Content uploaded by Martin W. Callmander
Author content
All content in this area was uploaded by Martin W. Callmander
Content may be subject to copyright.
Identification of priority areas for plant conservation in Madagascar
using Red List criteria: rare and threatened Pandanaceae indicate sites
in need of protection
Martin W. Callmander, George E. Schatz, Porter P. Lowry II, Michel O. Laivao, Jeannie Raharimampionona,
Sylvie Andriambololonera, Tantely Raminosoa and Trisha K. Consiglio
Abstract A major problem in establishing effective
protocols for conserving Madagascar’s biodiversity is
the lack of reliable information for the identification of
priority sites in need of protection. Analyses of field data
and information from herbarium collections for mem-
bers of the plant family Pandanaceae (85 spp. of
Pandanus; 6 spp. of Martellidendron) showed how risk
of extinction assessments can inform conservation
planning. Application of IUCN Red List categories and
criteria showed that 91% of the species are threatened.
Mapping occurrence revealed centres of richness and
rarity as well as gaps in Madagascar’s existing protected
area network. Protection of 10 additional sites would be
required to encompass the 19 species currently lacking
representation in the reserve network, within which east
coast littoral forests are particularly under represented
and important. The effect of scale on assessments of risk
of extinction was explored by applying different grid
cell sizes to estimate area of occupancy. Using a grid cell
size within the range suggested by IUCN overestimates
threatened status if based solely upon specimen data.
For poorly inventoried countries such as Madagascar
measures of range size based on such data should be
complemented with field observations to determine
population size, sensitivity to disturbance, and specific
threats to habitat and therefore potential population
decline. The analysis of such data can make an
important contribution to the conservation planning
process by identifying threatened species and revealing
the highest priority sites for their conservation.
Keywords IUCN Red List, Madagascar, Pandanaceae,
protected areas, threatened species.
This paper contains supplementary material that can
only be found online at http://journals.cambridge.org
Introduction
The Malagasy flora is characterized by high levels of
diversity (12,000–14,000 species; Goodman & Benstead,
2005; Phillipson et al., 2006) and endemism (.90%;
Schatz, 2001), and is threatened by unsustainable land
use practices, such as shifting cultivation and fire, that
have left ,10% of the original vegetation intact (Myers
et al., 2000). Madagascar is regarded as a high priority
for conservation and for the study of evolutionary
processes (Myers & Knoll, 2001).
At the 2003 World Parks Congress in Durban, South
Africa, the Malagasy President, Marc Ravalomanana,
announced plans to triple the size of the protected areas
network by 2008, from 1.7 to 6 million ha (nearly 10% of
the country). This goal has been integrated into Phase 3
of the National Environmental Action Plan, and an
initial set of target areas has been identified for inclusion
in an expanded protected area network. However,
research and analysis is needed to ensure maximal
inclusion of the country’s biodiversity. Application of
the IUCN Red List criteria, a widely recognized tool for
identifying threatened species, offers a powerful method
to identify priority sites for protection by providing
information on the conservation status of species in the
wild (Rodrigues et al., 2006). In Madagascar, however,
only 425 plant species have been assessed to date, of
which only 210 were evaluated using the current criteria
(IUCN, 2006). Many of the species that have not been
assessed are so poorly known or rare that up-to-date
distribution information is unavailable. To address this
we have developed an efficient, robust procedure for
Martin W. Callmander (Corresponding author) Missouri Botanical Garden,
P.O. Box 299, St. Louis, MO 63166-0299, USA, and University of Neucha
ˆtel,
Laboratoire de Botanique Evolutive, Case Postale 158, 2009 Neucha
ˆtel,
Switzerland. E-mail martin.callmander@mobot-mg.org
George E. Schatz, Porter P. Lowry II* and Trisha K. Consiglio Missouri
Botanical Garden, P.O. Box 299 St. Louis, MO 63166-0299, USA.
Michel O. Laivao,Jeannie Raharimampionona,Sylvie Andriambololonera
and Tantely Raminosoa Madagascar Research and Conservation Program,
Missouri Botanical Garden, BP 3391 Antananarivo 101, Madagascar.
*Also at: De
´partement Syste
´matique et Evolution, Muse
´um National d’Histoire
Naturelle, Case Postale 39, 57 rue Cuvier, 75231 Paris CEDEX 05, France.
Received 4 October 2005. Revision requested 24 April 2006.
Accepted 13 February 2007.
Oryx Vol 41 No 2 April 2007
ß2007 FFI, Oryx,41(2), 168–176 doi:10.1017/S0030605307001731 Printed in the United Kingdom
168
evaluating restricted range plant species using primary
occurrence data from herbarium collections and recent
field observations ( Schatz,2002; Raharimampionona et al.,
2006). This procedure also identifies sites with high
species richness and concentrations of rare species that
constitute prospective conservation priorities. Here we
provide an example of the application of these comple-
mentary approaches using the plant family Pandanaceae.
Pandanaceae comprise four genera of arborescent or
lianoid dioecious monocotyledons, two of which occur
in Madagascar, Martellidendron (7 spp., all but one
restricted to the island) and the widespread
Paleotropical genus Pandanus (c. 600 spp., with 85 spp.
endemic to Madagascar). The taxonomy of both genera
has been revised recently (Laivao et al., 2000, 2006;
Callmander et al., 2001, 2003a,b,c; Callmander & Laivao,
2002) and, combined with information from recent field
work, provides data suitable for illustrating our proce-
dures for identifying conservation priorities. Species of
Pandanaceae grow in all tropical forest types (dry to
humid) and exhibit both widespread and restricted
distributions. By applying IUCN Red List categories,
calculating richness and rarity, and identifying areas
with high diversity of endemic Malagasy Pandanaceae,
we (1) illustrate how assessments using herbarium data
can inform conservation planning, (2) show how the
assessments can be used to evaluate the effectiveness of
the existing protected area network and identify high
priority sites for establishing new protected areas and/
or implementing alternative conservation strategies, (3)
explore difficulties encountered in applying the Red List
criteria relating to species distributions, and (4) illustrate
some problems involved in using herbarium collections
and field observations for Red List assessments and
show how occurrence data can be analysed most
effectively to provide useful interpretations for conser-
vation.
Methods
Geographic data were compiled from .550 herbarium
specimens representing the 91 endemic species of
Malagasy Pandanaceae currently recognized (Fig. 1).
Geographical coordinates for recent collections were
recorded with a global positioning system or assigned
post facto for older specimens using a gazetteer of
botanical collecting localities (Schatz & Lescot, 1996).
All Malagasy Pandanaceae were assessed using the
IUCN Red List criteria (2001). Threatened species are
those evaluated as Critically Endangered (CR),
Endangered (EN), or Vulnerable (VU) with respect to
extinction risk in the wild. We have argued elsewhere
(Callmander et al., 2005) that the data deficient (DD)
category should be used with parsimony to avoid
underestimating the number of threatened species; this
approach has been followed here. We used the four Red
List criteria and subcriteria (IUCN, 2001) to determine
threat: rapid population decline (Criterion A), limited
geographic range and small population size linked with
fragmentation, decline or fluctuation (Criteria B and C),
and very small population size (Criterion D); we did not
attempt to analyse extinction risk quantitatively using
Criterion E.
For each species of Pandanaceae the geographical
information system Arcview v. 3.3 (ESRI, Redlands, USA)
was used to measure three parameters required for
categorization using the Red List criteria (IUCN, 2001):
extent of occurrence (EOO, minimum convex polygon
containing all points of occurrence), area of occupancy
(AOO, area estimated by superimposing a grid onto
occurrence points and calculating the cumulative area of
cells occupied by a species), and predicted future
decline (PFD, estimated continuing reduction in a
species’ AOO). For species restricted to primary vegeta-
tion PFD was estimated using (AOO outside protected
areas / total AOO) * 100, which is based on two
assumptions that reasonably reflect the current situation
for long-lived arborescent plants in Madagascar (Schatz
et al., 2000; Good et al., 2006): within a few decades (3
generations for most Pandanaceae, estimated to be 30–50
years, depending on the species) ongoing habitat
degradation and destruction will leave few significant
areas of primary vegetation outside the protected area
network, and loss of primary vegetation within pro-
tected areas will be minimal. While some unprotected
areas may escape destruction (especially on steep or
rugged terrain), it is nevertheless unlikely that many
populations of long-lived plants will survive outside
protected areas after 3 generations, given Madagascar’s
high rate of deforestation (Green & Sussman, 1990;
Achard et al., 2002). For species tolerant of habitat
disturbance the calculation of PFD using this method
would produce an overestimate of population reduction,
and therefore we have evaluated such species based on
direct field observations, indicating those known to be in
continuing decline (without attempting to calculate PFD)
so that the other Red List criteria can be applied.
The number of subpopulations within the total
population of a species was estimated by overlaying a
10 * 10 km grid onto the mapped locality records and
counting as distinct subpopulations the number of non-
contiguous occupied cells or cell clusters (Schatz et al.,
2000; Good et al., 2006). The number of subpopulations
located within protected areas was determined by
overlaying the distribution of subpopulations onto a
map of protected areas.
The grid cell size used to calculate AOO influences
Red List assessments (Willis et al., 2003) and because
169Malagasy Pandanaceae
ß2007 FFI, Oryx,41(2), 168–176
170
Fig. 1 All collections of the genera Martellidendron and Pandanus (Pandanaceae) in Madagascar mapped on the bioclimatic zones of Cornet
(1974) simplified to five categories (D, Dry; H, Humid; M, Montane; SA, Subarid; SH, Subhumid; Schatz, 2000), and the main existing
Protected Areas (vertical hatching).
M. W. Callmander et al.
ß2007 FFI, Oryx,41(2), 168–176
Malagasy Pandanaceae exhibit a wide range of distribu-
tion types they are an appropriate test group for
evaluating the impact of using alternative grid cell
sizes. The Red List guidelines (IUCN & SSC, 2004)
indicate that when selecting a grid ‘the appropriate scale
[for calculating AOO] depends on the taxon in question
and the comprehensiveness of the distribution data’,
with a 2 * 2 km cell size suggested for most situations.
Gardenfors (2001) recommended using smaller cells if a
species is suspected of being CR but this presupposes an
IUCN assessment (Willis et al., 2003). Schatz et al. (2000)
used 10 * 10 km cells to analyse species in Madagascar’s
endemic plant families, a size considered to correspond
to the average extent of an isolated subpopulation (Good
et al., 2006). Willis et al. (2003) proposed using grid cells
equal to 1/10 of the length between the most distant pair
of points of the EOO polygon. While this works well for
approximately circular to oval distributions (length to
width ratio (3:1), it is problematic for taxa with long,
narrow distributions, yielding very large cells. We
calculated estimates of AOO using three methods: that
of (1) Willis et al. (2003), (2) Schatz et al. (2000), and (3) a
3 * 3 km grid. The latter corresponds closely to the near
maximum for a taxon to be listed as CR under Red List
criterion B2 (for which the threshold is 10 km
2
), and better
reflects the overall poor level of geographic sampling of
plants in Madagascar than a 2 * 2 km cell size.
In calculating EOO ‘disjunctions or discontinuities
within the overall range of a taxon’ may be excluded
(IUCN & SSC, 2004), an idea explored but ultimately
rejected by Willis et al. (2003) because an element of
uncertainty is introduced. We tested the effect of
removing clearly unsuitable habitat (sea and areas
.800 m elevation) on estimates of EOO using data for
Martellidendron, all of which are restricted to lowland
habitats and threatened by fragmentation and forest
loss.
Geographic data on Malagasy Pandanaceae were
analysed with WORLDMAP (Williams, 2000) using J-
degree grid cells (c. 27 * 27 km) to calculate range
size rarity (expressed as the percentage aggregated
reciprocal range size for all species per grid cell), species
richness (a count of species per grid cell), irreplaceability
(presence in a cell of the only known populations of one
or more species), and rarity (the summed inverse range
size for all species present in a cell). The latter is a
function of both endemism and richness, and is also
referred to as endemism richness (Kier & Barthlott,
2001). WORLDMAP was also used to calculate the
correlation between richness and rarity using
Spearman’s non-parametric rank correlation coefficient
rho. Analyses were performed on the 87 species for
which adequate information is available; the 4 species
assessed as DD (Appendix) were excluded. Geographic
distribution was also plotted on maps of remaining
vegetation (Steiniger et al., 2003) and elevation (US
Geological Survey, 1996).
Results
A summary of the number of species assigned to each
Red List category using the three methods of assessment
is presented in Table 1. Application of the three methods
to data from herbarium specimens yielded striking
differences in the proportion of species assessed as
threatened, from 64% using the approach of Willis et al.
(2003) to 100% using a 3 * 3 km grid cell size. Using a 3 *
3 km grid but taking field observations of population
decline and sensitivity to disturbance into account
resulted in 95% of the species assessed as threatened
(Table 1). The Appendix gives the Red List assessments
of all 91 species assessed using this latter method. All
nine CR species belong to Pandanus, only one of which
(P.insuetus) occurs in a protected area. Four of the eight
CR species not encompassed within protected areas
occur in littoral forest, and one each in lowland humid
forest, mid elevation forest, remnant forest parcels on
the Central High Plateau, and western/northern decid-
uous forests. Among the 74 species categorized as EN or
VU, 12 are not known to occur in protected areas and
nine of these are endemic to littoral forest, two species
are restricted to low and mid elevation forest, and one to
171
Table 1 Number of Malagasy Pandanaceae (and % of 87 species assessed) assigned to Red List categories (IUCN, 2001) based on application
of the methodologies proposed by Schatz et al. (2000), Willis et al. (2003), a 3 * 3 km grid cell size, and a 3 * 3 km grid incorporating field
observations (see text for details). The four species categorized as Data Deficient are omitted.
Method
IUCN category* Total no. threatened
(CR+EN+VU)
CR EN VU LC
Willis et al. (2003) 10 (11) 21 (24) 25 (29) 31 (36) 56 (64)
Schatz et al. (2000) 6 (7) 41 (47) 38 (44) 2 (2) 85 (98)
3 * 3 km grid 9 (10) 62 (71) 16 (18) 0 (0) 87 (100)
3 * 3 km grid with field observations 9 (10) 31 (36) 43 (49) 4 (4) 83 (95)
*CR, Critically Endangered; EN, Endangered; VU, Vulnerable; LC, Least Concern
Malagasy Pandanaceae
ß2007 FFI, Oryx,41(2), 168–176
the central highlands. The four species (4%) assessed as
Least Concern (which indicates they are not threatened
with extinction according to the Red List criteria) are
widespread (EOO .20,000 km
2
). Four species (4%) were
categorized as DD because their taxonomic status has
not yet been fully clarified.
When calculations of EOO were adjusted by removing
clearly unsuitable habitat for the two species of
Martellidendron assessed as EN, the estimated EOO values
were reduced from 1,960 to 863 km
2
for M. androcepha-
lanthos and from 717 to 389 km
2
for M. karaka (Fig. 2C).
The highest local species richness of Malagasy
Pandanaceae (10 spp.) occurs in a single J-degree grid
cell containing the Manongarivo Special Reserve in the
north-east of the island (Fig. 1). Two other protected
areas (Tsaratanana and Perinet) each contain nine
species. Other areas of high species richness (7–8 spp.)
include two National Parks within humid forest
(Mananara-Nord and Marojejy) and several unprotected
littoral forest sites along the east coast (Fig. 2A).
Centres of high endemism, as indicated by the highest
endemism richness scores (7.66–14.35%) are found in
four protected areas (Tsaratanana, Manongarivo,
Marojejy and Montagne d’Ambre; Fig. 2A). Relatively
high rarity values (.3%) also occur at Perinet and in
several unprotected littoral forest sites. There was a
strong positive correlation (rho 50.782) between
endemism richness and species richness.
Nineteen species (21%) do not occur in any protected
area (Fig. 2B) and a minimum of 10 cells are necessary to
encompass them all, four of which are irreplaceable.
These same 10 cells collectively encompass 48% (44) of
all species. The minimum number of cells required to
ensure that all Malagasy Pandanaceae are represented at
least once includes 16 additional cells (i.e. 26 in total).
All 87 Malagasy Pandanaceae (excluding DD species)
occur within the cells collectively occupied by the 28
rarest species (i.e. those occurring in only one or two
cells).
Discussion
Pandanaceae are important components of many vege-
tation types in Madagascar, although they are best
represented in humid and subhumid forests and
riparian communities. They also present a wide variety
of distribution patterns, including widespread to locally
restricted taxa, species with nearly circular or linear
ranges, and with almost continuous to highly disjunct
and fragmented distributions. Although the method of
Willis et al. (2003) is useful for calculating AOO for
species with restricted ranges and approximately circu-
lar distributions, setting grid cell size using 1/10 of
the maximum distance between subpopulations is
inappropriate for taxa with widespread distributions
that span both suitable and intervening unsuitable
habitat. This method is also inappropriate for linear or
highly disjunct distributions because it leads to the
selection of large grid cells, resulting in an under-
assessment of threatened species (Table 1). Furthermore,
this method incorporates a dependency of AOO on
EOO, exacerbating an already inherent scale depen-
dency. In these instances the method proposed by
Schatz et al. (2000), using 10 * 10 km grid cells to
calculate AOO, is preferable in Madagascar (Good et al.,
2006); e.g. for littoral forest endemics restricted to a
narrow band (,10 km wide) along the east coast
(Consiglio et al., 2006). This method is, however, less
well suited for species with very restricted ranges
because AOO is overestimated, leading again to an
under-assessment of CR species (Table 1). A combined
approach using one or other of these methods, depend-
ing on distribution shape and fragmentation, would
exploit their respective strengths, although choosing
between them is necessarily subjective, introducing a
bias that could, however, be avoided by consistently
employing a single grid size, reducing uncertainty
(Keith et al., 2000). For Malagasy Pandanaceae use of a
3 * 3 km grid results in the highest number of species
assessed as threatened (Table 1) of the three methods
used, and may not best reflect biological reality. A more
realistic approach is to incorporate field observations
(Appendix, Table 1) that evaluate population decline,
although it remains impossible to eliminate assessor bias
altogether (Regan et al., 2005). In reality, complete
objectivity is unachievable; selecting a grid size for
uniform application, as recommended by IUCN, is itself
subjective.
Estimates of EOO can be improved by removing
unsuitable habitat (IUCN, 2001) or by using an alpha-
hull, in which a highly disjunct distribution is separated
into several distinct hulls (Burgman & Fox, 2003). Such
methods can, however, be difficult to apply using
collection data and may lead to substantial biases if
there are few records or if recorded locations are
imprecise or contain errors. While no change in threat
classification resulted when we removed unsuitable
habitat for species of Martellidendron, the different EOO
values obtained illustrate how such corrections could
significantly affect Red List assessments.
The highest levels of both species richness and
endemism richness occur in northern Madagascar,
especially in the higher mountains (Manongarivo,
Marojejy and Tsaratanana), where 95% of the species
growing at .800 m altitude are local endemics (i.e.
restricted to one or few cells). These results are not
surprising given that nearly one third of Malagasy
Pandanaceae (27 of 91 spp.) are endemic to the northern
172 M. W. Callmander et al.
ß2007 FFI, Oryx,41(2), 168–176
173
(A)
(C)
(B)
Fig. 2 (A) Endemism richness (see text for details) for all 87 species of Pandanaceae assessed (i.e. excluding the 4 species assessed as Data
Deficient) and the four Protected Areas (vertical hatching) with the highest scores. (B) Endemism richness of 19 species not currently
included in Madagascar’s protected area network. (C) Distribution of two Endangered species of Martellidendron in northern Madagascar,
with remaining forests as of 2003 (map from Steiniger et al., 2003).
Malagasy Pandanaceae
ß2007 FFI, Oryx,41(2), 168–176
part of the island (Callmander et al., 2003c). However, most
threatened Pandanaceae occur in low to mid elevation
eastern humid forests, including littoral forests (Schatz
et al., 2000; Randrianasolo et al., 2002; Good et al., 2006).
Our analyses show that 19 Malagasy Pandanaceae are
not included in the country’s protected area network but
can be encompassed in a near minimum set of 10 J-
degree cells, and these therefore represent the highest
priorities for Pandanaceae conservation. If these cells are
added to the existing protected area network, which
already includes the 16 additional cells required to
encompass all Malagasy Pandanaceae, each member of
the family would be protected in at least one site. These
10 new priority areas include four major vegetation
types (Table 2): littoral forests, natural vegetation on the
central highlands, deciduous forests in the north and
west, and rainforests in the east and north-west. The
first two are suffering exceptionally high rates of
deforestation; in 1990 they had ,10% of their original
cover remaining (Green & Sussman, 1990) and have
considerably less today. Neither of these vegetation
types has adequate coverage in protected areas, even
though they contain large numbers of locally endemic
species. Littoral forests are severely under-represented
in the protected area network, with just four small
parcels totalling 695 km
2
(ANGAP, 2005; Consiglio et al.,
2006). Of the 10 new sites, 3 have been granted
preliminary protected area status (Daraina, Analalava-
Tampolo, and Saint Luce).
One goal of conservation planners is maximum
inclusion of biodiversity in protected areas. Our study
shows that by conserving sites in Madagascar with the
rarest Pandanaceae (those occurring in only one or a few
areas) along with species not known to occur in
protected areas, we can achieve protection for the entire
family. This is also true for Madagascar’s endemic plant
families (Schatz et al., 2003; Raharimampionona et al.,
2006), some members of which are restricted to these
same areas. Application of this method to other groups
(now being conducted) is confirming the importance of
174
Table 2 Priority areas for Pandanaceae conservation, with size of area and threatened species of Pandanaceae occurring in each location.
Location Area (ha) Species (Red List category)*
Littoral forests
Baie d’Antongil (forests around the bay) 185 Martellidendron karaka (EN), Pandanus
bipyramidatus (VU), P. columnaris (VU),
P. concretus (VU), P. guillaumetii (VU),
P. insuetus (CR), P. imerinensis (VU),
P. longecuspidatus (VU), P. princeps (CR)
Analalava-Foulepointe-Soanierana-Ivongo-Tampolo 2,613 P. callmanderiana (CR), P. cephalothus (VU),
P. comatus (EN), P. concretus (VU),
P. guillaumetii (VU), P. malgassicus (VU),
P. neoleptopodus (EN), P. pseudocollinus (EN),
P. platyphyllus (EN), P. rollotii (VU),
P. sylvicola (CR)
Fort-Dauphin region (Saint-Luce) 457 P. bipyramidatus (VU), P. dauphinensis (CR),
P. longistylus (EN), P. peyriarasii (CR),
P. platyphyllus (VU), P. rollotii (VU)
Natural vegetation of the highlands
Andringitra region (E side outside the protected area) 46,649 P. ambalavaoensis (CR)
Itremo region (forest patches) 405 P. bakeri (CR)
Northern and Western deciduous forests
Cap Saint Andre
´(40 km south) 5,858 P. ambongensis (VU), P. perrieri (CR),
P. stellatus (VU)
Forest block in the Daraina region (Ambilondamba,
Ampondrabe, Antsahabe, Binara & Bobankora)
17,796 P. analamerensis (VU), P. coriaceus (VU),
P. grallatus (EN), P. latistigmaticus (EN),
P. perrieri (CR)
Mangoky basin (forest blocks along rivers in the Mangoky Basin) 84,477 P. ambongensis (VU)
Eastern and Western rainforests
Lakato forest 555 M. cruciatum (VU), P. freycinetioides (VU)
Galoko massif (NE of Ambanja) 8,720 M. androcephalanthos (EN), M. cruciatum (VU),
P. sp. nov ‘neosermollii’ ined. (CR),
P. mammillaris (EN)
South of Moramanga (forest patches around Anosibe an’ala) 1,730 P. ceratophorus (VU), P. kimlangii (VU),
P. leptopodus (VU), P. macrophyllus (CR)
*Species not occurring in the protected area network are in bold. CR, Critically Endangered; EN, Endangered; VU, Vulnerable
M. W. Callmander et al.
ß2007 FFI, Oryx,41(2), 168–176
these sites and revealing additional conservation prio-
rities. Ultimately, the identification of such key sites is a
fundamental component of developing a comprehensive
strategy for biodiversity protection, which can be
facilitated by carefully evaluating the conservation
status of well-delimited species using geo-referenced
data from natural history collections.
Our study confirms that conservation planning in
Madagascar must include as much of the remaining
lowland forest of the east coast as possible, especially
littoral forests (Consiglio et al., 2006); stands between the
Baie d’Antongil (central north-east) and Tampolo
(central east) are of particular importance for
Pandanaceae, as they are for the endemic plant families
(Schatz et al., 2003).
For Red List assessments the appropriate selection of
grid cell size is dependent upon the shape, size and
homogeneity of a species’ distribution, as manifested in
part by its extent of occurrence. By selectively applying
complementary methods for choosing grid size (Schatz
et al., 2000; Willis et al., 2003; IUCN & SSC, 2004),
depending on distribution attributes, we have shown
that there is no single best approach to selecting an
appropriate scale for calculating area of occurrence, and
have demonstrated that the method chosen can influ-
ence threat assessments. With 75% of Malagasy
Pandanaceae threatened, conservation action must be
taken immediately to ensure their continued survival.
The careful analysis of specimen data using appropriate
methods can make an important contribution to this
process by identifying threatened species and revealing
the highest priority sites for their conservation.
Acknowledgements
We thank the Parc Botanique et Zoologique de
Tsimbazaza, ANGAP (Association Nationale pour la
Gestion des Aires Prote
´ge
´es), and the Missouri Botanical
Garden’s office in Antananarivo (especially Lalao
Andriamahefarivo) for local assistance, and Bette
Loiselle (University of Missouri–St. Louis, USA), Prof.
Philippe Ku
¨pfer (Universite
´de Neucha
ˆtel) and Wendy
Strahm (IUCN, Gland, Switzerland) for support before
and during this study. We are grateful to Craig Hilton-
Taylor (IUCN, Cambridge, UK), Mark Burgman
(University of Melbourne, Australia) and Justin Moat
(Royal Botanic Gardens, Kew, UK) for kind responses to
queries regarding methods, and to two anonymous
reviewers for helpful comments. This work was sup-
ported by grants to MWC from the Swiss National
Science Foundation (No. PBNE2-102378), National
Geographic Society (No. 04065) and Conservation
International-Madagascar (Convention 234).
References
Achard, F., Eva, H.D., Stibig, H.-J., Mayaux, P., Gallego, J.,
Richards, T. & Malingreau, J.-P. (2002) Determination of
deforestation rates in the world’s humid tropical forests.
Science,297, 999–1002.
ANGAP (2005) Liste des Parcs et Re
´serves de Madagascar. Http://
www.parcs-madagascar.com/liste.htm [accessed 10 January
2007].
Burgman, M.A. & Fox, J.C. (2003) Bias in species range
estimates from minimum convex polygons: implications for
conservation and options for improved planning. Animal
Conservation,6, 19–28.
Callmander, M.W.,Chassot, P., Ku
¨pfer, P. & Lowry II, P.P.(2003a)
Recognition of Martellidendron, a new genus of Pandanaceae,
and its biogeographic implications. Taxon,52, 747–762.
Callmander, M.W. & Laivao, M.O. (2002) Re
´vision de Pandanus
sect. Dauphinensia St. John (Pandanaceae) a
`Madagascar.
Botanica Helvetica,112, 47–67.
Callmander, M.W., Laivao, M.O. & Wohlhauser, S. (2003b) Les
Pandanus sect. Acanthostyla Martelli (Pandanaceae) d’altitude
du Nord de Madagascar, avec description de deux nouvelles
espe
`ces. Candollea,58, 63–74.
Callmander, M.W., Schatz, G.E. & Lowry II, P.P. (2005) IUCN Red
List assessment and the Global Strategyfor Plant Conservation:
taxonomists must act now. Taxon,54, 1047–1050.
Callmander, M.W., Wohlhauser, S. & Gautier, L. (2003c) Notes
bioge
´ographiques sur les Pandanaceae du nord de
Madagascar. Candollea,58, 351–367.
Callmander, M.W., Wohlhauser, S. & Laivao, M.O. (2001) Une
nouvelle section du genre Pandanus (Pandanaceae) a
`
Madagascar: Pandanus sect. Tridentistigma. Adansonia, se
´r. 3, 23,
49–57.
Consiglio, T., Schatz, G.E., McPherson, G., Lowry II, P.P.,
Rabenantoandro, J., Rogers, Z.S., Rabevohitra, R. &
Rabehevitra, D. (2006) Deforestation and plant diversity of
Madagascar’s littoral forests. Conservation Biology,20,
1799–1803.
Cornet, A. (1974) Essai de cartographie bioclimatique a
`Madagascar.
Notice explicative N
˚55, ORSTOM, Paris, France.
Ga
¨rdenfors, U., Hilton-Taylor, C., Mace, G. & Rodrı
´guez, J.P.
(2001) The application of IUCN Red List Criteria at regional
levels. Conservation Biology,15, 1206–1212.
Good, T.C., Zjhra, M.L. & Kremen, C. (2006) Addressing data
deficiency in classifying extinction risk: a case study of a
radiation of Bignoniaceae from Madagascar. Conservation
Biology,20, 1099–1110.
Goodman, S.M. & Benstead, J.P. (2005) Updated estimates of
biotic diversity and endemismfor Madagascar. Oryx,39, 73–77.
Green, G.M. & Sussman, R.W. (1990) Deforestation history of
the eastern rainforest of Madagascar from satellite image.
Science,248, 212–215.
IUCN (2001) 2001 Categories and Criteria (version 3.1). IUCN,
Gland, Switzerland [http://www.redlist.org/info/
categories_criteria2001.html, accessed 20 February 2007].
IUCN (2006) 2006 IUCN Red List of Threatened Species. IUCN,
Gland, Switzerland [http://www.redlist.org, accessed 10
January 2007].
IUCN & SSC (2004) Guidelines for using the IUCN Red List
Categories and Criteria. IUCN, Gland, Switzerland.
Keith, D.A., Auld, T.D., Ooi, M.K.J. & Mackenzie, D.E. (2000)
Sensitivity analyses of decision rules in World Conservation
Union (IUCN) Red List Criteria using Australian plants.
Biological Conservation,94, 311–319.
175Malagasy Pandanaceae
ß2007 FFI, Oryx,41(2), 168–176
Kier, G. & Barthlott, W. (2001) Measuring and mapping
endemism and species richness: a new methodological
approach and its application on the flora of Africa.
Biodiversity and Conservation,10, 1513–1529.
Laivao, M.O., Callmander, M.W. & Buerki, S. (2006) Sur les
Pandanus (Pandanaceae) a
`stigmates saillants de la co
ˆte est de
Madagascar. Adansonia, se
´r. 3, 28, 267–285.
Laivao, M.O., Callmander, M.W. & Wohlhauser, S. (2000) Une
espe
`ce nouvelle de Pandanus sect. Martellidendron
(Pandanaceae) de la pe
´ninsule de Masoala, Madagascar.
Botanica Helvetica,110, 41–49.
Myers, N. & Knoll, H.A. (2001) The biotic crisis and the future
of evolution. Proceedings of the National Academy of Sciences,98,
5389–5392.
Myers, N., Mittermeier, R.A., Mittermeier, C.G., Da Fonseca,
G.A.B. & Kent, J. (2000) Biodiversity hotspots for
conservation priorities. Nature,403, 853–858.
Phillipson, P.B., Schatz, G.E., Lowry II, P.P. & Labat, J.-N. (2006)
A catalogue of the vascular plants of Madagascar. In
Taxonomy and Ecology of African Plants: Their Conservation
and Sustainable Use (eds S.A. Ghazanfar & H.J. Beentje), pp.
613–627. Proceedings 17th AETFAT Congress, Addis Ababa,
Ethiopia & Royal Botanic Gardens, Kew, London, UK.
Raharimampionona, J., Andriambololonera, S., Schatz, G.E.,
Lowry II, P.P., Rabarimanarivo, M., Ratodisoa, A. &
Ravololomanana, N. (2006) Identification des aires
prioritaires pour la conservation des plantes a
`Madagascar:
utilisation des donne
´es botaniques pour de
´finir les priorite
´s
en matie
`re de conservation. In Taxonomy and Ecology of
African Plants: Their Conservation and Sustainable Use (eds S.A.
Ghazanfar & H.J. Beentje), pp. 447–456. Proceedings 17th
AETFAT Congress, Addis Ababa, Ethiopia & Royal Botanic
Gardens, Kew, London, UK.
Randrianasolo, A., Miller, J.S. & Consiglio, T.K. (2002)
Application of IUCN Criteria and Red List Categories
on species of five Anacardiaceae genera in Madagascar.
Biodiversity and Conservation,7, 1289–1300.
Regan, T.J., Burgman, M.A., McCarthy, M.A., Master, L.L.,
Keith, D.A., Mace, G.M. & Andelman, S.J. (2005) The
consistency of extinction risk classification protocols.
Conservation Biology,19, 1969–1977.
Rodrigues, A.S.L., Pilgrim, J.D., Lamoreux, J.F., Hoffman, M. &
Brooks, T.M. (2006) The value of the IUCN Red List for
conservation. Trends in Ecology & Evolution,21, 71–76.
Schatz, G.E. (2000) Endemism in the Malagasy tree flora. In
Diversite
´et Ende
´misme a
`Madagascar (eds W.R. Lourenc¸o &
S.M. Goodman), pp. 1–9. Me
´moires de la Socie
´te
´
Bioge
´ographie, Paris, France.
Schatz, G.E. (2001)Generic Tree Flora of Madagascar. [Flore Ge
´ne
´rique
des Arbres de Madagascar]. Missouri Botanical Garden, Saint-
Louis, USA & Royal Botanic Gardens, Kew, UK.
Schatz, G.E. (2002) Taxonomy and herbaria in service of plant
conservation: lessons from Madagascar’s endemic families.
Annals Missouri Botanical Garden,89, 145–152.
Schatz,G.E.,Andriambololonera,S., Randriampionona,J. & Lowry
II, P.P. (2003) Rediscovering Lost Species. Http://
www.mobot.org/MOBOT/Madagasc/Rediscovering%
20Lost%20Species_files/frame.htm [accessed 10 January 2007].
Schatz, G.E., Birkinshaw, C., Lowry II, P.P., Randriantafika, F. &
Ratovoson, F. (2000) The endemic plant families of Madagascar
project: integrating taxonomy and conservation. In Diversite
´et
Ende
´misme a
`Madagascar (eds W.R. Lourenc¸o & S.M. Goodman),
pp. 11–24. Me
´moires de la Socie
´te
´Bioge
´ographie, Paris, France.
Schatz, G.E. & Lescot, M. (1996) Gazetteer to Malagasy Botanical
Collecting Localities. Http://www.mobot.org/MOBOT/
research/madagascar/gazetteer/ [accessed 10 January 2007].
Steiniger, M.K., Harper, G.J. & Tucker, C.J. (2003) Forest Cover
Fragmentation and Clearance in Madagascar. Centre for Applied
Biodiversity Science, Conservation International, Arlington,
USA & Nasa Goddard Space Flight Center, Washington, DC,
USA.
US Geological Survey (1996) Global 30-Arc-Second Elevation Data
Set. US Geological Survey, Sioux Falls, USA.
Williams, P.H. (2000) WORLDMAP.Software and help document
4.2. Privately distributed, London, UK.
Willis, F., Moat, J. & Paton, A. (2003) Defining a role for
herbarium data in Red List assessments: a case study of
Plectranthus from eastern and southern tropical Africa.
Biodiversity and Conservation,12, 1537–1552.
176
Appendix
The Appendix for this article is available online at
http://journals.cambridge.org
Biogeographical sketches
Martin W. Callmander’s research interests include the
inventory and systematics of the Malagasy flora, Malagasy
plant biogeography, and using botanical data to evaluate
endemism and conserve the remaining forests of this
biodiversity hotspot.
George E.Schatz’s research interests includethe woody flora of
Madagascar, the use of primary occurrence data for extinction
risk assessment, and species distribution modelling.
Porter P. Lowry II’s research interests include the inventory
and systematics of the Malagasy flora, and assessing
patterns of biological diversity to identify threatened species
and priority areas for conservation.
Michel O. Laivao, J. Raharimampionona, S. Andriam-
bololonera and T. Raminosoa work for Missouri Botanical
Garden’s research and conservation programme in
Madagascar. They are currently working on a threat
assessment of endemic Malagasy plant species.
Trisha Consiglio specializes in geographical information
systems, using tools that include remote sensing and species
distribution modelling to identify key areas of plant
diversity and endemism.
M. W. Callmander et al.
ß2007 FFI, Oryx,41(2), 168–176
