Content uploaded by Leslie Gordon Underhill
Author content
All content in this area was uploaded by Leslie Gordon Underhill on Nov 19, 2017
Content may be subject to copyright.
© 2007 South African National Biodiversity Institute DOI: 10.1111/j.1472-4642.2007.00391.x
Journal compilation © 2007 Blackwell Publishing Ltd www.blackwellpublishing.com/ddi
645
Diversity and Distributions, (Diversity Distrib.)
(2007)
13
, 645–653
BIODIVERSITY
RESEARCH
ABSTRACT
While poleward species migration in response to recent climatic warming is widely
documented, few studies have examined entire range responses of broadly distributed
sessile organisms, including changes on both the trailing (equatorward) and the
leading (poleward) range edges. From a detailed population census throughout
the entire geographical range of
Aloe dichotoma
Masson, a long-lived Namib Desert
tree, together with data from repeat photographs, we present strong evidence that a
developing range shift in this species is a ‘fingerprint’ of anthropogenic climate
change. This is explained at a high level of statistical significance by population
level impacts of observed regional warming and resulting water balance constraints.
Generalized linear models suggest that greater mortalities and population declines
in equatorward populations are virtually certainly the result, due to anthropo-
genic climate change, of the progressive exceedance of critical climate thresholds
that are relatively closer to the species’ tolerance limits in equatorward sites.
Equatorward population declines are also broadly consistent with bioclimatically
modelled projections under anticipated anthropogenic climate change but, as yet,
there is no evidence of poleward range expansion into the area predicted to
become suitable in future, despite good evidence for positive population
growth trends in poleward populations. This study is among the first to show a
marked lag between trailing edge population extinction and leading edge range
expansion in a species experiencing anthropogenic climate change impacts, a
pattern likely to apply to most sessile and poorly dispersed organisms. This provides
support for conservative assumptions of species’ migration rates when modelling
climate change impacts for such species.
Aloe dichotoma
’s response to climate
change suggests that desert ecosystems may be more sensitive to climate change
than previously suspected.
Keywords
Bioclimatic modelling, desertification, extinction, global warming fingerprints,
migration, range shift.
INTRODUCTION
Studies of the responses of terrestrial organisms to twentieth
century warming have focused on range shifts of motile
organisms (e.g. Hersteinsson & Macdonald, 1992; Parmesan
et al
., 1999; Warren
et al
., 2001; Parmesan & Yohe, 2003),
stressing poleward range boundary extensions (Parmesan
et al
.,
1999). The studies of sessile organisms (Keeling
et al
., 1996;
Menzel & Fabian, 1999; Abu-Asab
et al
., 2001; Fitter & Fitter,
2002) have focused on non-lethal changes in the growth
patterns. Both types of studies have focused most frequently on
only a portion of a species’ range.
There is concern that the adaptive responses of sessile
organisms to rapid climate change may be constrained, thus
causing population extinctions at the so-called ‘trailing edge’ of
species’ geographical ranges that are shifting in response to a
1
Global Change and Biodiversity Program,
2
Protea Atlas Project, South African National
Biodiversity Institute, Private Bag X7,
Claremont 7735, South Africa;
3
Threatened
Species Programme, South African National
Biodiversity Institute, Private Bag X101,
Pretoria 0001, South Africa;
4
Percy FitzPatrick
Institute of African Ornithology,
5
Department of
Botany,
6
Leslie Hill Institute of Plant
Conservation, University of Cape Town, Private
Bag X3, Rondebosch 7701, South Africa;
7
Avian
Demography Unit, Department of Statistical
Sciences, University of Cape Town, South Africa,
Private Bag X3, Rondebosch, 7701, South Africa;
8
Center for Applied Biodiversity Science,
Conservation International, 1919 M St NW,
Suite 600, Washington DC 20036, USA,
9
Environment Systems, ADAS, Woodthorne,
Wergs Road, Wolverhampton WV6 8TQ, UK,
10
Laboratoire d’Ecologie Alpine, UMR CNRS
5553, Université J Fourier, Grenoble Cedex 9,
France
*Correspondence: Wendy Foden, Threatened
Species Programme, South African National
Biodiversity Institute, Private Bag X101, Pretoria
0001, South Africa. E-mail: Foden@sanbi.org
Blackwell Publishing Ltd
A changing climate is eroding the
geographical range of the Namib Desert
tree
Aloe
through population declines
and dispersal lags
Wendy Foden
1,3,4
*, Guy F. Midgley
1,8
, Greg Hughes
9
, William J. Bond
5
,
Wilfried Thuiller
1,10
, M. Timm Hoffman
6
, Prince Kaleme
4
, Les G. Underhill
7
,
Anthony Rebelo
2
and Lee Hannah
8
W. Foden
et al.
© 2007 South African National Biodiversity Institute
646
Diversity and Distributions
,
13
, 645–653, Journal compilation © 2007 Blackwell Publishing Ltd
changing climate (Davis & Shaw, 2001). Such trailing edge
extinctions should be first detected in terrestrial organisms that
occupy extreme climatic environments such as deserts, where
climate-related stresses exert direct control over population
processes, especially at range margins (Jordan & Nobel, 1979).
It is implicitly assumed by bioclimatic modelling approaches that
trailing edge extinctions will be accompanied by simultaneous
leading edge range expansions (Parmesan
et al
., 1999), but lags
in population expansions due to limiting rates of dispersal and
establishment (Pitelka, 1997) undermined this assumption,
especially for sessile species. Such lags have been observed, for
example, in the difference in responsiveness of alpine plant range
responses relative to mobile organisms such as butterflies
(Walther
et al
., 2002), and predicted for temperate zone
American trees (Iverson
et al
., 2004). Such constraints are likely
to squeeze the ranges of non-motile organisms between a zone of
active population die-back and of constrained expansion,
thereby raising their risk of extinction, at least temporarily, even
if climate change trends begin to stabilize.
Here we investigate impacts of regional climate change on
population die-back in the long-lived, giant Namib Desert tree
(
Aloe dichotoma
Masson), prompted by casual observations of
widespread mortalities apparently induced by drought (i.e.
water balance constraints). We test whether this pattern of
population level mortality is in fact consistent with water balance
changes, and furthermore, whether a spatial pattern consistent
with the expected signal of anthropogenic climate change exists
(i.e. higher mortality in equatorward, trailing edge sites, and
lower mortality or stable/growing populations at poleward, lead-
ing edge sites). Such a pattern of die-back would provide credible
evidence for the first time of the negative impacts of anthropogenic
climate change on a sessile desert organism of the southern
Hemisphere, especially if it could be linked with observed
regional climatic changes, is consistent with modelled climate
changes, and concurs with future projections of anthropogenic
climate change impacts.
To accomplish this test, we conducted a detailed population
census and related studies on this plant species throughout
its entire geographical range in the Namib Desert and adjacent
arid regions of southern Africa (a reach of roughly 11 degrees
of latitude, between ~32
°
S and 21
°
S), in a region projected
to experience significant warming and drying due to anthropo-
genic climate change (Hulme
et al
., 2001). The stem-succulent
growth form, succulent leaves, and shallow root systems of
this species are common adaptations for rapid absorption
and storage of water to allow survival through frequent
droughts (Jacobsen, 1960; Barkhuizen, 1978). Individuals
grow up to 10 m tall and usually occur in dense populations
of up to 10,000 trees. A long life span [at least 200 years
(Vogel, 1974) and possibly up to 350 years], and large geo-
graphical range (~200,000 km
2
) make this a useful subject for
a study of the impacts of long-term climate trends. In particular,
because dead individuals decay relatively slowly
in situ
,
often remaining standing for many years, it is possible to obtain
a comparable measure of population mortality throughout
the species’ range.
METHODS
Characteristics of
Aloe
population mortality and mean
individual condition
Where possible, we selected large, discrete populations and
measured 100 live individuals in the densest part of each. We
measured smaller populations where they provided a useful
extension to the species’ range and where no larger populations
could be found. In such cases, the entire population was meas-
ured. Estimates for each live individual included degree of fungal
pathogen infection of leaves and trunk (scores from 0 = no infec-
tion to 3 = severe infection); extent of apparent animal damage
to the stem or trunk (scores from 0 = no damage to 3 = severe
damage); the percentage of the total potential canopy that had
been shed; and the percentage of the total potential canopy in
which leaf abscission (shedding) had occurred. These statistics
were averaged per population in the final analysis.
The number of dead trees within the area covered by a minimum
polygon around the sampled live trees was used to derive the
percentage of each population’s mortality. In order to test this
method for representivity of the whole population, we also
used a 1000
×
3 m linear transect spanning the population’s
altitudinal range and passing through its densest part to derive
an estimate of population density. This provided a second
estimate of relative mortality that was in agreement with the first
estimate. Mortality from the transect analysis of mortality was
positively correlated with that calculated from the survey method
(
r =
0.86,
n
= 36,
P
< 0.001, slope 0.86, intercept 5.15) with the
intercept of this regression not significantly different from 0,
and the slope not significantly different from 1.
Matched photographs
Photographs of
A. dichotoma
populations recorded 41–98 years
previously were relocated and precisely matched using tech-
niques broadly described by Rohde (1997). We counted all visible
individuals in both of each pair of matched photographs (i.e.
original and recent), and calculated the mean annual population
change as a percentage from the equation:
N
recent
=
N
original
·
r
t
or
r
= (
N
recent
/
N
original
)
1/
t
where
N
recent
is the total number of individuals in the recent photo-
graph,
N
original
the number in the original photograph,
r
is the
compounded annual growth rate (which is multiplied by 100 to
convert to percent), and
t
is the time difference in years between
the photographs. Mean birth and death rates could also be calcu-
lated from the photographs but are not reported here.
We attempted to track the rate of decay of dead individuals
between matched photographs, because a potential latitudinal
bias in decay rates might skew measures of mortality (e.g. a
potentially slower decay rate at drier sites towards the equator
may allow more dead individuals to accumulate). Long-term
Namib Desert trees feel the heat of climate change
© 2007 South African National Biodiversity Institute
Diversity and Distributions
,
13
, 645–653, Journal compilation © 2007 Blackwell Publishing Ltd
647
monitoring has now been established to measure decay rates of
dead trees at different sites across the range, but at this stage we
assume that decay rates are consistent throughout
A. dichotoma
’s
range. The lack of geographical trends in fungal infection of
both leaves and trunks supports this assumption, and it is even
possible that decay rates are higher towards the equator with
higher insolation levels and longer days leading to more rapid
photo-degradation. None of the dead trees visible in the initial
photographs was apparent after the 41–98 years interval between
repeat photographs, indicating that we are reporting mortality
trends in the past 40 years at most.
Climate records
To e xamine recorded past climatic trends within
A. dichotoma
’s
range, we used data from the 15 stations in Namibia and
north-western South Africa from which temperature records
for at least 24 years were available (Table 1). In order to ensure
that no periods of inconsistency or inaccuracy occurred in
the records, we tested each time series for stationarity by
visually assessing deviance from a straight line when cumulative
temperature and rainfall values for neighbouring stations
were plotted against each other (Gordon
et al
., 1992). No
stationarity effects were apparent. An index of water balance
was calculated from this data set, as the difference between
precipitation and potential evapotranspiration, the latter
calculated using monthly means of mean daily temperature and
total monthly precipitation (Thornthwaite, 1948). Annual trends
(calculated from April to March in order to represent a growth
year) were analysed using Robust MM Regression (Rosseeuw &
Yohai, 1984).
In order to test the relationship between recorded changes in
water balance and observed
A. dichotoma
population mortality,
water balance data from eight stations that were within 100 km
of one or more study populations were used in a simple
regression analysis. A single coastal station that qualified for this
analysis was not considered due to the likely steep environmental
gradient towards the interior from its coastal position.
We modelled adult mortality of
A. dichotoma
using generalized
linear models with a binomial distribution and logistic link
function (McCullagh & Nelder, 1989). In order to test if
measures of the condition of adult individuals and adult
mortality throughout the entire range of the species could be
related to climate change over the latter half of the twentieth
century (the maximum period over which dead individuals
could be observed), we also developed a set of key climate
parameters for 1960 and 2000, using the CRU CL 2.0 dataset
(New
et al
., 2000) to represent current climate (2000), and the
GCM HadCM3 to derive modelled changes between 1960 and
2000 (incorporating both natural and anthropogenic sources
of climate change). The CRU data were than adjusted using
modelled changes to provide matching 1960 climate surfaces for
each study site. These variables were used as explanatory variables
in the generalized linear modelling, as was the mean altitude of
the study site, and its latitude. Latitude data were converted
to a biologically more meaningful measure of mean annual solar
angle calculated by summing the zenith of the sun at midday for
each day of the year, dividing by 365 and subtracting this value
from 90
°
. The resulting index is the average angle of the sun
above the horizon at noon – large values indicate high mean
annual solar angle.
Bioclimatic modelling
To produce future scenarios of the possible impacts of climate
change across
A. dichotoma
’s range, we used the generalized
Table 1 Stations in Namibia and South Africa within the range of Aloe dichotoma Masson at which temperature and rainfall records were
available for 24 or more years. The table includes the mean decadal temperature change (°C) and the mean decadal water balance change (mm)
for each station. Figures in bold were found to show statistically significant trends (P < 0.01) using robust regression.
Weather station
Latitude
(degrees South)
Longitude
(degrees East)
Start
date
Time-series
duration (years)
Decadal temperature
change (degrees)
Decadal water
balance change (mm)
Okaukuejo 19.183 15.917 1975 26 0.213 –3.66
Sitrusdal 19.933 16.383 1976 24 0.481 –11.25
Windhoek 22.567 17.100 1960 43 0.199 –4.34
Keetmanshoop 26.533 18.117 1970 31 0.269 –2.34
Luderitz 26.633 15.100 1960 39 0.062 –1.07
Upington 28.450 21.250 1952 51 0.308 –1.37
Alexander Bay 28.620 16.480 1952 50 0.141 –1.02
Henkries 28.970 18.100 1960 41 0.273 –2.39
Douglas 29.070 23.750 1976 24 –0.012 0.56
Pofadder 29.130 19.380 1941 59 0.170 0.43
Port Nolloth 29.230 16.870 1960 42 0.191 –0.74
Okiep 29.620 17.880 1959 27 0.163 –1.55
Prieska 29.670 22.750 1959 44 0.364 –0.48
Brandvlei 30.470 20.480 1960 32 0.075 –1.80
Calvinia 31.470 19.770 1959 44 –0.031 –0.86
W. Foden
et al.
© 2007 South African National Biodiversity Institute
648
Diversity and Distributions
,
13
, 645–653, Journal compilation © 2007 Blackwell Publishing Ltd
additive modelling (GAM) function in the bioclimatic niche
modelling tool BIOMOD (McCarthy
et al
., 2001; Thuiller, 2003).
The CRU CL 2.0 data set (New
et al
., 2000) was used to represent
current (2000) climate, and future (2050) climate modelling was
based on the moderate climate change projections produced by
the HADCM3 General Circulation Model (Gordon
et al
., 2000)
using the B2 IPCC SRES scenario (Nakicenovic & Swart, 2000).
The sampled distribution of
A. dichotoma
and matching climatic
surfaces at a resolution of 10 min were then used to derive an
optimal bioclimatic model of
A. dichotoma
’s current geographical
range, and its response to future climate change modelled
spatially.
RESULTS AND DISCUSSION
Population-level mortality for the 53 populations examined
throughout the range of
A. dichotoma
ranged between 2% and
71% (mean = 32.1 ± 20.0%, Fig. 1), with a significant trend of
decreasing mortality from the equatorial to the poleward regions
of the species’ range (
r =
0.393,
n
= 53,
P
= 0.004; Fig. 2). A
regression of the residuals from this analysis against the altitude
of each population also produced a significant trend (
r =
0.331,
n
= 53,
P
= 0.016) – showing that low-altitude (warmer)
populations show higher mortalities than do high-altitude
(cooler) populations, particularly at the equatorial range edge.
Overall, populations at poleward and high-altitude extremes
tended to experience minimal mortality. Thus, population level
mortality decreased along latitudinal and altitudinal clines in a
pattern consistent with that expected under climate change,
with the highest mortality generally in the ‘trailing edge’
(Davis & Shaw, 2001) of the species’ geographical range.
We also sought to describe these data using more biologically
meaningful explanatory variables. The simplest biologically
meaningful model fitted using the generalized linear model
included four explanatory variables and accounted for 27.2% of
the deviance (Table 2). The explanatory variables were all abiotic.
As mean annual solar angle increased latitudinally towards
the equator, the probability of encountering live individuals in
a given population decreased. The probability of encountering
live individuals also increased with altitude. When both evapo-
transpiration in 1960 and the difference in evapotranspiration
between 1960 and 2000 were included as explanatory variables,
the model suggested that the probability of encountering
live individuals decreased at sites that had experienced larger
Figure 1 Map showing mortality of Aloe dichotoma Masson
populations across the species’ range. Green circles indicate
populations with low mortality and red circles represent
progressively higher mortality. Lighter and darker shades of
background grey represent lower to higher altitudes respectively.
For the four northernmost populations, altitudes have been
specified on the map because two of the populations occur on
inselbergs. Mortality appears patchy, but a general trend of lower
mortality in southern populations is apparent, irrespective of
altitude, and northern populations have increasingly greater
mortality, except at high altitudes.
Figure 2 Graph showing the relationship between latitude (degrees
South) and Aloe dichotoma Masson population mortality. Mortality
follows a latitudinal cline of increase from the poleward to the
equator-ward side of the species’ range, as indicated by the solid
regression line (r = 0.393, n = 53, P = 0.004). The mean altitudes
at which these populations occur are indicated by (< 700 m),
(700–1000 m) and (> 1000 m). When regressed against
altitude, the residuals of the above regression produced a significant
trend (r = 0.331, n = 53, P = 0.016) showing that where mortality
was lower than expected from the latitudinal trend, this can be
explained by higher altitude of the population and conversely, that
populations at low altitudes have higher than expected mortality.
In combination, these results strongly suggest that A. dichotoma is
undergoing the beginning of a systematic poleward and upwards
range shift.
Namib Desert trees feel the heat of climate change
© 2007 South African National Biodiversity Institute
Diversity and Distributions
,
13
, 645–653, Journal compilation © 2007 Blackwell Publishing Ltd
649
increases in evapotranspiration over the past four decades
(Table 2).
The generalized linear model for juveniles included three
explanatory variables (altitude was not found to be a significant
explanatory variable) and accounted for 14.7% of the deviance
(Table 3), about half that accounted for by the model for live
adult plants. Nonetheless, the model for juveniles suggests
the same general pattern as for live adult plants. We therefore
conclude that the same broad factors that lead to mortality also
inhibit germination, or diminish recruitment by causing juvenile
mortality.
Repeat photographs also suggest a latitudinal cline in
A. dichotoma
mortality. The rates of change in population size varied between
decreases of 4.73% per year and increases of 0.91% per year. These
survival rates were negatively related to the index of mean annual
solar angle (
r
7
= –0.70,
P
= 0.036), and negatively related to latitude
(Fig. 3), supporting the ‘snapshot’ observation of greater proportions
of dead plants towards the equator shown in Fig. 2. The altitudinal
range among the repeat photograph sites (300 m) was too small to
test for altitude effects, especially given this small sample size.
Repeat photographs show that during the twentieth century,
equatorward populations have experienced declines in population
numbers of between 0.7% and 5% per annum, and poleward
populations have increased their population size by between
0.25% and 0.91%. There is, further, a significant linear relation-
ship between latitude and population growth rate (
r =
0.70,
n
= 9,
P
< 0.05, Fig. 4), and while it is possible that this relation-
ship represents a threshold response, the data presented here,
especially for equatorward populations, are too few to test this
hypothesis more fully.
What is the mechanism causing adult mortality in equatorward
populations? In a large number of
A. dichotoma
individuals, we
observed that terminal leaf rosettes slowly withered and eventually
dropped to the ground, while the apical tips of supporting
branches died under drought conditions (we term this ‘shoot
shedding’). These branches never re-developed leaves, presumably
due to the loss of apical meristem. Our field observations and
anecdotal evidence from local experts strongly suggest that shoot
shedding is a typical response to water deficits in this species
(Van Jaarsveld & Chown, 2001). We found a strong logarithmic
relationship between the mean percentage of individuals’
canopies lost to shoot shedding and population mortality
(
r =
0.777,
n = 24, P < 0.001, Fig. 5), suggesting that high-
mortality populations are experiencing long-term water stress.
Water stress in plants is the result of the interaction between
precipitation and atmospheric vapour pressure deficit, itself a
function of air temperature and relative humidity. Desert rainfall
is variable and unsuited to trend analysis especially given the
length of records available, but atmospheric warming is more
reliably detected. Increasing air temperature is a key controller of
plant water status, and would increase the rate of water loss from
the succulent leaves of this CAM-photosynthetic species, espe-
cially during its night-time stomatal opening period, thereby
hastening leaf and shoot desiccation. This stress is well captured
by considering air temperature, the evaporative demand of the
air, and water balance that combines these two measures.
Continental-scale analyses of temperature records for the
twentieth century indicate that the Namib has undergone an
increase in temperature as well as a reduction in precipitation
(Hulme et al., 2001). Local climatic records revealed significant
Table 2 Results of the generalized linear model (binomial distribution, logit link function) relating the number of Aloe dichotoma Masson
individuals which were alive in a given population to abiotic explanatory variables. This model explained 27.2% of the deviance. The P-values
associated with all regression coefficients were less than 0.001 (i.e. t45 > 3.52).
Parameter
Regression
coefficient
Standard
deviation t-value t45
Constant 5.74 1.15 4.99
Mean annual solar angle (°)–0.1374 0.0168602 –8.20
Altitude (m) 0.001052 0.000102 10.30
Evaporation 1960 (mm) 0.03186 0.00287 11.11
Evaporation difference (mm) (2000–1960) –0.09144 0.00911 –10.03
Table 3 Results of the generalized linear model (binomial distribution, logit link function) relating the number of juvenile Aloe dichotoma
Masson individuals which were alive to explanatory variables. The binomial totals were the number of individuals, both dead and alive, at the study
sites. This model explained 14.7% of the deviance. The P-values associated with all regression coefficients were less than 0.001 (i.e. t48 > 3.50).
Parameter
Regression
coefficient
Standard
deviation t-value t48
Constant 3.2556 2.068 1.5863
Mean annual solar angle (°)–0.132690 0.0294331 –4.51
Evaporation 1960 (mm) 0.0292947 0.0047768 6.146
Evaporation difference (mm) (2000–1960) –0.0677 0.00142 –4.78
W. Foden et al.
© 2007 South African National Biodiversity Institute
650 Diversity and Distributions, 13, 645–653, Journal compilation © 2007 Blackwell Publishing Ltd
regional climate warming. We examined temperature and rain-
fall records from all available long-term weather stations in the
regions in which A. dichotoma occurs (see Table 1). Fifty-three
per cent of stations showed significant increases in temperature
over the last 25–60 years while none showed a significant decline.
There was no relationship between the duration of weather sta-
tions’ time series and the magnitude of their mean temperature
change. The mean decadal increase across all stations during this
interval was 0.2 ± 0.1 °C. Water balance, a composite measure of
temperature and rainfall reflecting the water available to plants,
showed a significant decline at 33% of stations over the last
25–60 years. No stations showed a significant increase in water
balance over this period. The relative severity of cumulative
water stress in A. dichotoma is reflected by the percentage of
months during the past decade in which water balance fell below
–90 mm. Population mortality at study sites within 100 km of
the long-term weather stations is positively correlated with this
measure of cumulative water stress (r = 0.491, n = 22, P = 0.021;
Fig. 6), strongly suggesting that a combination of water and heat
stress is responsible for the increased mortality in declining A.
dichotoma populations.
Although the areas in which A. dichotoma occurs are too arid
for agriculture, parts of its range are used for stock and game
farming. The species is considered to be unpalatable, yet we
observed a degree of herbivory by sheep, goats, donkeys,
antelope, baboons, and porcupines, but no correlation between
canopy herbivory and population mortality. We found mortality
to be weakly related to degree of stem bark damage, presumably
by herbivores (r = 0.406, n = 35, P < 0.02), but as it is unrelated
to herbivore density (measured as the frequency of dung pellets
in the area; r = 0.11, n = 28, P > 0.56), we conclude that some
herbivory of A. dichotoma trunks (likely by porcupines) probably
occurs only under conditions of extreme drought when more
palatable food and water sources are unavailable. However, when
we examined all other reasonable explanations for the observed
mortality patterns in A. dichotoma, such as intraspecific com-
petition, pathogen infection, pollution damage, and including
exposure to human or animal persecution, we found that the
Figure 3 Repeat photographs of populations of Aloe dichotoma Masson taken in 1904 and 1918 and precisely matched in 2002. Photos (a) and
(b) were taken at Hantamsberg, Calvinia (31°12.5′-South, 19°43.3′-East) in 1904 (Marloth) and 2002, respectively. From this and another pair
of matched photographs taken at the site, a decadal population increase of 0.76% per year is inferred. Photographs (c) and (d) were taken in 1918
(Evans) and 2002 in the Westerberg, near Koegas (29°19.9′ S, 22°18.5′ E). A decadal population decrease of –0.85% per year is inferred from two
matched photograph pairs taken at this site.
Namib Desert trees feel the heat of climate change
© 2007 South African National Biodiversity Institute
Diversity and Distributions, 13, 645–653, Journal compilation © 2007 Blackwell Publishing Ltd 651
addition of any of these factors as explanatory variables to the
generalized linear model of Table 2 resulted in only small changes
to the regression coefficients for the abiotic variables, but did not
modify the overall patterns described earlier.
Although the southern third and northern two-thirds of A.
dichotoma’s range are separated by the political border between
Namibia and South Africa, neither historical nor current land
use and management methods differ markedly between the two
countries. The areas in which Aloe dichotoma occurs are generally
too sparsely vegetated for interspecific plant competition to limit
recruitment success. The ‘nurse rocks’ and ‘nurse shrubs’ in
which juveniles were frequently found germinating, probably
due to the ameliorated microhabitats they create, are present at
almost all localities, throughout the species’ range and in the
regions south of it.
Finally, niche-based spatial modelling techniques (Thuiller,
2004) show changes between the years 2000 and 2050 in the
frequency of occurrence of A. dichotoma within each 10-min
band of latitude that indicates a projected poleward shift in this
species’ geographical range that is consistent with the local
extinction of equatorward populations currently showing high
rates of mortality (Fig. 7). The mean altitude of 10′ pixels in
which climate change models project the species’ presence
increases from 806 ± 354 m in 2000 to 885 ± 329 m in 2050
while the projected mean latitude shifts from 26.68 ± 2.81° S in
2000 to 27.90 ± 2.5° S. These modelled shifts equate to a mean
altitudinal increase of 16 m per decade and a poleward range
shift of 23 km per decade, considerably higher than the
6.1 ± 2.4 km per decade poleward range shifts recently collated
for a broad variety of organisms including woody and herba-
ceous plants (Parmesan & Yohe, 2003). Given an estimated
recruitment frequency of 15 years, A. dichotoma must disperse
approximately 35 km southwards and 24 m upwards with each
recruitment event in order to keep pace with its climatic niche.
No records exist of new populations of this conspicuous
species establishing in areas projected to become suitable for the
Figure 4 The relationship between latitude and mean annual
population growth rate of Aloe dichotoma Masson populations as
derived from repeat photographs taken between 41 and 98 years
apart (r = 0.70, n = 9, P < 0.05).
Figure 5 The relationship between A. dichotoma Masson
population-level mean loss of canopy due to shoot shedding and
population-level mortality (r = 0.777, n = 24, P < 0.001).
Figure 6 The relationship between the percentages of very dry
months in the last 10 years (when water balance fell below –90 mm)
and population-level mortality in Aloe dichotoma Masson at sites
within 100 km of the weather stations at which mortalities were
measured (r = 0.49, n = 22, P < 0.02).
Figure 7 Comparisons between present and bioclimatically
modelled future projections of the distribution range of Aloe
dichotoma Masson. The figure shows the modelled frequency of
A
. dichotoma in each 10-min latitude band of its range for 2000
(grey bars) and 2050 (black bars). The projected mean latitude
of the species’ range shifts from –26.68 ± 2.81 degrees South in
2000 to –27.90 ± 2.5 degrees South in 2050. This equates to a mean
altitudinal shift of 16 m per decade and a poleward range shift of
23 km per decade.
W. Foden et al.
© 2007 South African National Biodiversity Institute
652 Diversity and Distributions, 13, 645–653, Journal compilation © 2007 Blackwell Publishing Ltd
species in the poleward parts of its range. This is despite having
copious small, light wind-dispersed seeds and that it has been
successfully planted, and recruits autonomously, well beyond its
poleward range margin (I. Oliver, pers. comm.). While soil type
and biotic interactions (e.g. competition) seem an obvious limit-
ing factor for poleward migration for this species, soils of similar
nature are clearly widespread further South of its southern range
margin, and the issue of competition is not likely to be important
in these open, low-density communities. More work is needed to
determine if any of these factors, or simply the availability of
establishment sites, may be limiting range expansion of this species.
The Succulent Karoo biodiversity hotspot lies almost entirely
within the range of A. dichotoma, and represents the planet’s
richest arid biodiversity hotspot by far (Myers et al., 2000). The
hotspot contains over 5000 species, 40% of which are endemic
(Cowling & Hilton-Taylor, 1999). Recent empirical work has
shown the sensitivity of endemic Karoo succulent species to
warming treatments in the field (Musil et al., 2005).
Although it is extremely widespread and abundant, A. dichotoma
qualifies as Endangered (criterion A3ce) according to the IUCN
Red List Categories and Criteria for Red List Assessments (Standards
and Petitions Working Group, 2005) based on population loss
corresponding to modelled range loss (assuming a linear rela-
tionship) under a ‘null migration’ scenario. But in comparison
with other species in its biome and even globally, succulence,
gigantism, and a broad distribution range and bioclimatic niche
probably make A. dichotoma relatively more robust to drought
and climatic fluctuation. The species therefore provides a
conservative indicator of the impacts of regional warming and
drying in the Namib region. While insufficient data are available
to model range shifts of all species, it seems likely that doing so
would result in a substantial increase in the number of species
qualifying as Threatened (defined as ‘in immediate danger of
extinction’) according the IUCN Red Listing Categories and Cri-
teria (Standards and Petitions Working Group, 2005). Thus, a
large and rapid shift in conservation approach is clearly needed.
CONCLUSIONS
Our results suggest that A. dichotoma, a species with an extended
juvenile period, is experiencing population declines at its equa-
torward limits (i.e. its ‘trailing edge’, sensu Davis & Shaw, 2001) in
response to anthropogenic climate change trends. Generalized
linear modelling shows that A. dichotoma populations in equa-
torward regions are relatively closer to critical climate limits, and
that observed and modelled climate changes in the latter half of
the twentieth century have likely caused these to be exceeded,
resulting in elevated mortality and population declines. At the
same time, we report the species’ failure to expand polewards in
relation to its shifting climatic envelope, despite good evidence
suggesting positive population growth trends in established
populations at poleward latitudes.
The geographical range of A. dichotoma is therefore apparently
becoming progressively squeezed between an advancing equa-
torward zone of range contraction due to population declines,
and a poleward zone of lagging range expansion. This study is
among the first to document such an imbalance between con-
traction and expansion trends by looking at population status
throughout the entire geographical range of a species, but the
pattern could well be repeated for sessile and poorly dispersed
organisms globally.
Many projections of climate change impacts on biodiversity
attempt to incorporate uncertainty due to migration constraints
by contrasting ‘full migration’ and ‘null migration’ assumptions
(e.g. Peterson et al., 2002; Thomas et al., 2004; Thuiller, 2004).
The findings provided here provide support for conservative
assumptions of migration rate in sessile organisms, with impor-
tant implications for projections of species diversity under future
climate change scenarios.
Global assessments suggest that deserts will show a relatively
muted biodiversity response to climate change (Sala et al., 2000).
The results of this study argue against this conclusion and suggest
that desert ecosystems are likely to become increasingly hostile to
endemic biota, and thus more species-poor with intensifying
global warming.
ACKNOWLEDGEMENTS
Graeme Ellis provided invaluable field support. Bronwyn Stiles,
Nicola Berg, Trygve Cooper, and Shayne Fuller also provided
assistance in the field. The comments of three referees are much
appreciated. Nathan Wells and Alison Joubert assisted with
meteorological calculations. Meteorological data were pro-
vided by the South African Weather Services and the Namibian
Meteorological Services. This work was funded by the Centre for
Applied Biodiversity Science and the South African National
Biodiversity Institute. Many thanks to the communities and con-
servators of Namibia and the Northern Cape Province for their
cooperation and hospitality.
REFERENCES
Abu-Asab, M., Peterson, P., Shetler, S. & Orli, S. (2001) Earlier
plant flowering in spring as a response to global warming in
the Washington, DC, area. Biodiversity and Conservation, 10,
597– 612.
Barkhuizen, B.P. (1978) Succulents of Southern Africa. Purnell,
Cape Town, South Africa.
Cowling, R.M. & Hilton-Taylor, C. (1999) Plant biogeography,
endemism and diversity. The Karoo: ecological patterns and
processes (ed. by W.R.J. Dean and S.J. Milton), pp. 42–56.
Cambridge University Press, Cambridge, UK.
Davis, M.B. & Shaw, R.G. (2001) Range shifts and adaptive
responses to Quaternary climate change. Science, 292, 673–
679.
Fitter, A.H. & Fitter, R.S.R. (2002) Rapid changes in flowering
time in British plants. Science, 296, 1689–1691.
Gordon, C., Cooper, C., Senior, C.A., Banks, H., Gregory, J.M.,
Johns, T.C., Mitchell, J.F.B. & Wood, R.A. (2000) The simulation
of SST, sea ice extents and ocean heat transports in a version of
the Hadley Centre coupled model without flux adjustments.
Climate Dynamics, 16, 147–168.
Namib Desert trees feel the heat of climate change
© 2007 South African National Biodiversity Institute
Diversity and Distributions, 13, 645–653, Journal compilation © 2007 Blackwell Publishing Ltd 653
Gordon, N.D., McMahon, T.A. & Finlayson, B.L. (1992) Stream
hydrology: an introduction for ecologists. John Wiley & Sons,
Chichester, UK.
Hersteinsson, P. & Macdonald, D.W. (1992) Interspecific competition
and the geographical distribution of red and arctic foxes Vulpes
vulpes and Alopex lagopus. Oikos, 64, 505–515.
Hulme, M., Doherty, R., Ngara, T., New, M. & Lister, D. (2001)
African climate change: 1900–2100. Climate Research, 17,
145–168.
Iverson, L.R., Schwartz, M.W. & Prasad, A.M. (2004) How fast
and far might tree species migrate in the eastern United States
due to climate change? Global Ecology and Biogeography, 13,
209–219.
Jacobsen, H. (1960) A handbook of succulent plants. Blandford
Press, London.
Jordan, P.W. & Nobel, P.S. (1979) Infrequent establishment
of seedlings of Agave deserti (Agavaceae) in the north-
western Sonoran Desert. American Journal of Botany, 66,
1079–1084.
Keeling, C.D., Chin, J.F.S. & Whorf, T.P. (1996) Increased activity
of northern vegetation inferred from atmospheric CO2
measurements. Nature, 382, 146–149.
McCarthy, J.J., Canziani, O.F., Leary, N.A., Dokken, D.J. &
White, K.S. (2001) Climate change 2001: impacts, adaptation,
and vulnerability. Contribution of Working Group II to the
Third Assessment Report of the Intergovernmental Panel on
Climate Change Cambridge University Press, New York.
McCullagh, P. & Nelder, J.A. (1989) Generalized linear models.
Chapman & Hall, London.
Menzel, A. & Fabian, P. (1999) Growing season extended in
Europe. Nature, 397, 659.
Musil, C.F., Schmiedel, U. & Midgley, G.F. (2005) Lethal impacts
of elevated atmospheric temperature. New Phytologist, 165,
539–547.
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.
Nakicenovic, N. & Swart, R. (2000) Emissions scenarios: a special
report of Woking Group III of the Intergovernmental Panel on
Climate Change. Cambridge University Press, Cambridge, UK.
New, M., Lister, D., Hulme, M. & Makin, I. (2000) A high-
resolution data set of surface climate over global land areas.
Climate Research, 21, 1–25.
Parmesan, C., Ryrholm, N., Stefanescu, C., Hill, J.K., Thomas,
C.D., Descimon, H., Huntley, B., Kaila, L., Kullberg, J.,
Tammaru, T., Tennent, W.J., Thomas, J.A. & Warren, M. (1999)
Poleward shifts in geographical ranges of butterfly species
associated with regional warming. Nature, 399, 579–583.
Parmesan, C. & Yohe, G. (2003) A globally coherent fingerprint
of climate change impacts across natural systems. Nature, 421,
37–42.
Peterson, A.T., Ortega-Huerta, M.A., Bartley, J., Sanchez-
Cordero, V., Soberon, J., Buddemeier, R.H. & Stockwell,
D.R.B. (2002) Future projections for Mexican faunas under
global climate change scenarios. Nature, 416, 626–629.
Pitelka, L.F. (1997) Plant migration and climate change.
American Scientist, 85, 464–473.
Rohde, R.F. (1997) Looking into the past: interpretations of
vegetation change in Western Namibia based on matched
photography. Dinteria, 25, 221–249.
Rosseeuw, P.J. & Yohai, V. (1984) Robust regression by means of
S-estimates. Lecture Notes in Statistics, 26, 256–272.
Sala, O.E., Chapin, F.S. III, Armesto, J.J., Berlow, E., Bloomfield,
J., Dirzo, R., Huber-Sanwald, E., Huenneke, L.F., Jackson, R.B.,
Kinzig, A., Leemans, R., Lodge, D.M., Mooney, H.A., Oesterheld,
M.I.N., Poff, N.L., Sykes, M.T., Walker, B.H., Walker, M. &
Wall, D.H. (2000) Global biodiversity scenarios for the year
2100. Science, 287, 1770 –1774.
Standards and Petitions Working Group (2005) Guidelines for
using the IUCN red list categories and criteria, Version 3.1.
Prepared by the Standards and Petitions Working Group of the
IUCN SSC Biodiversity Assessments Sub-Committee in April
2005.
Thomas, C.D., Cameron, A., Green, R.E., Bakkenes, M.,
Beaumont, L.J., Collingham, Y.C., Erasmus, B.F.N., de Siqueira,
M.F., Grainger, A., Hannah, L., Hughes, L., Huntley, B., van
Jaarsveld, A.S., Midgley, G.F., Miles, L., Ortega-Huerta, M.A.,
To w nsend Peterson, A., Phillips, O.L. & Williams, S.E. (2004)
Extinction risk from climate change. Nature, 427, 145–148.
Thornthwaite, C.W. (1948) An approach toward a rational
classification of climate. Geographic Review, 38, 55 –94.
Thuiller, W. (2003) BIOMOD – optimizing predictions of
species distributions and projecting potential future shifts
under global change. Global Change Biology, 9, 1353 –1362.
Thuiller, W. (2004) Patterns and uncertainties of species’ range
shifts under climate change. Global Change Biology, 10, 2020–
2027.
Van Jaarsveld, A.S. & Chown, S.L. (2001) Climate change and its
impacts in South Africa. Trends in Ecology & Evolution, 16, 13 –
14.
Vogel, J.C. (1974) The life span of the Kokerboom. Aloe, 12, 66–
68.
Walther, G.-R., Post, E., Convey, P., Menzel, A., Parmesan, C.,
Beebee, T.J.C., Fromentin, J.-M., Hoegh-Guldberg, O. &
Bairlein, F. (2002) Ecological responses to recent climate
change. Nature, 416, 389–395.
Warren, M.S., Hill, J.K., Thomas, J.A., Asher, J., Fox, R., Huntley,
B., Roy, D.B., Telfer, M.G., Jeffcoate, S., Harding, P., Jeffcoate,
G., Willis, S.G., Greatorex-Davies, J.N., Moss, D. & Thomas,
C.D. (2001) Rapid responses of British butterflies to
opposing forces of climate and habitat change. Nature, 414,
65–69.
