See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/230867947

Cryptic biodiversity loss linked to global climate change

Article  in  Nature Reports Climate Change · September 2011

DOI: 10.1038/NCLIMATE1191

CITATIONS READS

340 2,602

9 authors, including:

Miki Bálint Sami Domisch

Senckenberg Biodiversität und Klima - Forschungszentrum Leibniz-Institute of Freshwater Ecology and Inland Fisheries
11 PUBLICATIONS   743 CITATIONS    87 PUBLICATIONS   3,568 CITATIONS   

SEE PROFILE SEE PROFILE

Peter Haase Kathrin Theissinger

Senckenberg Research Institute LOEWE centre for Translational Biodiversity Genomics Senckenberg Nature Resea…
236 PUBLICATIONS   8,727 CITATIONS    94 PUBLICATIONS   1,288 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Phylogeography View project

Range shifts o aquatic odonates View project

All content following this page was uploaded by Steffen U Pauls on 01 June 2014.

The user has requested enhancement of the downloaded file.

 LETTERS
PUBLISHED ONLINE: 21 AUGUST 2011 | DOI: 10.1038/NCLIMATE1191

Cryptic biodiversity loss linked to global

climate change
M. Bálint1,2 , S. Domisch1,3 , C. H. M. Engelhardt3 , P. Haase1,3 , S. Lehrian3 , J. Sauer1 , K. Theissinger4,5 ,
S. U. Pauls1 *† and C. Nowak1,3 *†

Global climate change (GCC) significantly affects distribu- from nine aquatic insect species of European mountain regions. We
tional patterns of organisms1 , and considerable impacts on examined seven species of caddisfly (Chaetopterygopsis maclachlani,
biodiversity are predicted for the next decades. Inferred ef- Drusus discolor, Drusus romanicus, Hydropsyche tenuis, Rhyacophila
fects include large-scale range shifts towards higher alti- aquitanica, Rhyacophila carpathica and Rhyacophila pubescens;
tudes and latitudes2 , facilitation of biological invasions3 and Trichoptera), the mayfly Ameletus inopinatus (Ephemeroptera) and
species extinctions1,3 . Alterations of biotic patterns caused the stonefly Arcynopteryx compacta (Plecoptera; Supplementary
by GCC have usually been predicted on the scale of taxo- Table S1). The selected species have alpine, Arctic–alpine or
nomically recognized morphospecies1 . However, the effects of endemic distributions (Supplementary Table S1) and are restricted
climate change at the most fundamental level of biodiversity— to cold, fast-running, oxygen-saturated streams with low organic
intraspecific genetic diversity—remain elusive4 . Here we show matter input7 . These habitats are limited to higher altitudes, leading
that the use of morphospecies-based assessments of GCC to insular distribution patterns of the species in the European
effects will result in underestimations of the true scale of biodi- mountains. Strong genetic population differentiation and high
versity loss. Species distribution modelling and assessments of proportions of regionally endemic haplotypes have been shown
mitochondrial DNA variability in nine montane aquatic insect for all but one range-restricted species (Supplementary Table S1),
species in Europe indicate that future range contractions will indicating very limited potential for long-distance dispersal. The
be accompanied by severe losses of cryptic evolutionary lin- species were chosen for their thermal sensitivity, their dispersal
eages and genetic diversity within these lineages. These losses limitation and the availability of unbiased locality data and range-
greatly exceed those at the scale of morphospecies. We also wide phylogeographic data sets. We applied an ensemble forecasting
document that the extent of range reduction may be a useful of species distributions to project future ranges of our study
proxy when predicting losses of genetic diversity. Our results species under the Intergovernmental Panel on Climate Change
demonstrate that intraspecific patterns of genetic diversity (IPCC) 2080 A2a (‘business as usual’) and the IPCC 2080 B2a
should be considered when estimating the effects of climate (‘reduced CO2 emissions’) emission scenarios8 . Assuming only
change on biodiversity. short-distance dispersal, primarily in the form of local altitudinal
Numerous studies document the effects of GCC on biodiversity shifts in response to climate warming, we inferred genetic diversity
both at the ecosystem and species level, but not at the level of losses by associating predicted range losses with present haplotype
intraspecific genetic diversity. This is surprising, given that the use distributions at the basin scale (see Methods).
of molecular techniques in biodiversity research increasingly results Projections of future range shifts (SDM) showed significant
in the recognition of high levels of cryptic biodiversity below the losses of climatically suitable areas for all species under both
morphospecies level5 . Efforts to delimit evolutionarily significant IPCC emission scenarios (Fig. 1 and Supplementary Fig. S1 and
units (ESUs) for biodiversity-related fields acknowledge the fact Table S3). Range contractions were most severe at the low-latitude
that the morphospecies concept seems increasingly insufficient edges of species ranges and in low mountain ranges (Fig. 1).
for holistic biodiversity estimates6 . However, a concise framework These areas are well known for high species diversity and for the
for estimating the effects of GCC on cryptic biodiversity and frequent occurrence of endemic intraspecific lineages9,10 . Three
spatial genetic differentiation is still lacking. Here we introduce species with restricted distribution are directly threatened by
a widely applicable approach based on the use of range-wide GCC-mediated extinction, as their present ranges are projected to
phylogeographic mitochondrial DNA data and species distribution become completely unsuitable or strongly reduced depending on
modelling (SDM), which allows estimation of climate-related the respective emission scenario (Fig. 1 and Supplementary Fig. S1
future changes of genetic and cryptic biodiversity. We reason that and Table S3). More extensive suitable areas under GCC were
species with strong genetic population structure will experience projected for each of the remaining six morphospecies. At present,
massive losses of cryptic diversity and ESUs under GCC, and that all of these species have relatively large distributions. Suitable
examining GCC effects solely at the level of morphospecies will areas will generally shift from the Central European highlands
underestimate the extent of climate-driven biodiversity loss. to higher elevations in the Alps. Extensive suitable areas were
To test our hypotheses, we used range-wide mitochondrial projected in the north for the two Arctic–alpine species (Fig. 1 and
sequence data (mitochondrial cytochrome c oxidase subunit I) Supplementary Fig. S1).

1 Biodiversityand Climate Research Centre (BiK-F), Senckenberganlage 25, 60325 Frankfurt am Main, Germany, 2 Molecular Biology Center, Babeş-Bolyai
University, Treboniu Laurian 42, 400271 Cluj, Romania, 3 Senckenberg Research Institutes and Natural History Museums, Department of Limnology and
Conservation, Clamecystrasse 12, 63571 Gelnhausen, Germany, 4 Institute of Zoology, Department of Ecology, Johannes Gutenberg-University Mainz,
Becherweg 13, 55128 Mainz, Germany, 5 Institute for Environmental Sciences, University of Koblenz-Landau, Fortstrasse 7, 76829 Landau, Germany. † These
authors contributed equally to this work. *e-mail: steffen.pauls@senckenberg.de; carsten.nowak@senckenberg.de.

NATURE CLIMATE CHANGE | VOL 1 | SEPTEMBER 2011 | www.nature.com/natureclimatechange 313

© 2011 Macmillan Publishers Limited. All rights reserved
LETTERS NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1191

Present

A2a

B2a

Arctic¬alpine distribution (A. compacta) Alpine distribution (D. discolor) Range-restricted distribution (D. romanicus)

Figure 1 | Projections of climatically suitable regions for three montane aquatic insect species representing different distribution types in Europe.
Distribution types are listed according to ref. 10. Top: present conditions; middle and bottom: IPCC 2080 emission scenarios A2a and B2a. Coloured
shading indicates modelled present and future ranges; dots show sampling localities. Grey areas display elevations >500 m above sea level. Left:
A. compacta; middle: D. discolor; right: D. romanicus. Analogous projections for the other species are shown in Supplementary Fig. S1. Projections of future
range shifts (SDM) show significant losses of climatically suitable areas for all species under both emission scenarios (paired t-test on
square-root-normalized absolute values of suitable area sizes of each species: present–A2a, t = 8.5127, df = 8, P = 0.00002; present–B2a, t = 10.7075,
df = 8, P = 0.00001).

Five examined species comprise two or more ESUs based on 2% under the severe emission scenario, and 68% under the more mod-
sequence divergence threshold (Fig. 2), and eight species comprise erate scenario (Figs 2 and 3 and Supplementary Table S3). These
two or more ESUs based on the generalized mixed Yule-coalescent estimated losses of genetic diversity correlate with losses of suitable
(GMYC) delineation (Supplementary Figs S2–S4 and Table S3). We areas (Fig. 4). This indicates that the extent of GCC-induced range
found that substantially more ESUs face extinction by the loss of reduction may be a good predictor of changes in genetic diversity,
climatically suitable areas when compared with the morphospecies, at least for species with limited migration potential.
regardless of the method used for ESU delineation (Figs 2 and 3 The highest levels of intraspecific genetic diversity losses
and Supplementary Figs S2–S4 and Table S2). Depending on the measured as number of haplotypes and lost ESUs are projected
method of ESU delimitation, 52% and 59% of ESUs will be lost in the Mediterranean region followed by the Central European
under the moderate CO2 emission scenario (Fig. 3); 65 to 79% of highlands and the Carpathians (Supplementary Tables S4–S6).
ESUs will be lost under the more severe greenhouse-gas emission In many European species, genetic diversity is greatest in
scenario. The results highlight the greater resolution achieved when regions of their putative Pleistocene glacial refugia in southern
using ESUs rather than morphospecies when predicting the impact Europe10 . In general, ancestral refugial populations also show
of GCC on biodiversity. This is particularly relevant as detailed stud- stronger genetic differentiation and exhibit the oldest splits9 .
ies of intraspecific genetic variation often yield previously unrec- Under our climate change projections, all but two populations
ognized, morphologically cryptic evolutionary lineages5 . Losses of in the classical southern refugia are projected to become
diversity under scenarios of GCC are even more severe when trans- extinct (two high-altitude Pyrenees populations of A. compacta
lated into haplotypes. A large number of haplotypes are geographi- are the exception; Supplementary Table S4 and S5). Thus,
cally restricted to areas that will not be climatically suitable for them ancestral, highly diverse populations will be lost. Among our
in the future. On average, 84% of haplotypes lose all of their areas study species, deep ancestral splits are seen in D. discolor,

314 NATURE CLIMATE CHANGE | VOL 1 | SEPTEMBER 2011 | www.nature.com/natureclimatechange

© 2011 Macmillan Publishers Limited. All rights reserved
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1191 LETTERS
A2a
AC l
AI ll A. compacta

A. inopinatus

AI l

D. discolor

DD l

RP lV

DD ll
DD III
R. pubescens DD IV
RP lll DD V
RP ll
RP l DD Vl
R. carpathica RC l DD Vll
DR l
RA ll DR ll
DR lll D. romanicus
R. aquitanica
RA l
HT l C. maclachlani
CM l
H. tenuis

B2a
AI ll AC l
A. compacta

A. inopinatus

AI l

D. discolor

DD l

RP lV

DD ll
R. pubescens DD III
DD IV
RP lll
RP ll DD V
RP l DD Vl
R. carpathica RC l DD Vll
DR l
RA ll DR ll
DR lll
R. aquitanica D. romanicus
RA l
HT l C. maclachlani
H. tenuis CM l

Figure 2 | Predicted loss of mitochondrial cytochrome c oxidase subunit I haplotypes for nine montane aquatic insect species in Europe under two IPCC
2080 CO2 emission scenarios. Top: A2a ‘business as usual’ scenario; bottom: B2a ‘reduced CO2 emission’ scenario. Shown is a combined UPGMA tree for
all species. For clarity, species are separated by background colour. White branches show lost haplotypes (no climatically suitable areas projected). Highly
diverged lineages (2% sequence divergence molecular operational taxonomic units) are marked on the outer circle. Molecular operational taxononmic
units codes are given outside the circle and follow Supplementary Table S5.

D. romanicus and R. pubescens. The other species with particularly severe in the Mediterranean region3 and many ancestral
Mediterranean populations (A. compacta and H. tenuis) show and genetically diverse southern lineages of temperate species are
shallow splits resulting from postglacial recolonization of Central likely to become extinct.
Europe from southern refugia and greatest diversity in the GCC effects on genetic diversity may differently impact leading-
recolonized regions of the Carpathians and Central European edge, central and rear-edge populations2,13 . Our study supports
highlands, respectively. This reflects the special situation of this pattern, with marked decreases of intraspecific diversity
montane species in Europe with high genetic diversity in the Central mainly predicted for the genetically most diverse low-latitude and
European highlands9,11,12 . In general, our study supports the idea low-mountain populations. However, the coarse resolution of our
that climate change-induced biodiversity losses are projected to be modelling approach cannot discount future suitable habitat patches

NATURE CLIMATE CHANGE | VOL 1 | SEPTEMBER 2011 | www.nature.com/natureclimatechange 315

© 2011 Macmillan Publishers Limited. All rights reserved
LETTERS NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1191

Morphospecies 2% MOTUs data and SDM, the drawn conclusions rely on several assumptions.
We used neutral mitochondrial DNA haplotype variation as a
1 surrogate for intraspecific genetic diversity. We specifically chose
8
mitochondrial DNA as it is the most common marker for large-scale
2 assessments of genetic population structure15 and it is widely used
12 for the identification of cryptic evolutionary lineages16 . Thus, the
6 particular strength of the approach lies in the wide availability
3 of such homologous data from varying organism groups of
different geographic regions and ecological guilds. With decreases
in sequencing costs, comparative homologous multi-locus data
GMYC species Haplotypes sets will become more readily available, allowing further improved
estimates of genetic diversity losses4 .
54 An important issue for SDM-based predictions of GCC effects
16
on biodiversity is the difficulty of implementing dispersal and
58 migration scenarios into projections of future distribution17 . The
44 discrepancy between the widespread use and importance of SDM
15 233
in ecological studies on the one hand, and the rare implementation
of sophisticated approaches for assessing dispersal on the other
hand, is largely due to a limited understanding of dispersal rates and
migration patterns for many taxa17 . Dispersal of immature stages
Figure 3 | Losses of morphospecies, ESUs and mitochondrial DNA (eggs, larvae and pupae) of aquatic insects is often accidental or
haplotypes combined for nine montane aquatic insect species in Europe passive and usually restricted to within-stream movement, whereas
under two IPCC 2080 CO2 emission scenarios. Absolute numbers of units dispersal among streams or catchments happens during the winged
that will persist under both future climate projections (blue), units that will adult life stages18 . Here we assumed that over the short timescales
be lost only under the ‘business as usual’ scenario (A2a; pink) and units relevant to this study, cold-adapted insects will generally be limited
lost under both scenarios (grey). MOTUs, molecular operational to short-distance compensatory altitudinal shifts in response to
taxonomic units. climate warming. Our rationale for restricting dispersal to within
basin movements is based on the general dispersal limitation of
in small pockets of favourable microclimate conditions, for example our study taxa. This assumption is supported by high rates of
in springs. It has been proposed13 that low-latitude rear-edge pop- genetic differentiation found between populations (Supplementary
ulations of numerous species survived past climate warming in very Table S1). For instance, it was shown that D. discolor migrated
restricted climatic islands. We also consider future microrefugia altitudinally but not longitudinally during warm phases in the
possible for highland species; however, persisting populations will interglacial9 . Thus, we considered gene flow among river basins
probably be highly vulnerable as a result of severely reduced effec- rare—particularly in the timeframe of ∼70 years relevant for
tive population sizes. Although GCC will probably result in genetic this study—and used watersheds as proxies for dispersal barriers
impoverishment in rear-edge populations, shifting ranges may also for larvae and adults. This approach seems suitable for other
enhance global intraspecific genetic diversity. This may result from montane or Arctic–alpine aquatic taxa with restricted over-land
local adaptation in leading-edge populations2 or increased local dispersal, particularly in hololimnic organisms18 . For other species
intraspecific genetic diversity through hybridization of historically that show greater long-distance dispersal propensities, however, the
isolated lineages after a climate-driven secondary contact14 . implementation of dynamic models that consider more complex
Although our data indicate the usefulness of combining dispersal scenarios17 will be of crucial importance when inferring
phylogeographic data sets based on mitochondrial DNA sequence GCC effects on genetic diversity.
70 60
A2a B2a

60
50

50
40
Haplotypes lost

Haplotypes lost

40
30
30

20
20

10
10

0 0
0 10,000 20,000 30,000 40,000 50,000 0 10,000 20,000 30,000 40,000 50,000
Area lost (km2) Area lost (km2)

Figure 4 | Correlation between predicted losses of area and inferred losses of haplotypes. Left: A2a IPCC 2080 ‘business as usual’ CO2 emission
scenario; n = 9, R = 0.79, P = 0.012; Right: B2a IPCC 2080 ‘reduced CO2 emissions’ scenario; n = 9, R = 0.73, P = 0.025. Pearson correlations were carried
out on square-root-normalized absolute values of haplotype and area losses.

316 NATURE CLIMATE CHANGE | VOL 1 | SEPTEMBER 2011 | www.nature.com/natureclimatechange

© 2011 Macmillan Publishers Limited. All rights reserved
NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1191 LETTERS
Genetic diversity has important ecological consequences for WorldClim and used it for modelling as an important proxy for the occurrence of
populations, communities and ecosystems19 , because it provides montane aquatic insects. SDMs were fitted using six algorithms as implemented in
the basis for phenotypic plasticity, local adaptation and adaptability the BIOMOD (ref. 30) package version 1.1.5 in R. Potential future distributions
of the species under 2080 climatic conditions were estimated using three climate
to changing environmental conditions20 including GCC21 . For models (CSIRO, HadCM3 and CCCMA) under two 2080 CO2 emission scenarios8 .
most species, including our study species, we have a limited We summarized the results of SDMs in consensus projections for each species
understanding of physiological responses to varying environmental and scenario with weighted averages based on area under the receiver operating
conditions22 . There is some evidence for life-history plasticity characteristic curve (AUC) values of single-model outputs for each species
and algorithm following ref. 31. See Supplementary Information for a detailed
in our study species: the Arctic–alpine stonefly A. compacta
description of the modelling procedure.
inhabits vastly different ecosystems in the Alps (springs) and the
high-latitude regions (springs and lake shores)23 , and the mayfly Estimating the survival of haplotypes and ESUs. We evaluated haplotype
A. inopinatus inhabits headwaters in Central Europe, larger rivers distribution and potential haplotype losses based on haplotype/basin association.
in the Tatra Mountains and small streams and lake shores in On the basis of collection localities, we assigned each haplotype to the basins
where it was sampled. Haplotype fate was determined on the basis of the projected
high-latitude regions12 . It is unknown whether these differences availability of climatically suitable areas in the associated basin. We assumed
reflect local adaptation or true plasticity. It is also unclear unlimited dispersal within a basin and no minimum area required for population
how strongly climatic niches are conserved in aquatic insects. persistence. Thus, we considered haplotype survival even if only a single grid cell
However, under an increasingly variable climate8 , genetic diversity with suitable climatic conditions was projected within a basin. This approach
will become increasingly important for population and species leads to a rather conservative estimate of haplotype losses. We assumed ESU
survival if at least one haplotype per ESU was projected to survive under the
survival. Moreover, the loss of cryptic evolutionary lineages reduces above-outlined approach.
evolutionary potential and ends ongoing diversification processes We used sixth Pfafstetter level drainage basins derived from vector streamlines
that will impact future biodiversity. and flow direction (Supplementary Fig. S5). Most of these basins (96%) extend
Estimating losses in genetic diversity is one of the key challenges below 400 m above sea level. The selected species rarely find suitable ecological
in biodiversity research when assessing the impact of GCC. Taking conditions in water bodies at altitudes below 400 m above sea level. At these
lower latitudes, O2 -saturation is likely to be the limiting factor for the study
intraspecific genetic diversity and its geographic distribution into species. Many water bodies below this altitude are also strongly affected by human
account provides improved insights into the effects of GCC on impact. This makes them unsuitable for the high habitat-quality demands of
global and regional patterns of biodiversity. Our approach of the selected species. Thus, larval dispersal between basins through low-altitude
applying SDM projections and assessments of spatial genetic connections is very unlikely.
structure is adaptable across taxa and makes use of well-developed
tools and data sets in phylogeography15 , DNA taxonomy24 and DNA Received 16 April 2011; accepted 25 July 2011; published online
barcoding16,25 . Thus, the approach can be easily adapted to other 21 August 2011
study systems with reliable phylogeographic and faunistic data sets. References
Extensions of the approach to other systems and the incorporation 1. Thomas, C. D. et al. Extinction risk from climate change. Nature 427,
of loci under climate-driven selection are the logical next steps 145–148 (2004).
that will provide more accurate predictions of GCC effects on 2. Hill, J. K., Griffiths, H. M. & Thomas, C. D. Climate change and evolutionary
genetic and cryptic diversity. Ultimately, only the fusion of SDM adaptations at species’ range margins. Annu. Rev. Entomol. 56, 143–159 (2011).
3. Sala, O. E. et al. Biodiversity—Global biodiversity scenarios for the year 2100.
or dynamic modelling approaches, phylogeography and ecological Science 287, 1770–1774 (2000).
genomics (by revealing climate-affected loci under selection), 4. Habel, J. C., Rodder, D., Schmitt, T. & Neve, G. Global warming will
combined with deeper insights into species interactions, promises affect the genetic diversity and uniqueness of Lycaena helle populations.
to unravel the mechanisms of how GCC affects species-level and Glob. Change Biol. 17, 194–205 (2011).
intraspecific biodiversity. This knowledge may soon be of crucial 5. Bickford, D. et al. Cryptic species as a window on diversity and conservation.
Trends Ecol. Evol. 22, 148–155 (2007).
importance to implement well-directed conservation strategies in a 6. Fraser, D. J. & Bernatchez, L. Adaptive evolutionary conservation:
rapidly changing world. Towards a unified concept for defining conservation units. Mol. Ecol.
10, 2741–2752 (2001).
Methods 7. Schmidt-Kloiber, A. & Hering, D. (eds) The Taxa and Autecology
Genetic data analysis and lineage delineation. We assessed neutral genetic Database for Freshwater Organisms version 4.0 (2011); available at
variation using mitochondrial cytochrome c oxidase subunit I sequence data from http://www.freshwaterecology.info.
1,778 specimens of the nine species to project GCC-caused losses of genetic diversity 8. IPCC, Climate Change 2001: The Scientific Basis (eds Houghton, J. T. et al.)
and cryptic evolutionary lineages. Alignments for each species varied between (Cambridge Univ. Press, 2001).
475 base pairs (bp) and 620 bp in length (Supplementary Table S1). UPGMA 9. Pauls, S. U., Lumbsch, H. T. & Haase, P. Phylogeography of the montane
(unweighted pair group method with arithmetic mean) trees were constructed from caddisfly Drusus discolor: Evidence for multiple refugia and periglacial survival.
haplotypes under the Jukes–Cantor model to visualize the mapped genetic diversity Mol. Ecol. 15, 2153–2169 (2006).
as haplotypes without phylogenetic inferences. We identified ESUs based on 2% 10. Schmitt, T. Molecular biogeography of Europe: Pleistocene cycles and
sequence divergence (regularly applied in DNA barcoding of aquatic insects25 ), postglacial trends. Front. Zool. 4, e11 (2007).
and on a GMYC model of species delineation26 . The GMYC model assesses the 11. Engelhardt, C. H. M., Haase, P. & Pauls, S. U. From the Western Alps across
highest likelihood of transition from intraspecific divergence to interspecific lineage Central Europe: Postglacial recolonisation of the tufa stream specialist
formation using the predicted difference in branching rate on a tree under the two Rhyacophila pubescens (Insecta, Trichoptera). Front. Zool. 8, e10 (2011).
models of lineage evolution (neutral coalescent processes occurring within species 12. Taubmann, J. et al. Modelling range shifts and assessing genetic diversity
and branching among species). The method recovers independently evolving distribution of the montane aquatic mayfly Ameletus inopinatus in Europe
lineages as putative species26 . Branch lengths were estimated with Bayesian relaxed under climate change scenarios. Conserv. Genet. 12, 503–515 (2011).
lognormal clock analysis in BEAST v. 1.5 (ref. 27) using a coalescent prior according 13. Hampe, A. R. & Petit, J. Conserving biodiversity under climate change: The
to ref. 26 (see Supplementary Information for details). Multiple GMYC thresholds rear edge matters. Ecol. Lett. 8, 461–467 (2005).
were estimated separately for Trichoptera, Plecoptera and Ephemeroptera using 14. Scriber, J. M. Impacts of climate warming on hybrid zone movement:
the ‘SPLITS’ R package28 . Geographically diffuse and biologically porous ‘species borders’. Insect Sci. 18,
121–129 (2011).
Species distribution modelling. We modelled the potential present range of the 15. Hickerson, M. J. et al. Phylogeography’s past, present, and future: 10 years after
selected species on the basis of observation data and ten environmental variables Avise, 2000. Mol. Phylogenet. Evol. 54, 291–301 (2010).
at a resolution of 2.5 arc min (approximately 5 km2 ). Nine climatic variables were 16. Hebert, P. D. N., Penton, E. H., Burns, J. M., Janzen, D. H. & Hallwachs, W.
obtained from the public WorldClim database29 : mean diurnal range, temperature Ten species in one: DNA barcoding reveals cryptic species in the neotropical
annual range, mean temperature of wettest quarter, mean temperature of driest skipper butterfly Astraptes fulgerator. Proc. Natl Acad. Sci. USA 101,
quarter, mean temperature of warmest quarter, annual precipitation, precipitation 14812–14817 (2004).
seasonality, precipitation of warmest quarter and precipitation of coldest quarter. 17. Franklin, J. Moving beyond static species distribution models in support of
In addition, we extrapolated a ‘slope’ layer from the 2.5 arc min ‘altitude’ layer of conservation biogeography. Divers. Distrib. 16, 321–330 (2010).

NATURE CLIMATE CHANGE | VOL 1 | SEPTEMBER 2011 | www.nature.com/natureclimatechange 317

© 2011 Macmillan Publishers Limited. All rights reserved
LETTERS NATURE CLIMATE CHANGE DOI: 10.1038/NCLIMATE1191

18. Hughes, J. M. Constraints on recovery: Using molecular methods to study 30. Thuiller, W., Lafourcade, B., Engler, R. & Araújo, M. B. BIOMOD—a
connectivity of aquatic biota in rivers and streams. Freshwater Biol. 52, platform for ensemble forecasting of species distributions. Ecography 32,
616–631 (2007). 369–373 (2009).
19. Hughes, R. A., Inouye, B. D., Johnson, M. T. J., Underwood, N. & Vellend, M. 31. Marmion, M., Parviainen, M., Luoto, M., Heikkinen, R. K. & Thuiller, W.
Ecological consequences of genetic diversity. Ecol. Lett. 11, 609–623 (2008). Evaluation of consensus methods in predictive species distribution modelling.
20. Lewontin, R. C. The Genetic Basis of Evolutionary Change (Columbia Univ. Divers. Distrib. 15, 59–69 (2009).
Press, 1974).
21. Davis, M. B. & Shaw, R. G. Range shifts and adaptive responses to quaternary Acknowledgements
climate change. Science 292, 673–679 (2001). We thank J. Hughes (Griffith University), A. Mulch (BiK-F), M. Pfenninger (BiK-F) and
22. Chown, S. L. et al. Adapting to climate change: A perspective from evolutionary K. Schwenk (University of Landau) for helpful comments. This study was financially
physiology. Clim. Res. 43, 3–15 (2010). supported by the research funding program LOEWE—Landes-Offensive zur Entwicklung
23. Lillehammer, A. Norwegian stoneflies. II. Distribution and relationship to the Wissenschaftlich-ökonomischer Exzellenz of Hesse’s Ministry of Higher Education,
environment. Nor. Entomol. Tidsskr. 21, 195–200 (1974). Research, and the Arts.
24. Vogler, A. P. & Monaghan, M. T. Recent advances in DNA taxonomy.
J. Zool. Syst. Evol. Res. 45, 1–10 (2007). Author contributions
25. Zhou, X., Adamowicz, S., Jacobus, L., DeWalt, R. E. & Hebert, P. D. N. M.B., C.N. and S.U.P. conceived the research. M.B., C.N. and S.U.P. conducted
Towards a comprehensive barcode library for arctic life—Ephemeroptera, the research. S.D., C.H.M.E., P.H., S.L., J.S., K.T., M.B. and S.U.P. contributed new
Plecoptera, and Trichoptera of Churchill, Manitoba, Canada. Front. Zool. 6, reagents/analytic tools. M.B. and S.D. analysed the data. M.B., C.N. and S.U.P. wrote the
e30 (2009). manuscript. C.N. and S.U.P. contributed equally to the study. All authors discussed the
26. Monaghan, M. T. et al. Accelerated species inventory on Madagascar using results and commented on the manuscript.
coalescent-based models of species delineation. Syst. Biol. 58, 298–311 (2009).
27. Drummond, A. J. & Rambaut, A. BEAST: Bayesian evolutionary analysis by Additional information
sampling trees. BMC Evol. Biol. 7, e214 (2007). The authors declare no competing financial interests. Supplementary information
28. Ezard, T., Fujisawa, T. & Barraclough, T. Species delimitation by threshold accompanies this paper on www.nature.com/natureclimatechange. Reprints and
statistics version 1. https://r-forge.r-project.org/projects/splits/ (2009). permissions information is available online at http://www.nature.com/reprints.
29. http://www.worldclim.org. Correspondence and requests for materials should be addressed to S.U.P. or C.N.

318 NATURE CLIMATE CHANGE | VOL 1 | SEPTEMBER 2011 | www.nature.com/natureclimatechange

View publication stats

© 2011 Macmillan Publishers Limited. All rights reserved