological fl ows and linkages, movements of water,
nutrients and animals into successively large units.
Hence, the approach stresses not just fl oristic similarity
or consistency between sites, but also the importance
of internal linkages and exchanges of ‘ information ’
within the territory of a single ecoregional unit. Each
unit at a particular point in the hierarchy is linked by
cross - boundary exchanges to other units at the same
level, rather as small streams join together to form
higher order drainage basins. Hence, the approach provides a multi - layered scheme for classifying regions, for
Basic biogeography: estimating biodiversity and mapping nature
purposes such as land management and planning,
which stresses functional properties of ecosystems in
preference to composition.
The use of ecoregional schemes in conservation has
been growing and it follows that tests of ecoregional
schemes that explore the extent to which the boundaries between ecoregional units mark meaningful
ecological boundaries are thus of increasing interest
(Magnusson, 2004 ). Magnusson (2004) argued that if
distance affects similarity in ecological attributes, any
arbitrary division of the landscape will result in objects
within a unit being signifi cantly different from objects
in other divisions. It follows that rigorous tests of ecoregions have to be undertaken with due attention to artefacts arising from spatial autocorrelation. He suggested
testing for natural breaks in species ’ distributions
across ecoregional boundaries to test how well they
refl ect divisions in ecological composition.
This form of analysis is likely to be informative from
a conservation perspective even if it is, in essence,
using a compositionalist criterion to judge a scheme
focused on functional properties. Just such an analysis
was undertaken by Karanth et al . (2006) for birds in
North American ecoregions. Supporting the validity of
ecoregions, they reported lower species richness,
higher local turnover and higher extinction probabilities at the edges of ecoregions, but not for all
regions.
A less encouraging test was provided by Riitters et al .
(2006) who compared ecoregional boundaries (derived
both from Omernik ’ s and Bailey ’ s work) with a detailed
land - cover map of the USA. They report that whereas
ecoregions accounted for 65 per cent to 75 per cent of
the total variance of per cent agriculture and per cent
forest, this is unsurprising because dominant land -
cover is included in practice in ecoregional defi nitions
within these schemes. By contrast, ecoregions
explained only 13 to 34 per cent of the variance of
the other seven landscape - level pattern indices they
examined.
They concluded that the ecoregional stratifi cations
tested were not effective mapping units for land - cover
pattern, because within - unit variance of land - cover
pattern was typically two to four times larger than
between - unit variance. Notwithstanding the equivocal
nature of these recent tests, ecoregional schemes have
found increasing application in conservation science.
The most important such scheme, the WWF - ecoregions
approach, is discussed in Chapter 5 .
4.5.2 Ecoregions
If biome schemes are important foundations for several
strategic conservation schemes discussed in Chapter 5 ,
they are rather coarse units for landscape level application, while fl oristic (compositionalist) approaches for
identifying natural communities at fi ner scales of
analysis (above) are subject to criticism for artifi cially
imposing a rigid structure on what are typically rather
variable assemblages in space and through time
(Chapter 3 ). Moreover, these approaches are rather
dependent on prevailing views in phytosociology
(vegetation science) (Carri ó n & Fern á ndez, 2009 ).
The problem, then, is how to classify ecosystems and
habitats independently of composition, at fi ne scales of
analysis and for purposes such as land management. It
was to this end that ecoregional schemes were developed by R.G. Bailey (e.g. 1996 , 1998 ) and J.M. Omernik
(1987) , following the coining of the term by Crowley
(1967) . Omernik ( 1987 , p. 123) defi nes ecoregions as
‘ regions of relative homogeneity in ecological systems
or in relationships between organisms and their
environments ’ .
The ecoregion approach, as exemplifi ed by Bailey ’ s
cited works, essentially follows a controlling factors
methodology, starting at the coarsest (global and continental) scales with climatic parameters (much like
Figure 4.11 ), by which the major Ecoregional Divisions
(units of the order of 10
5 km
2
) are recognized. It
then recognizes major azonal topographical features
(mountain chains), before proceeding to successively
fi ner landscape mosaics (10
3 down to 10 km
2
) in which
edaphics and human transformation infl uence ecosystem form and function.
The approach of focusing on controlling factors is
underpinned by the concept that ecosystems do not
function independently from one another, but that
smaller systems (habitat patches) are connected by
ecological fl ows and linkages, movements of water,
nutrients and animals into successively large units.
Hence, the approach stresses not just fl oristic similarity
or consistency between sites, but also the importance
of internal linkages and exchanges of ‘ information ’
within the territory of a single ecoregional unit. Each
unit at a particular point in the hierarchy is linked by
cross - boundary exchanges to other units at the same
level, rather as small streams join together to form
higher order drainage basins. Hence, the approach provides a multi - layered scheme for classifying regions, for