Download SUPPORTING INFORMATION Divergent and narrower climatic

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
SUPPORTING INFORMATION
Divergent and narrower climatic niches characterize polyploid species of
European primroses in Primula sect. Aleuritia
Spyros Theodoridis, Christophe Randin, Olivier Broennimann, Theofania Patsiou and
Elena Conti
Journal of Biogeography
Appendix S1 Detailed description of the methods and results.
MATERIALS AND METHODS
Species distribution and occurrence data
In this study we used four species of Primula sect. Aleuritia (Primula farinosa, 2n =
2x = 18; Primula halleri, 2n = 4x = 36; Primula scotica, 2n = 6x = 54; Primula
scandinavica, 2n = 8x = 72) that are roughly distributed along a loosely defined
latitudinal and elevational gradient in Europe.
Primula farinosa occurs in the Pyrenees (Cantabrian range: Spain); the entire Alpine
range; the Jura mountains (France, Germany); the Vitosha, Pirin, Rila, Rhodope
ranges (Bulgaria); the Carpathian range (with a few populations in Romania); the
Tatra mountains (Slovakia); and also in England, Denmark (nearly extinct), Sweden
to 64o N, the Finnish Archipelago, and in the Baltic states to 60o N (Tutin et al., 1972;
Richards, 2003; S. Theodoridis, pers. obs.). Primula halleri occurs mainly in the
eastern Alps and Tatra mountains; parts of the Romanian and Russian Carpathians;
several mountain systems in the former Yugoslavia and in Albania; and in the Rila
and Pirin ranges in Bulgaria. Primula scotica is endemic to northern Scotland and the
Orkney Islands and P. scandinavica is endemic in Scandinavia, occurring in the
mountain ranges of southern Norway away from the coast, in both coastal and inland
areas of northern Norway and in a few sites in Sweden at the Norwegian border
(Tutin et al., 1972; Richards, 2003; S. Theodoridis, pers. obs.).
Niche variation and quantification in environmental space
To quantify niche overlap/divergence between the species and the niche breadth of
each species in multidimensional environmental space we used a method recently
developed by Broennimann et al. (2012). It calculates the density of occurrences and
environmental factors along the environmental axes of a multivariate analysis, such as
a principal components analysis (PCA), and measures niche overlap and niche
breadth along the gradients of this multivariate analysis. This approach was shown to
be an important complement to ecological niche models (ENM) for niche comparison
(Petitpierre et al. 2012). Here, we consider the first two axes of a PCA calibrated on
the entire environmental space of the study area, which was defined by spatially
restricting the extent of the biomes occupied by the four species to the European
region. The reduced environmental space formed by these two axes was divided into a
grid of 100 × 100 cells, where each cell corresponds to a unique vector of
environmental conditions present at one or more sites in geographical space.
Niche breadth of each species was assessed by randomly sampling 100 pixels
in the niche of each species (i.e. pixels were sampled according to the density of the
species occurrences), extracting their scores along the two PCA axes, and calculating
the standard deviation of the scores along the two PCA axes. This procedure was
repeated 1000 times and the distribution of the breadth values was visualized using
boxplots for each PCA axis. Niche overlap between each species pair was expressed
by Schoener’s D similarity index (Schoener, 1970; Broennimann et al., 2012). All
analyses were performed in R 2.14.1 (R Development Core Team, 2011).
Testing for niche similarity
Niche divergence between species might either be the result of an effective niche
differentiation between two species, meaning that the species occupy different
habitats, or simply reflect differences in the spatial autocorrelation of the climatic
variables between regions. To check for the effect of spatial autocorrelation, we
applied niche similarity tests in environmental (Broennimann et al., 2012) and
geographical space (Warren et al., 2008, 2010). These tests require the definition of a
background area reflected both in environmental and geographical space. This area
should ideally include suitable habitats for the species and the way it is delimited
might influence the analysis (Warren et al., 2008, McCormack et al., 2010).
To test for the robustness of our results under different ways of delimiting the
background area, we followed two different approaches. In the first approach, we
used backgrounds defined by each species’ ENM set to a base-line threshold that
maximizes the sum of sensitivity and specificity of the test data (Liu et al., 2005;
Hernandez et al., 2006). We then combined each species’ predicted background and
used a common background for the four Aleuritia species. In the second approach, we
defined species backgrounds by applying 20-km buffer zones around the occurrence
points of each species, given the lack of any obvious adaptation for long-distance
dispersal in Primula seeds (B. Keller et al., Institute of Systematic Botany, Zurich,
unpublished data). Overlaying of grids, visualization of the models and background
delimitation were conducted in ARCGIS 9.3 (ESRI, Redlands, CA).
The similarity test in environmental space examines whether the observed
niche overlap between two species with different available environmental
backgrounds is different from the overlap between the observed niche of one species
and niches selected at random from the available environmental background of the
other species in the environmental space defined by PCA axes (Broennimann et al.,
2012). For this test, the entire observed density of occurrences of one species is
randomly shifted within its defined environmental background and the overlap of the
new simulated niche with the observed niche of the other species is calculated. The
distribution of the simulated overlap values (null distribution) was generated with 100
repetitions and the process was repeated for both species being compared. Observed
overlap values greater or lower than the 95% of the null distribution indicate that
ecological niches of the species under comparison are more similar or more different
respectively than expected by random. All tests were performed in R.
The similarity test in geographical space compares the observed niche overlap
between two species to a distribution of 100 simulated overlap values generated by
comparing the ENM of one species to an ENM created from random points drawn
from the geographical background area of the other species (Warren et al., 2008). The
number of random points was equal to the number of actual occurrences of the
species from whose range the points were drawn. This process was repeated for both
species under comparison, thus two distributions per analysis were generated. This
test is also a two-tailed test, as the one conducted in environmental space. All tests
were performed in ENMTOOLS (Warren et al., 2010) and in Perl environment.
RESULTS
ENMs
Models inferred from occurrence records accurately predict the observed distribution
of all four species (Fig. 4). Additionally, high area under the receiver operating
characteristic curve (AUC) values (> 0.9 for all models) and low test omission rates
indicate excellent model performance for the four species. The predicted distribution
of P. farinosa is much broader than those of the three other species and it also
includes regions where this species is known to occur, but no georeferenced
occurrence records are available (e.g. Baltic states). ENM for P. halleri predicts all
the main mountain ranges of central-eastern Europe (Alps, Carpathians), except the
Bulgarian ranges, as suitable areas, but also predicts Pyrenees and Caucasus as
potential areas of distribution with low probability of occurrence. However,
occurrences of P. halleri in the latter regions have not been reported in previous
taxonomical studies. The narrow, known distribution of P. scotica is reflected in its
predicted potential distribution, which is confined to a few sites in Great Britain and
Iceland; no occurrences are known from the latter island. ENM predictions for P.
scandinavica include the entire region where this species occurs, but also parts of the
main European mountain ranges (i.e. Alps, Carpathians, Bulgarian ranges, Pyrenees
and Caucasus) and also a few parts of Northern Europe (e.g. England and Denmark).
Niche similarity tests
Observed niche overlap values in the multivariate niche space consistently fall within
the 95% confidence limits of the null distributions under all three different
backgrounds and all comparisons, except one (Table 1, see also Appendix S3). In all
cases but one, these results suggest that the ecological differences between species are
not more or less similar than expected due to the climatic differences between the
areas in which they occur. However, this does not imply that there are no differences
in the climatic preferences of the species. It rather suggests that these differences
reflect the environmental heterogeneity between the habitats available (i.e.
background) to the species and they are not necessarily the result of habitat selection.
The only case where the observed overlap is greater than the null distribution is when
we compared the observed overlap between P. halleri and P. scotica to the null
distribution produced by randomly shifting the niche of P. scotica within its available
background delimited by ENM predictions. However, the comparison in the opposite
direction did not produce significant deviation from the null distribution (Table 1).
Results of background similarity tests in geographical space are more complex,
but significant niche differentiation is detected in 23 of 30 comparisons (observed
overlap lower than 95% of the null distribution; Table 2). Analysis using ENMs
showed that all three pairwise comparisons P. farinosa–P. scotica, P. farinosa–P.
scandinavica and P. scotica–P. scandinavica support niche divergence with respect to
both null distributions. Most comparisons between P. halleri and P. scotica also
support greater divergence than that expected from the environmental differences of
their background area, with the exception of the comparison between P. scotica with
P. halleri background (Table 1). Most of the comparisons (four out of six) between P.
farinosa and P. halleri show that the niches that the two species occupy are less
divergent than expected.
REFERENCES
Broennimann, O., Fitzpatrick, M.C., Pearman, P.B., Petitpierre, B., Pellissier, L.,
Yoccoz, N.G., Thuiller, W., Fortin, M.J., Randin, C., Zimmermann, N.E.,
Graham, C.H. & Guisan, A. (2012) Measuring ecological niche overlap from
occurrence and spatial environmental data. Global Ecology and Biogeography,
21, 508–512.
Hernandez, P.A., Graham, C.H., Master, L.L. & Albert, D.L. (2006) The effect of
sample size and species characteristics on performance of different species
distribution modeling methods. Ecography, 29, 773–785.
Liu, C., Berry, P.M., Dawson, T.P. & Pearson, R.G. (2005) Selecting thresholds of
occurrence in the prediction of species distributions. Ecography, 3, 385–393.
McCormack, J.E., Zellmer, A.J. & Knowles, L.L. (2010) Does niche divergence
accompany allopatric divergence in Aphelocoma jays as predicted under
ecological speciation? Insights from tests with niche models. Evolution, 64,
1231–1244.
Petitpierre, B., Kueffer, C., Broennimann, O., Randin, C., Daehler, C. & Guisan, A.
(2012) Climatic niche shifts are rare among terrestrial plant invaders. Science,
335, 1344–1348.
R Development Core Team (2011) R: a language and environment for statistical
computing, version R 2.14.1. R Foundation for Statistical Computing, Vienna,
Austria. Available at: http://www.r-project.org/.
Richards, J. (2003) Primula. BT Batsford Ltd, London.
Schoener, T. (1970) Nonsynchronous spatial overlap of lizards in patchy habitats.
Ecology, 51, 408–418.
Tutin, T.G., Heywood, V.H., Burges, N.A., Valentine, D.H., Walters, S.M. & Webb,
D. A. (1972) Flora Europea, Vol. 3. Cambridge University Press, Cambridge.
Warren, D.L., Glor, R.E. & Turelli, M. (2008) Environmental niche equivalency
versus conservatism: quantitative approaches to niche evolution. Evolution, 62,
2868–2883.
Warren, D.L., Glor, R.E. & Turelli, M. (2010) ENMTools: a toolbox for comparative
studies of environmental niche models. Ecography, 33, 607–611.
Related documents