APPLICATIONS IN
CONSERVATION
with
Conservation International (CI):
quantitative
methods at the Upper Guinea region priority-setting workshop
A. Balmford1, T. Brooks123, N. Burgess24, F. Corsi5,
L. Hansen2, J. Lovett6, C. Rahbek2, P. Williams7
1 Conservation Biology Group, Department
of Zoology, University of Cambridge, Downing Street, Cambridge CB2
3EJ, UK
2 Zoological Museum, Universitetsparken
15, DK-2100, Copenhagen Ø, Denmark
3 Center for Applied Biodiversity Science, Conservation
International, 2501 M Street NW, Suite 200, Washington DC 20037,
USA
4 WWF-US, Conservation Science
Programme, 1250 Twenty-Fourth Street NW, Washington DC 20037, USA
5 Institute of Applied
Ecology, Via L. Spallanzani 32, 00161 Rome, Italy
6 Environment Department,
University of York, York Y010 5DD, UK
7 Biogeography and
Conservation Laboratory, The Natural History Museum, Cromwell Road,
London SW7 5BD, UK
One of the central aims of the African biodiversity database project
of the University of Copenhagen, The Natural History Museum (London),
and the University of Cambridge, is to make quantitative biodiversity
information available to workshops on conservation priorities. Here
we summarise how this quantitative information was used at the workshop
for the Upper Guinea region organised by Conservation International
at Elmina, Ghana, in December 1999. This study was funded by the Isaac
Newton Trust of the University of Cambridge, DANIDA (Danish Ministry
of Foreign Affairs, Danida, Department of Development Research), the
Center for Applied Biodiversity Science of Conservation International,
and The Natural History Museum, London.
Pre-workshop
Following conference calls in June and July it was agreed that the database
project would provide range maps for all species occurring within the
region for use during the workshop. There would also be opportunities
for specialists to make attributed modifications to the database during
the workshop.
Workshop (days 1 and 2)
For the first two days, the 'Biogeography Group' used WORLDMAP
to identify near-minimum sets of one-degree grid cells to represent,
for each of birds, mammals, snakes and amphibians and the 5% of plants,
and for all species combined, the following:
- All species across the whole of Africa, but looking only at the
contribution of the Upper Guinea region to the continental priorities.
- All species occurring in the Upper Guinea region (by removing the
fauna of other parts of the continent from the database).
- All species endemic to the Upper Guinea region (by removing all
species occurring outside of the region).
We had planned one other analysis. This was to identify near-minimum
sets for all species occurring in the Upper Guinea forest, but this
proved unfeasible because of a lack of habitat preference information
for most species.
We found that analysis (2) identify near-minimum sets for representing
all species occurring in the Upper Guinea region proved uninformative,
because many of the selected grid cells were located in the margins
of the region, caused by the peripheral presence of species more widespread
outside the region.
In addition, we found that (3) the near-minimum sets for representing
all species endemic to the Upper Guinea region gave an overly narrow
view because of the relatively small number of species completely restricted
to Guinea, Sierra Leone, Liberia, Cote dIvoire and Ghana.
Consequently, we concentrated on approach (1), which shows the contribution
of Upper Guinea to conservation priorities for the databased species
in sub-Saharan Africa. In the map below, red cells are irreplaceable
in that they have records for species known from nowhere else but these
cells. The orange cells are needed to represent species that are more
widespread, so that other (flexible) areas could be substituted in their
place (although in some cases it would take more than one). The numbers
refer to the number of species that each area contributes uniquely to
the continental picture for representing all of the sub-Saharan species
in the database:
Workshop (days 3 and 4)
We refined the large-scale continental-derived priorities found on days
1 and 2 of the workshop to identify forest areas (polygons) comparable
in scale to those generated by the working groups for the various major
taxa.
We started with a map of the near-minimum set of one-degree grid cells
needed to represent all birds, mammals, snakes, amphibians and sampled
plants, at least once, calculated for the entire continent. Within the
Upper Guinea region, feedback from taxon-group experts led us to exclude
cells picked solely because they contained non-forest species.
Next, we overlaid our grid cells on the WCMC forest cover map, and
within each minimum-set cell drew polygons around remaining forest cover.
Where individual forest blocks extended into neighbouring cells, the
forest-area polygons were continued into those cells. These polygons
represent our continental level priorities for forest species across
Upper Guinea, shown here in red:
We then compared our priorities with a preliminary overlay of the combined
priorities chosen by groups of specialists in each taxon (birds, mammals,
herps and plants), shown here stippled in grey:
In general, there was remarkable agreement between the two maps, with
most of our priority polygons (shown below in red) overlapping the taxon-group
polygons (shown below outlined in green) and vice versa. This cross-validation
of two largely independent methods is encouraging:
However, there were some areas picked by just one of the two approaches.
Examining how these differences arose is illuminating. Consultation
with taxon-group experts helped us to identify five reasons for these
differences:
- Taxon groups picked some areas we did not because they had records
which were not in our database (e.g. forests around Mt Bolo in southern
Cote d'Ivoire, chosen for plants not in our taxonomically limited
plant database).
- Taxon groups picked some areas we did not because they believed
they were very likely to be important for their group, even though
they did not have records confirming this (e.g. forests of eastern
Cote d'Ivoire, chosen for its likely importance for herps). Obviously
a records-based approach such as our cannot identify such areas.
- Taxon groups picked one area we did not because, for the groups
covered in our database, the important local species are represented
in other cells we had picked (e.g. Ziama Massif in Guinea).
- We picked some areas which the taxon groups did not because the
records that led us to select them are taxonomically or geographically
dubious, or extremely old (e.g. the eastern forests of southern Ghana).
- We picked some areas which taxon groups did not because they had
apparently overlooked some key records (e.g. forest fragments in southern
Fouta Djallon, where we had records of three narrowly distributed
forest plants).
Conclusions
- The degree of overlap between the two sets of (largely) independently
derived priority polygons is encouraging. It also suggests that similar
sampling biases to those in our database are also present in the data
and approach used by the specialist taxon groups.
- It is quite likely, however, that some of the congruence is inevitable,
at least insofar as both approaches suffer from similar sampling problems,
and (perhaps more importantly) identify priorities based on the same
maps of the remaining forest patches.
- We believe that continuing both approaches would be useful, for
two reasons. First, the dialogue between the taxon groups and the
biogeography group at this workshop was mutually beneficial. We have
been able to improve our database and taxon specialists have offered
to continue helping us with this. In a few cases, we were also able
to help the taxon groups by identifying some key species which they
initially overlooked.
- Second, and looking ahead, the quantitative approach has the potential
to integrate biological and socio-economic concerns in accountable
and potentially powerful ways. We explored this briefly in the Upper
Guinea workshop, by examining (at the grid-cell level) how biological
priority selection might be modified to take account of human population
patterns. We were encouraged to do this after finding a noisy but
highly significant (at p < 0.001) positive correlation between
total species richness and human population density among Upper Guinea
grid cells: within this region, more people live in biologically rich
areas than in less species rich areas. We addressed this issue by
modifying our cell-selection criteria so that, at each step in the
algorithm, we picked not the cell with the highest complementary species
richness, but the cell with the highest ratio of complementary species
richness to a simple index of population density. This procedure preferentially
selects cells with relatively few people for their biological value.
The resulting priorities generally contained more species between
them than sets of cells containing the same total number of people,
but picked using a straightforward near-minimum set algorithm. We
are very keen at future workshops to explore and expand these approaches,
using more realistic quantitative indices of conservation threats
and opportunities.
