The distribution of diversity: challenges and applications
111
generations of the hotspots scheme reveals that the
original ten areas were each long, thin areas, comprising a stretch of the Atlantic coastal forest of Brazil, a
thin band in the uplands of Western Amazonia, the
eastern mountains and coastal plain of Madagascar
and so on. The rule base for deciding where to draw the
borders of these units was not specifi ed (see Myers,
1988, 1990 ).
As the number of hotspots has expanded, so too
have the geographical extents of most of these original
small areas, at least as represented on the maps. The
most dramatic expansion in the size of the envelopes
comes with the CI - 2004 scheme, in which the problem
of the underlying map base has now been partially
addressed by the pragmatic step of aligning hotspot
boundaries with WWF Ecoregions (themselves an
amalgam of pre - existing and novel compositionalist
and functionalist classifi cations).
‘ Finally, delineating hotspots is by no means an exact
science. It requires that a line – that might be easily
discernible or rather vague on the ground – must be
drawn to represent a transition between two habitats. The map of Ecoregions developed by the World
Wildlife Fund - US is now the most widely used system
for such bioregional classifi cation. In order to facilitate analysis, interoperability, and collaboration, we
have therefore gone to considerable lengths to ensure
that … the boundaries of the hotspots (and those of
the high biodiversity wilderness areas) correspond
directly to those of the World Wildlife Fund - US
Ecoregions. ’
( www.biodiversityhotspots.org/xp/hotspots/
hotspotsscience/Pages/hotspots_revisited.aspx )
In addition to expanding the size of envelopes in the
CI - 2004 framework, the number of hotspots has
expanded to 34. One important goal was to greatly
expand the coverage of islands of high biodiversity
value (especially in the Pacifi c) that might otherwise
have ‘ slipped through the net. ’ This is a pragmatic
expansion, described as such by Mittermeier et al .
(2004) , who acknowledge that the fl oristic affi liations
and boundaries of the resulting hotspot units are
sometimes tenuous.
Taking the Mediterranean as an example, neither
the 1988 nor the 1990 scheme featured a
Mediterranean hotspot, but it is included in the CI versions. This hotspot, as of the CI - 2004 version, comprises the major part of the Iberian Peninsula (but not
contained 35 per cent of the terrestrial vertebrate
species of the world as endemics, while the 2004
version is claimed to embrace over 50 per cent of the
world ’ s plant species and 42 per cent of terrestrial vertebrate species as endemics (Mittermeier et al ., 2004 ).
The second criterion presupposes that we are able to
assess general drivers of species loss as a function of
habitat loss, and that we can determine accurate baseline data for pristine habitats. In addition to the two
stated criteria, a third key determinant of the outcome
of the hotspots analysis is how the underlying areas
are determined in the fi rst place.
As the CI hotspots scheme is so important and infl uential, it is especially relevant to evaluate the scientifi c
merits of the scheme critically. The diversity of sources
used in the analyses makes it hard to assess independently the quality and comparability of the diversity and
habitat loss data used in the hotspots analyses. As discussed in Chapter 4 , for plants we still lack systematic
species range maps for most of the globe, with knowledge of plant species diversity regarded as poor or very
poor across large parts of South America, Africa, Asia
and Australia (Kier et al ., 2005 ). Similarly, the assessment of how much of a region ’ s primary vegetation
cover has been lost can be highly dependent on assumptions made about the nature of the ‘ original ’ vegetation cover (see Chapter 3 ).
In practice, some commonly assumed baselines for
pre - Anthropocene states are based on false assumptions about the degree to which a region originally
supported forest habitats (Virah - Sawmy et al ., 2010 ),
while on the other hand some highly biodiverse areas
of tropical forest have been found to be underlain by
evidence of once dense human populations and former
episodes of forest clearance (e.g. Brncic et al ., 2007 ).
In addition, estimates of habitat conversion, particularly with regard to the tropical forest regions, have
been shown to produce highly inconsistent outcomes.
Surveys conducted at different points in time, or by
different teams, have produced confl icting results
(Grainger, 2008 ). However, it would be fair to counter
that any alternative metric of threat would encounter
some operational diffi culties, and that the use of the
70 per cent threshold provides a reasonable fi rst
approximation.
A further concern, as with all such analyses (Chapter
4 ), is the problem of deciding what constitutes the
most appropriate underlying map of nature, i.e. the
appropriate geographical units to be used in the analysis. Examination of the output maps from the four
111
generations of the hotspots scheme reveals that the
original ten areas were each long, thin areas, comprising a stretch of the Atlantic coastal forest of Brazil, a
thin band in the uplands of Western Amazonia, the
eastern mountains and coastal plain of Madagascar
and so on. The rule base for deciding where to draw the
borders of these units was not specifi ed (see Myers,
1988, 1990 ).
As the number of hotspots has expanded, so too
have the geographical extents of most of these original
small areas, at least as represented on the maps. The
most dramatic expansion in the size of the envelopes
comes with the CI - 2004 scheme, in which the problem
of the underlying map base has now been partially
addressed by the pragmatic step of aligning hotspot
boundaries with WWF Ecoregions (themselves an
amalgam of pre - existing and novel compositionalist
and functionalist classifi cations).
‘ Finally, delineating hotspots is by no means an exact
science. It requires that a line – that might be easily
discernible or rather vague on the ground – must be
drawn to represent a transition between two habitats. The map of Ecoregions developed by the World
Wildlife Fund - US is now the most widely used system
for such bioregional classifi cation. In order to facilitate analysis, interoperability, and collaboration, we
have therefore gone to considerable lengths to ensure
that … the boundaries of the hotspots (and those of
the high biodiversity wilderness areas) correspond
directly to those of the World Wildlife Fund - US
Ecoregions. ’
( www.biodiversityhotspots.org/xp/hotspots/
hotspotsscience/Pages/hotspots_revisited.aspx )
In addition to expanding the size of envelopes in the
CI - 2004 framework, the number of hotspots has
expanded to 34. One important goal was to greatly
expand the coverage of islands of high biodiversity
value (especially in the Pacifi c) that might otherwise
have ‘ slipped through the net. ’ This is a pragmatic
expansion, described as such by Mittermeier et al .
(2004) , who acknowledge that the fl oristic affi liations
and boundaries of the resulting hotspot units are
sometimes tenuous.
Taking the Mediterranean as an example, neither
the 1988 nor the 1990 scheme featured a
Mediterranean hotspot, but it is included in the CI versions. This hotspot, as of the CI - 2004 version, comprises the major part of the Iberian Peninsula (but not
contained 35 per cent of the terrestrial vertebrate
species of the world as endemics, while the 2004
version is claimed to embrace over 50 per cent of the
world ’ s plant species and 42 per cent of terrestrial vertebrate species as endemics (Mittermeier et al ., 2004 ).
The second criterion presupposes that we are able to
assess general drivers of species loss as a function of
habitat loss, and that we can determine accurate baseline data for pristine habitats. In addition to the two
stated criteria, a third key determinant of the outcome
of the hotspots analysis is how the underlying areas
are determined in the fi rst place.
As the CI hotspots scheme is so important and infl uential, it is especially relevant to evaluate the scientifi c
merits of the scheme critically. The diversity of sources
used in the analyses makes it hard to assess independently the quality and comparability of the diversity and
habitat loss data used in the hotspots analyses. As discussed in Chapter 4 , for plants we still lack systematic
species range maps for most of the globe, with knowledge of plant species diversity regarded as poor or very
poor across large parts of South America, Africa, Asia
and Australia (Kier et al ., 2005 ). Similarly, the assessment of how much of a region ’ s primary vegetation
cover has been lost can be highly dependent on assumptions made about the nature of the ‘ original ’ vegetation cover (see Chapter 3 ).
In practice, some commonly assumed baselines for
pre - Anthropocene states are based on false assumptions about the degree to which a region originally
supported forest habitats (Virah - Sawmy et al ., 2010 ),
while on the other hand some highly biodiverse areas
of tropical forest have been found to be underlain by
evidence of once dense human populations and former
episodes of forest clearance (e.g. Brncic et al ., 2007 ).
In addition, estimates of habitat conversion, particularly with regard to the tropical forest regions, have
been shown to produce highly inconsistent outcomes.
Surveys conducted at different points in time, or by
different teams, have produced confl icting results
(Grainger, 2008 ). However, it would be fair to counter
that any alternative metric of threat would encounter
some operational diffi culties, and that the use of the
70 per cent threshold provides a reasonable fi rst
approximation.
A further concern, as with all such analyses (Chapter
4 ), is the problem of deciding what constitutes the
most appropriate underlying map of nature, i.e. the
appropriate geographical units to be used in the analysis. Examination of the output maps from the four