Data Sources. Species' distribution data are available for 2,294 plants (20),
comprising ≈20% of the total European flora, sampled between 1972 and
1996. Modeling was conducted by using available data for Europe on a 50
× 50 km grid. The mapped area comprises western, northern and southern
Europe, but excludes most of the eastern European countries where
recording effort was both less uniform and less intensive (21).
After removing species with <20 records, we considered range
responses of 1,350 plant species of Europe. We assume this sample can
be taken as representative of the responses of European plant species
to climate change because it includes most of the life forms and
phytogeographic patterns found among plant species in Europe.
Climate data were obtained from the Climatic Research Unit (www.cru.uea.ac.uk)
and included mean annual, winter, and summer precipitation, mean annual
temperature and minimum temperature of the coldest month (MTC), growing
degree days (>5°) and an index of moisture availability (22). These variables were chosen because of their strong link with the physiology and growth of plant species (23, 24).
For instance, MTC discriminates species based on their ability to
assimilate soil water and nutrients, and continue cell division,
differentiation and tissue growth at low temperatures (lower limit),
and chilling requirements for processes such as bud break and seed
germination (upper limit). The moisture index discriminates species
through processes related to phenology, rooting strategy, leaf
morphology, and xylem vulnerability to cavitation. However, because
there is surprisingly little experimental work for any particular
species to guide the choice of bioclimatically limiting variables, the
variables are generic and represent a hypothetical minimum basic set
for niche-based modeling. Climate data were averaged for the 1961-1990
period. The data were supplied on a 10-foot (1 ft = 0.3 m) grid
covering Europe. They were aggregated by averaging to 50 × 50 km
Universal Transverse Mercator (UTM) to match the species data grid.
Niche-based models were calibrated on the 50 × 50 km UTM grid, and
modeled species distributions were projected back onto the 10′ grid for
current and future climate.
Future projections were derived by
using climate model outputs made available through the
Intergovernmental Panel on Climate Change (IPCC) Data Distribution
Centre (ipcc-ddc.cru.uea.ac.uk).‡‡ The modeled climate anomalies were scaled based on four scenarios proposed by the IPCC (12).
The A1 scenario describes a globalized world with rapid economic growth
and global population that peaks in mid-century and declines thereafter
and assumes rapid introduction of new and more efficient technologies.
Concentrations of CO2 increase from 380 ppm in 2000 to 800 ppm in 2080, and temperature rises by 3.6 K (12).
The A2 scenario describes a heterogeneous world with regionally
oriented economic development. Per capita economic growth and
technological change are slower than in the other scenarios. Global
concentrations of CO2 increase from 380 ppm in 2000 to 700
ppm in 2080, and temperature rises by 2.8 K. The B1 scenario describes
a convergent world with global population that peaks in mid-century and
declines thereafter, as in A1, but with a rapid change toward a service
and information economy and the introduction of clean and
resource-efficient technology. Concentrations of CO2
increase from 380 ppm in 2000 to 520 ppm in 2080, and temperature rises
by 1.8 K. The B2 scenario describes a world in which the emphasis is on
local solutions to socioeconomic and environmental sustainability. It
is a world with continuously increasing global population (at a rate
lower than A2), intermediate levels of economic development, and less
rapid and more diverse technological change than in the B1 and A1
scenarios. Concentrations of CO2 increase from 380 ppm in 2000 to 550 ppm in 2080, and temperature rises by 2.1 K (12).
did not assess the impacts of land-use change, even though this factor
will potentially compound the effects of climate change on species
However, given the spatial extent and resolution of our data and the
magnitude of climate change in most projections, the effect of land use
would be most likely overridden by climate (26, 27).
Niche-Based Models of Species Climatic Envelops. We used the biomod
framework, which capitalizes on several widely used niche-based
modeling techniques (generalized linear models, generalized additive
models, classification tree analysis, and artificial neural networks)
to provide alternative spatial projections (13).
For each climate change scenario, models relating species distributions
to the seven bioclimatic variables were fitted by using biomod
and projected into the future. Then, a consensus principal component
analysis was run to explore central tendencies in projections and
select the niche-based model representing the greatest commonality
among projections (8).
is increasing evidence that model projections can be extremely
variable, and there remains a need to test the accuracy of models and
to reduce uncertainties (8, 28, 29).
One recent analysis has however provided the first test of the
predictive accuracy of such models by using bird observed species'
range shifts and climate change in two periods of the recent past (30).
This work provides validation of niche-based models under climate
change and demonstrated how uncertainty can be reduced by selecting the
most consensual projections, as done in this study. We are therefore
confident that this strategy provides a robust and defensible approach
to species range projections for the purposes of conservation planning
and biodiversity management.
To evaluate species extinction
risks, we summed the number of pixels lost, potentially gained (under
universal migration), or stable by each species for the different
climate-change scenarios. We assigned each species to an International
Union for Conservation of Nature and Natural Resources (IUCN) threat
category (IUCN 2001). Those that were not listed were classified as
lower risk, depending on the projected reduction in range size from
present to 2080. Present and future range sizes (area of occupancy)
were estimated from the number of pixels where species occurred. Loss
in range size was calculated by subtracting future potential range size
from present potential range size. In line with IUCN Red List criterion
A3(c), the following thresholds were then used to assign a species to a
threat category (IUCN 2001). Extinct is a species with a projected
range loss of 100% in 50 or 80 years, critically endangered has a
projected range loss of >80%, endangered has a projected range loss
of >50%, and vulnerable has a projected range loss of >30%.
Although this Red Listing approach is simplistic and considers only the
effects of climate change, it provides a synthetic overview of
species-specific threats due to climate change.
To evaluate the percentage of extinctions for a given area, we summed the number of species lost (L)
by pixel and related it to current species richness by pixel. The
procedure was the same to assess the percentage of species gained (G)
by pixel (under assumptions that species could reach a new suitable
climate space). Percentage of species turnover by pixel, under the
assumption of universal migration, is then given by T = 100 × (L + G)/(SR + G) where SR is the current species richness.