M10: Network Refinement Protocol
Learning Objectives: This module
develops a protocol for refining conservation area networks selected at stage 6 (see M8: Place Prioritization). Learners will draw from what was learned in previous stages and be asked to reflect on how to utilize
these methods in the process of refining a conservation network plan.
Identifying
a candidate (nominal) conservation area network that satisfies biodiversity surrogate
representation targets (see M8: Place
Prioritization) is only the beginning of the process of selecting a
network for implementation in the field.
Recall
that representation of biodiversity surrogates in a conservation area network
is useless if the prognosis for their survival is not good —see M9: Vulnerability and Persistence Analysis.
This
module will consider how persistence considerations can be taken into account
to refine an initial prospective conservation area network.
Results
from the vulnerability assessment (see M9:
Vulnerability and Persistence Analysis) will be crucial.
Careful
attention must be given to the representation targets that were used and their
influence in planning because these targets often do not have very good
biological justification.
The
next stage (see M11: Multi-Criteria Analysis)
will take up the incorporation of sociopolitical criteria.
The
network refinement protocol can also be implemented at other stages, for
instance, before site prioritization (see below).
Experts
with local knowledge and experience are crucial to the refinement process (see below).
The
basic idea is to drop selected areas that either have high vulnerability of the
surrogates in them or, for other reasons, cannot be put under a conservation
plan, and then run place prioritization again without these areas (see M8: Place Prioritization) to ensure adequate
representation of the surrogates.
If
an area itself has high vulnerability, then it should be removed from the
conservation area network.
This
means that the prognosis is not good for all surrogates in that area.
If
only some surrogates at an area have low viability then what must be determined
is whether that area is necessary for the adequate representation of those
surrogates (reaching the target level for that surrogate).
If these surrogates are adequately represented and are viable in the other areas within the network, then the area need not be dropped from the network becuase of these specific viability problems.
Throughout
the network refinement protocol, "adequate representation" means that a
surrogate is represented at least up to its target level.
Viability
assessments of taxa are difficult to carry out because
of the reasons discussed in M9: Vulnerability
and Persistence Analysis, such as requirements for large amounts of
demographic data.
This
viability assessment strategy--simultaneous population viability analysis of all
surrogate species--has apparently so far not been used in any application
because of such difficulties.
Quite
often a stage of site assessment which is equivalent to network refinement is carried out even before initial place
prioritization.
Areas
with high vulnerability can be dropped even before areas are prioritized to
create an initial conservation area network—see Example 10.1.
The
most obvious situation when this should be done is when some areas are very
seriously anthropogenically transformed (e.g., through
urbanization, industrialization, etc.).
This
process is known as "masking" a site--essentially coding it for non-selection.
Most
software tools for area prioritization (for instance, ResNet)
allow such masking.
However,
if the prognosis of individual surrogates must be taken into account, this is a
cumbersome option.
The
viability of surrogates must be estimated for all areas in the planning region,
and not only for the selected areas.
Given
how difficult it is to estimate such viabilities (see M9: Vulnerability and Persistence Analysis)
this is extremely cumbersome to achieve in practice.
When
hundreds of species are used as surrogates, they must all have their
viabilities assessed. In practice this is impossible.
Additionally,
"masking" an area when using individual surrogates is not acceptable if other
surrogates have viable populations within the site. Whether or not the area
could be important for the representation of these other surrogates can only be
judged from the results of area prioritization.
|
Example 10.1
Excluding
Anthropogenically Transformed Areas from a Conservation
Area Network for the Transvolcanic Belt of Mexico
(Fuller
et al. 2006)
|
|
This example was also discussed in M9: Vulnerability
and Persistence Analysis —see Example 9.2.
Fuller et al.'s (2006) identification of priority areas for the Transvolcanic Belt of Mexico excluded (or "masked") all anthropogenically transformed sites as candidates either
for conservation or restoration action—see Figure 10.1.
|
|
|
The region was divided into three types of landscape: those that
had primary vegetation relatively intact, those that had secondary
vegetation, and those that had neither. The first type was used to identify
conservation areas, the second type to identify areas that could potentially
be restored. The third type was excluded from the analysis because the prognosis
for the persistence of any biota in them was deemed unlikely.
|
Figure 10.1
Anthropogenically
Transformed Areas in the Transvolcanic Belt of Central Mexico
|
|
|
These are the areas that were excluded from the
analysis or "masked" of Fuller et al. (2006) before site prioritization
algorithms were applied. 36 % of the total area was excluded in this way.
|
|
Yet
another method of dealing with vulnerability is to use features related to
vulnerability as criteria in Multi-criteria analysis
(MCA) (see M11: Multi-Criteria Analysis)—see Example 10.2.
These
features include the human population of an area, its distance from an anthropogenically transformed area, the distance to an
existing conservation area, etc.
|
Example 10.2
Priority
Areas for Ecuador
(Sarkar
et al. 2004b)
|
|
Sarkar et al. (2004b) used 46 vegetation types as biodiversity
surrogates to augment the National Reserve System (NRS) of Ecuador
to include 10 % of the
habitat of each vegetation type. Continental Ecuador (248 750 sq km) was
divided into 2 × 2 sq km cells in a grid for the analysis. From 100 different
solutions, the best two (Figures 10.2a, b) were selected using multi-criteria analysis (see M11: Multi-Criteria Analysis) using six criteria (1)
the aggregate number of conservation areas, which should be minimized
to achieve spatial cohesiveness of the network; (2) the average area of each conservation area, which should
be maximized to encourage larger conservation areas; (3) the variance of the areas, which should be minimized to
discourage further the selection of very small areas; (4) the aggregate distance of the selected cells to existing units of the NRS, which
should be minimized, again to increase cohesiveness; (5) the aggregate distance to anthropologically transformed areas, which should be
maximized to decrease the threat of habitat destruction; (6) the total area of the selection cells,
which should be minimized to decrease the cost of acquisition of the added
cells.
Figure 10.2
|
|
(a)
|
(b)
|
In
addition to refining the network before the place prioritization process,
refinement of initially selected conservation area networks can be used to
mitigate the effect of the targets that were imposed on surrogates.
Recall
that these targets often do not have a very firm or strong biological basis—see
M6: Conservation Targets and Goals.
Therefore it makes sense to modify conservation area networks in such a way that there is some robustness with respect to the use of targets.
If
targets are set as percentages of the total land, or percentages of the
habitats of surrogates, then the targets may change because of land use
patterns by humans.
This
problem can be avoided by carefully refining the initially selected conservation
area network—see Example 10.3.
|
Example 10.3
A Conservation Plan for Papua New Guinea
(Faith et al. 2001) |
|
Faith et al. (2001) used network refinement for the Papua New Guinea region in a non-standard way,
in that they refined the conservation area network of Papua New Guinea
through the
targets of representation, rather than the viability of the species. In conventional cases, an a priori target,
such as 10 % of the habitat (see M6: Conservation Targets and Goals), is usually chosen to represent
biodiversity surrogates.(This example
was also used in M4: Data Compilation, Assessment, and
Treatment see Example 4.1.) In the
Papua New Guinea
case, Faith et
al. (2001) argued that the conventional approach to target representation in
conservation planning does not fully support the main goals of biodiversity
representation and persistence. They
contended that, to achieve biodiversity representation and persistence fully, percentage targets should first be calculated and
assessed with no constraints on biodiversity representation from other
criteria. This means that the targets of representation used in the presence
of constraints (usually used in the initial analysis of a region) can only
emerge after an initial analysis of a region.
For example, if one were to use a target of 10 % of representation in
conservation planning for a network, Faith et al. argue that representation
should be manifested in a target (or set of targets) of 10% of a total area
and/or habitat types without incorporating humans, opportunity costs, land
use history, etc. (things that normally get factored into the standard method
of setting targets) in Figure 10.3a, this is indicated by the white
circle.This is called the baseline
analysis.The next step is to use this
initial baseline target (ie. amount/type of
surrogates within the initial 10%) in the area, this time incorporating
humans, opportunity costs, land use history, and existing protected
areas.In Figure 10.3a, the triangle
represents the trade-off curve with only the constraint of cost taken into
account. The square represents the
trade-off curve with additional constraints including existing conservation
areas.Essentially, conservation planning includes
the minimizing of opportunity cost and the minimization of biodiversity
vulnerability.Thus, initializing
planning with a baseline solution (without constraints) will help to develop
a more representative plan for persistence of biodiversity.
|
|
Figure 10.3a
Trade-off curves
(Faith et al. 2001)
|
|
Each area for the conservation network was depicted in a Resource
Mapping Unit (RMU) rather than a grid cell. Using a comparison of the
opportunity cost to the complementarity value of
each site, if the complementarity value of a site
exceeded the opportunity cost (i.e. timber value/volume Figure 10.3b), then
the site could potentially be added to the network.Other such data criteria that were taken
into account were: opportunity costs (timber value and
agricultural potential); commitments (existing conservation areas Figure
10.3c); masks (land use intensity--high to moderate--and small RMUs depicting small site areas which are undesirable);
preferences (i.e. low human population density); and priority conservation
areas (those sites satisfying the criteria of low land use, low human
population density, etc. Figure 10.3d).
The final map of the conservation network (RMUs)
with the five data criteria taken into account is demonstrated in Figure
10.3e.
|
|
Figure 10.3b
Timber Volume Classes
Yellow = highest volume class
Red = second highest class
|
Figure 10.3c
Existing Protected Areas
The areas were modified to fit the RMU
shapes changing their shapes in the process.
|
|
Figure 10.3d
High Priority Areas
The areas were modified to fit the RMU shapes
changing their shapes in the process.
|
Figure 10.3e
Areas with High
Conservation Value
The areas were modified to fit the RMU
shapes changing their shapes in the process.
|
Refinement
is also used in some cases to include areas that were not selected in the
original conservation area network but are known to be obviously of
conservation interest.
For
instance, there may be biological features that are known to be important but
not explicitly used as surrogates.
For
migratory birds and flying insects, patches of habitat added to an existing
network at requisite intervals (wherever necessary using knowledge of the
dispersal patterns and rates of the species) may provide valuable
connectivity —see Example 10.4.
In
such cases, anthropogenic transformation of these intervening lands may not be
highly deleterious.
|
Example 10.4
Stepping
Stones and Enlargement of Conservation Areas in the Netherlands (van Langevelde et al. 2002)
|
|
In an analysis of an area in the southern Netherlands
van Langevelde et al. (2002) used the nuthatch (Sitta europaea)
as a surrogate for other birds. The choice was made because the nuthatch is
an umbrella species for other birds.
(Note that this is a controversial choice because umbrella species are often poor surrogates for other species - see M5: Surrogate Identification and Analysis). The area included both deciduous woods and farmland. Initial
conservation areas were selected on the basis of their suitability for
nuthatch habitat and unsuitability for agriculture. These are the areas in
black in Figure 10.4a. Stepping stones (or intervening lands) were then
added to ensure connectivity between populations in the conservation
areas— these are shown in Figure 10.4b. Finally, the areas were all expanded
to promote higher viability of all the populations—see Figure 10.4c
|
Figure 10.4
(a)
|
|
(b)
|
(c)
|
Refining
initial plans often requires the participation of experts with specific
knowledge of the planning region.
Computer
algorithms may miss important local idiosyncrasies.
Refinement
is a highly context-dependent process, and knowing the appropriate context
requires familiarity with local issues, both biological and sociopolitical
issues.
The
emphasis should always be on the fact that planning protocols are to be used by
experts, not to replace experts.
