Environmental Data
To estimate the environmental conditions necessary for successful chick-rearing, we used sea
ice concentrations (SIC), sea surface temperatures (SST), chlorophyll concentrations (CHL), and
bathymetric data from multiple data sources (Fig. S1). Since these environmental parameters were
measured on different spatial scales, each parameter was interpolated to the same 25 x 25 km polar
stereographic grid surrounding the Antarctic continent
(ftp://sidads.colorado.edu/pub/DATASETS/seaice/polar-stereo/tools/). Then, we created annual
climatologies for each environmental parameter consisting of the penguins’ chick-rearing period
(December-February; Wilson et al. 2001; Croxall and Davis 1999; Kooyman, 2002). Using these annual
climatologies, we estimated the rate of change during the chick-rearing period for SIC (1978-2010), SST
(1981-2010), and CHL (1978-1986, 1997-2010) using the slope of the linear regression over time.
Significant changes in these environmental parameters over time were assessed at the 0.05 level (Fig. 2).
All analyses and statistics were carried out in R (www.r-project.org, R Development Core Team, 2004).
From 1978-2011, SIC were obtained from NSIDC’s Nimbus-7 Scanning Multichannel
Microwave Radiometer (SMMR) and the Defense Meteorological Satellite Program’s (DMSP) Special
Sensor Microwave/Imager (SSM/I; ftp://sidads.colorado.edu/pub/DATASETS/seaice/polarstereo/nasateam/final-gsfc/south/monthly/; ftp://sidads.colorado.edu/pub/DATASETS/seaice/polarstereo/nasateam/near-real-time/south/).
CHL data from 1978-1986 were acquired from the Coastal Zone Color Scanner (CZCS)
(http://oceandata.sci.gsfc.nasa.gov/CZCS/Mapped/Monthly/4km/chlor/); from 1997-2007 from Sea
Viewing Wide Field of View Sensor (SeaWiFS;
http://oceandata.sci.gsfc.nasa.gov/SeaWiFS/Mapped/Monthly/9km/chlor/); and from 2008-2011 from
the MODIS Aqua sensor (http://oceandata.sci.gsfc.nasa.gov/MODISA/Mapped/Monthly/4km/chlor/).
All statistical analyses using CHL data were computed in log10 units.
SST data from 1981-1984 was obtained from the NOAA Advanced Very High Resolution
Radiometer (AVHRR; ftp://podaacftp.jpl.nasa.gov/allData/avhrr/L3/pathfinder_v51/monthly/day/04km/) and from 1985-2009
(ftp://podaac-ftp.jpl.nasa.gov/allData/avhrr/L3/pathfinder_v5/monthly/day/04km). In addition, SST data
from 2010-2011 was obtained from NASA’s MODIS Aqua satellite
(http://oceandata.sci.gsfc.nasa.gov/MODISA/Mapped/Monthly/4km/SST/).
Bathymetry data was obtained from the International Bathymetric Chart of the Southern Ocean
(http://www.ngdc.noaa.gov/mgg/aboutmgg/oceanmapping.html).
Table S1. Mean AUC for CRH models trained on all Adélie penguin colony locations and WAP
Adélie penguin colony locations and mean percent contribution for each environmental variable
in the model.
All Adélie
WAP Adélie
.89 ± 0.001
.95 ± 0.001
14.4 ± 0.39
50.9 ± 0.39
Sea Ice Percent
Contribution
40.5 ± 0.32
7.3 ± 0.25
Bathymetry
Percent
Contribution
45.1 ± 0.39
41.8 ± 0.39
Mean AUC
SST Percent
Contribution
Figure S1. Mean (a) SIC, (b) SST and (c) CHL around Antarctica during the austral summer
(Dec-Feb).
Figure S2. (a) Sea ice percent and (b) SST training data for CRH models for continental Adélie,
WAP Adélie, Gentoo and Chinstrap penguins.
Figure S3. A rotating plot of niche spaces defined by SIC, SST and bathymetry for continental
Adélie penguins, WAP Adélie penguins, Gentoo penguins and Chinstrap penguins similar to Fig.
4.
Figure S4. Mean CRH suitability from a MaxEnt model trained (a) on all Adélie penguin colony
locations and (b) on WAP Adélie penguin colony locations. Change in the CRH suitability from
a MaxEnt model trained (c) on all Adélie penguin colony locations and (d) on WAP Adélie
penguin colony locations.
Figure S5. Mean SST (a,d), CHL (b,e), SIC (c,f), within our polar stereographic grid (top row)
and within 200 km from land (bottom row) during the chick-rearing period. Measurements were
taken from different satellites.
Figure S6. Significant changes in log10 CHL from 1997-2011 around Antarctica during the
austral summer (Dec-Feb). Black areas indicate non-significant trends at the 0.05 level.
Figure S7. Probability density of CHL concentrations from 1997-2011 for continental Adélie
penguins (CA), WAP Adélie penguins (WA), Gentoo penguins (G) and Chinstrap penguins (C).
Significant differences (p< 0.05) were computed using a non-parametric Kruskal-Wallis test and
a multiple comparison test after Kruskal-Wallis. Letters denote non-significant differences in the
mean. The thick black box indicates the interquartile range, the thinner black lines are the upper
and lower adjacent values, the white points are the median and gray x’s are the mean.
Figure S8. Mean CRH suitability for Adélie penguins from each CRH model output from 19822010. Red areas indicate suitable conditions while dark blue areas are unsuitable conditions. The
WAP, a region of rapid change, is highlighted in the bottom left corner.
Figure S9. Mean CRH suitability for Gentoo penguins from each CRH model output from 19822010. Red areas indicate suitable conditions while dark blue areas are unsuitable conditions. The
WAP, a region of rapid change, is highlighted in the bottom left corner.
Figure S10. Mean CRH suitability for Chinstrap penguins from each CRH model output from
1982-2010. Red areas indicate suitable conditions while dark blue areas are unsuitable
conditions. The WAP, a region of rapid change, is highlighted in the bottom left corner.
Figure S11. Significant changes in CRH suitability from 1982-2010 for Adélie penguins. Warm
colors indicate an increase in suitability while cool colors display regions with decreases in
suitability. Black areas indicate non-significant trends at the 0.05 level. The WAP, a region of
rapid change, is highlighted in the bottom left corner.
Figure S12. Significant changes in CRH suitability from 1982-2010 for Gentoo penguins. Warm
colors indicate an increase in suitability while cool colors display regions with decreases in
suitability. Black areas indicate non-significant trends at the 0.05 level. The WAP, a region of
rapid change, is highlighted in the bottom left corner.
Figure S13. Significant changes in CRH suitability from 1982-2010 for Chinstrap penguins.
Warm colors indicate an increase in suitability while cool colors display regions with decreases
in suitability. Black areas indicate non-significant trends at the 0.05 level. The WAP, a region of
rapid change, is highlighted in the bottom left corner.