naturejobs transferable skills for postdocs
10 July 2003 International weekly journal of science
424, 113-236 10 July 2003 www.nature.com/nature no.6945
Why a tree grows better
in New York City
Why a tree grows better
in New York City
The pressure builds
The hydrogen economy
A nuclear option?
The pressure builds
The hydrogen economy
A nuclear option?
10.7 cover US 3/7/03 2:57 pm Page 1
the coseismic mean stress change. For a homogeneous half-space we have
ð1 2 2nÞ
where m is shear modulus, and R is the euclidean distance between x and y. Therefore,
equation (1) can be written as:
2pmð1 þ n
To determine surface displacements due to complete draining of a permeable surface layer
with thickness D, overlying impermeable rocks, we integrate equation (3) from the surface
to depth D. Note that equation (3) scales with (n
Received 23 January; accepted 23 May 2003; doi:10.1038/nature01776.
1. Harris, R. A., Simpson, R. W. & Reasenberg, P. A. Inﬂuence of static stress changes on earthquake
locations in southern California. Nature 375, 221–224 (1995).
2. Dieterich, J. A constitutive law for rate of earthquake production and its application to earthquake
clustering. J. Geophys. Res. 99, 2601–2618 (1994).
3. Toda, S., Stein, R. S. & Sagiya, T. Evidence from the AD 2000 Izu islands earthquake swarm that
stressing rate governs seismicity. Nature 419, 58–61 (2002).
4. Freed, A. M. & Lin, J. Delayed triggering of the 1999 Hector Mine earthquake by viscoelastic stress
transfer. Nature 411, 180–183 (2001).
5. Nur, A. & Booker, J. R. Aftershocks caused by pore ﬂuid ﬂow? Science 175, 885–887 (1972).
6. Pollitz, F. F., Wicks, C. & Thatcher, W. Mantle ﬂow beneath a continental strike-slip fault: Post-seismic
deformation after the 1999 Hector Mine earthquake. Science 293, 1814–1818 (2001).
7. Zebker, H., Rosen, P., Goldstein, R., Gabriel, A. & Werner, C. On the derivation of coseismic
displacement ﬁelds using differential radar interferometry: The Landers earthquake. J. Geophys. Res.
99, 19617–19643 (1994).
8. Einarsson, P. Earthquakes and present-day tectonism in Iceland. Tectonophysics 189, 261–279 (1991).
9. Einarsson, P. & Eirı
ksson, J. Earthquake fractures in the Districts Land and Ranga
rvellir in the South
Iceland Seismic Zone. Jo
kull 32, 113–119 (1982).
10. Pedersen, R., Sigmundsson, F., Feigl, K. L. & A
ttir, T. Coseismic interferograms of two M
earthquakes in the South Iceland Seismic Zone, June 2000. Geophys. Res. Lett. 28, 3341–3344 (2001).
11. Pedersen, R., Jo
nsson, S., A
ttir, T., Sigmundsson, F. & Feigl, K. Fault slip distribution of two June
¼ 6.5 earthquakes in South Iceland estimated from joint inversion of InSAR and GPS
measurements. Earth Planet. Sci. Lett. (in the press).
ttir, T. et al. Crustal deformation measured by GPS in the South Iceland Seismic Zone due to
two large earthquakes in June 2000. Geophys. Res. Lett. 28, 4031–4033 (2001).
13. Einarsson, P., Bjo
rnsson, S., Foulger, G., Stefa
nsson, R. & Skaftado
ttir, Th. in Earthquake Prediction
An International Review Vol. 4 (eds Simpson, D. W. & Richards, P. G.) 141–151 (American
Geophysical Union, Washington DC, 1981).
rgmann, R. et al. Time-space variable afterslip on and deep below the Izmit earthquake rupture.
Bull. Seismol. Soc. Am. 92, 126–137 (2002).
15. Peltzer, G., Rosen, P., Rogez, F. & Hudnut, K. Poro-elastic rebound along the Landers 1992 earthquake
surface rupture. J. Geophys. Res. 103, 30131–30145 (1998).
16. Sigmundsson, F. Post-glacial rebound and asthenosphere viscosity in Iceland. Geophys. Res. Lett. 18,
17. Sigmundsson, F. & Einarsson, P. Glacio-isostatic crustal movements caused by historical volume
change of the Vatnajokull ice cap, Iceland. Geophys. Res. Lett. 19, 2123–2126 (1992).
18. Pollitz, F. F. & Sacks, I. S. Viscosity structure beneath northeast Iceland. J. Geophys. Res. 101,
19. Hofton, M. & Foulger, G. R. Postrifting anelastic deformation around the spreading plate boundary,
north Iceland: 1. Modeling of the 1987-1992 deformation ﬁeld using a visco-elastic earth structure.
J. Geophys. Res. 101, 25403–25421 (1996).
rnsson, G., Flovenz, O
. G., Saemundsson, K. & Einarsson, E. H. in Proc. 26th Workshop on
Geothermal Reservoir Engineering SGP-TR-168, 327–334 (Stanford University, Stanford, 2001).
21. Roeloffs, E. Poroelastic techniques in the study of earthquake-related hydrologic phenomena. Adv.
Geophys. 27, 135–195 (1996).
22. Clifton, A. & Einarsson, P. in Autumn Meeting 2000
Abstracts (ed Jo
nsson, S. S.) 1 (Geoscience
Society of Iceland, Reykjavı
23. Segall, P. Induced stresses due to ﬂuid extraction from axisymmetric reservoirs. Pure Appl. Geophys.
139, 535–560 (1992).
24. Rice, J. R. & Cleary, M. P. Some basic stress diffusion solutions for ﬂuid-saturated elastic porous media
with compressible constituents. Rev. Geophys. Space Sci. 14, 227–241 (1976).
25. DeMets, C., Gordon, R. G., Argus, D. F. & Stein, S. Current plate motions. Geophys. J. Int. 101, 425–478
26. Dziewonski, A. M., Ekstrom, G. & Maternovskaya, N. Centroid-moment tensor solutions for April-
June 2000. Phys. Earth Planet. Inter. 123, 1–14 (2001).
27. Chen, C. W. & Zebker, H. A. Two-dimensional phase unwrapping with use of statistical models for
cost functions in nonlinear optimization. J. Opt. Soc. Am. 18, 338–351 (2001).
28. Wang, H. F. Theory of Linear Poro-Elasticity with Applications to Geomechanics and Hydrogeology
(Princeton Univ. Press, Princeton, 2001).
Acknowledgements We thank the European Space Agency for providing the SAR data. We also
thank G. Guðmundsson and R. Stefa
nsson for providing preliminary earthquake locations from
the South Iceland Lowland (SIL) seismic network; A. Clifton and P. Einarsson for providing data
of the mapped surface ruptures; F. Sigmundsson, T. A
ttir, K. A
stsson, E. Roeloffs and
K. Feigl for discussions; and R. Bu
rgmann for comments and suggestions that improved the
Competing interests statement The authors declare that they have no competing ﬁnancial
Correspondence and requests for materials should be addressed to S.J. (email@example.com).
Urbanization effects on tree growth
in the vicinity of New York City
Plants in urban ecosystems are exposed to many pollutants and
higher temperatures, CO
and nitrogen deposition than plants in
. Although each factor has a detrimental or ben-
eﬁcial inﬂuence on plant growth
, the net effect of all factors and
the key driving variables are unknown. We grew the same
cottonwood clone in urban and rural sites and found that
urban plant biomass was double that of rural sites. Using soil
transplants, nutrient budgets, chamber experiments and mul-
tiple regression analyses, we show that soils, temperature, CO
nutrient deposition, urban air pollutants and microclimatic
variables could not account for increased growth in the city.
Rather, higher rural ozone (O
) exposures reduced growth at
rural sites. Urban precursors fuel the reactions of O
resulted in lower cumulative
exposures compared to agricultural and forested sites
throughout the northeastern USA. Our study shows the over-
riding effect of O
despite a diversity of altered environmental
factors, reveals ‘footprints’ of lower cumulative urban O
exposures amidst a background of higher regional exposures,
and shows a greater adverse effect of urban pollutant emissions
beyond the urban core.
Urbanization of the globe is accelerating, with potentially large
impacts on vegetation in cities and surrounding areas
. Urban air
contains high concentrations of many gaseous, particulate and
photochemical pollutants (such as NO
and volatile organic compounds)
; and urban soils are high in
heavy metals and can be more hydrophobic and acidic than
surrounding rural environments
. Although many of these con-
taminants have detrimental effects on plant growth, urban environ-
ments also have higher rates of nutrient and base-cation
, warmer temperatures (urban ‘heat-island’ effect)
and increased CO
factors that often, but not
invariably, enhance plant growth. Given the potential for inter-
and the relative absence of studies
examining more than two or three factors in combination, under-
standing the net effect of multiple anthropogenic environmental
changes in an urban environment and the relative importance of the
individual factors remains a major challenge.
We used an inherently fast-growing clone of Eastern cottonwood
(Populus deltoides) as a ‘phytometer’
to integrate the net growth
response to multiple anthropogenic environmental changes in New
York City compared to surrounding rural environments. Rapid
growth rates, continuous growth throughout the season, and
responsiveness to a range of climatic and pollutant variables
make this widespread riparian and early successional tree species a
suitable indicator. Soil transplants, nutrient budgets, chamber
replication of ﬁeld conditions and multiple regression approaches
were then used to determine the key driving variables. Urban and
rural site comparisons were selected from known steep pollution
across relatively short spatial scales (,100 km). Local
variation in light and precipitation was minimized by growing
plants in open ﬁelds with drip irrigation. Temperature effects on
season length were controlled by synchronizing transplant and
Department of Integrative Biology, University of California, Berkeley, California 94720, USA (T.E.D.).
letters to nature
NATURE | VOL 424 | 10 JULY 2003 | www.nature.com/nature 183
harvest dates within each growing season. The remaining micro-
climatic and pollutant differences were monitored at adjacent
climate and air quality monitoring stations (http://climod.nrcc.
cornell.edu, http://www.dec.state.ny.us, and http://www.ecostudies.
The relative importance of atmospheric (that is, pollutants and
residual microclimatic variation) versus soil effects was determined
by growing cottonwoods in soils reciprocally transplanted from
remnant urban and rural primary forest stands previously shown to
differ in pH, hydrophobicity, conductivity, heavy metals (including
Pb, Ni, S, Zn, Cu, Al and Mn) and base cation concentrations (Ca
. Potting soil with slow-release fertilizer was also used
to estimate maximum growth potential at all sites independent of
soil transplants. Given the net growth responses to site and soil
treatments, we used the nutrient budgets to assess effects of wet and
dry deposition; the chamber experiments to simulate urban and
rural thermal, CO
environments; and multiple regression
analyses to relate ﬁnal season biomass from ﬁeld experiments to the
remaining variables potentially responsible for observed growth
Contrary to expectations, cottonwoods grew twice as large amid
the high concentration of multiple pollutants in New York City
compared to rural sites (Fig. 1). Greater urban plant biomass was
found for all urban–rural site comparisons, two separate planting
dates in the ﬁrst year and two further consecutive growing seasons.
Urban–rural growth differences occurred for the faster-growing
trees in the fertilized potting soils and the slower-growing trees in
the forest soil treatments, but there was no signiﬁcant effect of soils
transplanted from urban versus rural forests (F
P ¼ 0.124). The consistently greater urban plant biomass, indepen-
dent of soil type, indicated that growth differences between urban
and rural sites were due to atmospheric rather than soil alterations.
The beneﬁcial effects of increased nutrient deposition or higher
urban temperatures and CO
concentrations were primary factors
potentially responsible for an increase in plant growth in the urban
atmosphere (Table 1). However, the twofold growth differences
between urban and rural sites occurred for the fertilized potting soil
treatments where cottonwoods had access to three to ﬁve orders of
magnitude more N, P, K
in fertilizer than from
atmospheric deposition (11.1, 4.8, 9.2, 3.2 and 0.2 g respectively in
fertilizer, compared to 28.1, 0.07, 0.31, 7.3 and 1.8 mg deposited to
urban plants, respectively). Furthermore, atmospheric nutrient
inputs were highest in year 2 when saplings grew the least (for
example 2.2, 2.7 and 2.0 kg N ha
per season via wet deposition at
study site NY
for years 1, 2 and 3 respectively), and PO
trend toward higher deposition at the rural sites (Table 1). The
fertilization effects of nutrient deposition therefore did not appear
to account for greater urban cottonwood biomass.
While temperature effects on season length were controlled
via simultaneous transplant and harvest dates, a greater increase in
C gain relative to respiratory C losses at the warmer daily tempera-
tures (þ1.8 8C mean growing period temperatures
enhanced cottonwood growth in the urban environment
the relationship between C ﬁxation and CO
Figure 1 Cottonwood growth in urban and rural sites. Final season shoot and root
biomass (mean ^ s.e., potting soils) for cottonwoods grown in urban (ﬁlled, NY
rural (open; HV
) sites in the vicinity of New York City for three consecutive growing
seasons (a–c). Site locations in Methods and Fig. 3b. Values that fall below the zero line
are for belowground biomass. F and P statistics are for linear contrasts of analyses of
variance comparing total biomass for urban versus rural sites. Independent comparisons
for above- and belowground biomass gave the same result. Bars with different letters
indicate values signiﬁcantly different using the Tukey–Kramer HSD.
Table 1 Urban and rural atmospheric pollutants near New York City.
Atmospheric gases (p.p.b.): x
annual mean (^s.e.)
18.7 (0.3) 2.3 (0.0)
NO 39.3 (3.5) 0.5 (0.06)
37.7 (0.7) 6.2 (0.25)
* 16.0 (1.5) 28.0 (0.6)
† 408 (0.2) 358 (0.4)
Suspended particulates (.10
annual mean (^s.e.)
Pb 0.09 (0.00) 0.04 (0.00)
5.47 (0.4) 0.44 (0.04)
12.4 (0.8) 4.3 (0.2)
Total 57.3 (5.2) 19.4 (2.3)
Wet deposition (mg m
third quarter total (^s.e.)
913.8 (151.2) 725.6 (65.9)
519.9 (74.4) 465.7 (31.6)
142.2 (10.0) 60.7 (6.9)
0.6 (0.01) 0.8 (0.1)
2.5 (1.4) 2.8 (0.6)
59.5 (31.9) 17.0 (4.0)
14.8 (8.8) 5.1 (1.0)
72.9 (36.3) 14.4 (2.1)
133.7 (17.3) 41.3 (4.4)
18.0 (4.9) 19.0 (1.1)
pH‡ 4.3 (0.1) 4.2 (0.1)
Atmospheric pollutant concentrations at urban and rural sites over the three years of experiments.
Urban atmospheric gases, suspended particulates and wet deposition data were monitored at the
New York State Department of Environmental Conservation’s Morrisania, Green Point and
Eisenhower Park air monitoring stations
, respectively (adjacent to sites NY
, locations in
Methods and Fig. 3b). Rural data are from HV
, with the exception of SO
, Pb and CO
available from Belleayre
(,70 km northwest of HV
(,70 km southwest of HV
. Bold values show exposures that were signiﬁcantly higher (paired t-tests, P , 0.05). Italics
indicate data available in only one year in which case statistics represent intra-annual
comparisons. Rural NO
and urban/rural CO
concentrations were collected in years sub-
sequent to ﬁeld experiments, but follow well-documented urban–rural patterns
September; † 14 h, p.p.m., August 1996
. ‡ Precipitation weighted mean.
letters to nature
NATURE | VOL 424 | 10 JULY 2003 | www.nature.com/nature184
increases most with the initial rise in CO
above ambient con-
, incrementally elevated urban CO
(þ50 p.p.m.) could also have increased growth in urban sites.
However, a series of chamber experiments simulating urban and
rural thermal and CO
environments failed to reveal individual or
combined effects of elevated urban temperatures and CO
trations on total (F
1,12Te m p
¼ 0.74, P ¼ 0.55; F
P ¼ 0.56; F
1,12Te m p þ CO2
¼ 1.55, P ¼ 0.43), above- or below-
ground biomass. The absence of a CO
effect is in agreement with
other studies on this clone at higher CO
regression analysis also failed to reveal any relationship between
ﬁnal season biomass in urban and rural ﬁeld sites and ambient
temperature regimes whether the data were summarized as the
maximum, minimum or mean of the daily temperature proﬁles or
as cumulative growing degree-days (base 15 8C; P ¼ 0.882, 0.895,
0.226 and 0.160, respectively; forward stepwise regression analysis).
The urban pollution haze can reduce maximum incoming
photosynthetically active radiation)
by up to 18%
, but light levels remained far above photosynthetic
saturation for this species (700
. High concen-
trations of condensation nuclei can also increase precipitation in
, but irrigation waters were supplied at a rate far in
excess of precipitation (6.2 cm H
) and there were no
consistent differences in precipitation between our urban and
rural sites (F
¼ 0.85, P ¼ 0.36). Whereas lower relative humid-
(210.3% mean growing period comparisons
), increased CO
concentrations and even the pollutants themselves could all have
reduced stomatal conductance in the urban sites, thereby minimiz-
ing pollutant impacts
, the offset of detrimental impacts could not
account for a relative increase in plant growth in urban compared to
Overall, then, the collective results of all experiments and
environmental comparisons provided no evidence that greater
urban cottonwood biomass was due to enhanced growth in the
urban atmosphere. Yet the same pattern could have arisen if
detrimental effects reduced growth in the country. Because nutrient
budgets, temperature and CO
regressions and microclimatic comparisons could not account for
an increase in plant growth in the city, clearly these factors also
could not account for reduced growth in the country. As expected,
most atmospheric gases, suspended particulates and wet deposition
components that could reduce plant growth were either higher in
New York City or did not differ between urban and rural sites (Table
1). However, O
was signiﬁcantly higher at rural sites, and thus
could have reduced growth in the country. Primary O
are emitted in cities, but must react in sunlight to form O
masses move to rural environments
. Ozone exposures were there-
fore consistently higher for rural sites both to the north and the east
of the city in all consecutive growing seasons (paired t-tests,
P , 0.001).
An open-top chamber experiment exposing cottonwood to
ambient and greater O
exposures representative of those at the
urban and rural sites (33 versus 59 p.p.b. growing period mean
Figure 2 Cottonwood biomass related to O
exposure. Final season cottonwood shoot
biomass (mean ^ s.e., potting soils) at urban (ﬁlled) and rural (open) ﬁeld sites versus
exposure (growing period 12-hour mean, p.p.b.; data available for NY
; ref. 16). Squares, circles and triangles represent years 1, 2 and 3,
respectively. F and P statistics show signiﬁcance of O
effects from a multiple regression
analysis with growing degree-days (base 15 8C; F
¼ 0.2, P ¼ 0.685) and urban
versus rural sites (F
¼ 0.04, P ¼ 0.866) as additional independent variables.
Variance inﬂation factors ,10 demonstrated the lack of collinearity.
Figure 3 Urban and rural O
exposures in the northeastern USA. a, Histograms of O
exposures (12-hour mean p.p.b., May–September, year 2) for all urban and rural sites in
the US EPA AIRS database for the northeastern USA (excluding Maine and Pennsylvania;
rural grey: forested; rural white: agricultural). Exposures were signiﬁcantly lower in urban
compared to rural sites (t
¼ 4.1, P , 0.001). Arrows show mean exposures. Asterisks
show exposures at sites used in this study. Results were consistent for mean and
cumulative peak O
(AOT40 or SUM06
) comparisons. b, Inverse distance weighted
interpolation of O
exposure (units as above) for all US EPA AIRS sites in the New York/New
Jersey/Connecticut region in year 2. The lower cumulative urban exposures appear as the
green-yellow ‘footprints’ up to 500 km
in size. Note that footprints of lower urban O
appear only for areas with extensive urban, suburban and rural monitoring stations. Urban
), Hudson Valley (HV
) and Long Island (LI
) site locations are also shown.
letters to nature
NATURE | VOL 424 | 10 JULY 2003 | www.nature.com/nature 185
concentrations, respectively) showed a 50% reduction in cotton-
wood biomass at the greater O
P ¼ 0.016)
an effect magnitude comparable to that found between
urban and rural ﬁeld sites. Multiple regression analysis showed that
ﬁnal season biomass was signiﬁcantly inversely related to ambient
exposures across all ﬁeld sites and years of experiments,
accounting for 93% of variation (Fig. 2). Within-season compari-
sons also showed that incremental changes in leaf area production
were inversely related to cumulative O
P ¼ 0.004; F
¼ 11.5, P , 0.001) independent of effects due
to site, soil, time through the season or growing degree-days. Less
leaf area was produced in the sites with the highest O
this pattern held, despite the switch in sites with the highest
exposures between measurement intervals.
Analysis of the US Environmental Protection Agency’s Alliance of
Information and Referral Systems (AIRS) database (http://www.airs.
org) showed that O
exposures at our rural sites were representative
of mean non-urban agricultural and forested exposures throughout
the northeastern USA (Fig. 3a). Consequently, the detrimental
effect of higher rural than urban O
exposures was not due to
extreme high exposures downwind of an urban centre
urban exposures were signiﬁcantly lower than non-urban sites
throughout the region. Spatial interpolation of the O
data for the
New York/New Jersey/Connecticut region revealed footprints con-
sisting of relatively low cumulative O
exposures in urban areas with
a background of higher regional exposures (Fig. 3b). While NO
titration reactions have been shown to reduce O
within the urban
, regional interpolations speciﬁcally omit the urban data
the footprints of lower cumulative urban O
exposures have not
previously been documented.
Of the many factors that could affect plant growth in urban
environments, soil factors, nutrient deposition, temperature, CO
urban air pollutants and microclimatic variables could not account
for the greater urban plant biomass. Rather, all evidence indicated
that the greatest effect of the multiple anthropogenic environmental
changes was the secondary reactions: these produced the higher
exposures that reduced growth in the country. Because
cottonwood is mid-range in O
sensitivity, with many species
showing greater responses to ambient O
growth in response to higher rural O
exposures is unlikely to be
restricted to this cottonwood clone. These results do not negate the
known detrimental effects of multiple urban pollutants, but show
the greater effect of secondary reactions that create higher cumu-
exposures beyond the urban core. Although individual 1-
hour peak concentrations are typically higher in urban centres
data indicate that the higher cumulative exposures at rural sites had
the greatest impact.
Our research thus determines the relative importance of multiple
anthropogenic environmental changes under current ﬁeld con-
ditions, and reveals a number of counter-intuitive results. (1)
There was greater plant growth amid multiple pollutants in urban
compared to rural environments. (2) Higher urban temperatures,
concentrations and N deposition could not account for
increased growth in the city. (3) Ozone was the single overriding
factor accounting for observed growth differences among multiple
anthropogenic environmental changes. (4) The most detrimental
effects of multiple urban pollutant emissions occurred in rural
environments, with (5) footprints of reduced impact of lower
cumulative urban O
exposures on a background of higher regional
These ﬁndings are in contrast to the extensive monitoring,
warning and effects research within city centres, suggestions that
ecological differences between urban and rural environments could
be due to higher urban than rural O
, and the pervasive
perception that rural environments are safe havens from urban
pollutant emissions. As such, our work highlights the need to
reconsider relative pollutant impacts in urban and rural environ-
ments as air sheds merge throughout the globe. Although extensive
global change research has studied potential impacts of tempera-
and N deposition, this study shows overriding O
amid these other factors. A
Cottonwood growth in New York City was compared to a northern rural site in the
Hudson Valley (HV) and eastern rural sites on Long Island (LI) for three consecutive
growing seasons (1992–1994). Rooted cuttings (Clone ST109; initial height 10–25 cm;
5–12 leaves) were transplanted to the ﬁeld on 4–7 July, drip irrigated (3.8 litres day
four intervals at 06:00, 10:00, 14:00 and 18:00 hours) and harvested before bud set and leaf
senescence on 13–15 September. A second planting was also made on 6 August in year 1.
Final season biomass comparisons were supplemented with measurements of total leaf
area (area ¼ 3.772 2 1.611 £ length þ 0.745 £ length
¼ 0.976, P , 0.001)
measured biweekly on all plants in year 1 and tri-weekly on three plants per soil treatment
in year 2. Urban and rural sites (locations shown in Fig. 3b) were located at The New York
Botanical Garden, Bronx (NY
); Hunts Point Water Works, Bronx (NY
); Con Edison Fuel
Depot, Austoria (NY
); Eisenhower Park, Hempsted (NY
); The Institute of Ecosystem
Studies, Millbrook (HV
); Cornell University Horticultural Research Laboratory,
); and Brookhaven National Laboratory, Upton (LI
). Herbivory was
negligible (,1% total leaf area) and did not vary between urban/rural site and soil
treatments (ANOVA, P . 0.5).
Soils were collected from the organic and top soil horizons of remnant oak-dominated
urban and rural forest stands and transplanted to urban and rural sites. Soils were
transplanted from one urban and one rural forest to all sites in year 1 (ﬁve plants per soil
origin), and from two urban and two rural forests from each of the HVand LI comparisons
to all respective HVand LI sites in year 2 (four forest soils per site, ten plants per soil origin,
eight forest soils in total). All soils were ﬁne sandy loam: Charlton and Hollis series for the
urban–rural HV comparisons and Montauk series for LI comparisons. Soils were sieved
(1-cm mesh), mixed thoroughly, placed in 19-litre pots, transported to each site, and
buried on 1-m centres. Holes were dug 50% below pot depth and backﬁlled with gravel
and sand for drainage. HV soils were collected from Van Cortland and Pelham Bay Parks,
Bronx, and Housetonic and Mohawk State Forests, northwestern Connecticut. LI soils
were collected from Cunningham and Alley Pond Parks, Queens, The David Weld
Preserve, Nissequogue, and Edward Stevenson’s woods, Mt Sinai. The potting soil
treatment, also used at all sites and all years of experiments (ten plants per site; 15 plants
per site for the second planting of year 1), was perlite:topsoil:peat (1:2:1 v/v), with
limestone (10.2 g per pot), 5N-10P-5K fertilizer (13.6 g per pot), phosphate (20%, 10.2 g
per pot) and slow-release fertilizer (Osmocote 14N-14P-14K, Scotts-Sierra Co.; 113.4 g per
pot). Site, soil and site £ soil were the main effects in analyses for the ﬁrst planting of year 1
and these factors were nested within HV and LI comparisons in year 2. Collinearity
between time and growing degree-days for intra-annual foliar comparisons forced
removal of the non-signiﬁcant temperature effect from the regression model
¼ 0.00, P ¼ 0.988; F
¼ 1.4, P ¼ 0.232).
N, P, K
and base-cation deposition were calculated from NH
concentrations in precipitation
with dry deposition assumed to equal
that in rainfall
. Nutrient comparisons in otherwise clear and odourless irrigation waters
showed that As
2þ,3 þ ,4 þ
were all below detection limits. The remaining detectable constituents (NH
, alkalinity, hardness, total solids,
pH, anion-cation balance, conductivity and turbidity) were not different between urban
and rural sites (t-tests, P . 0.05), and were not related with phytometer biomass (forward
stepwise regression analysis, P . 0.25).
Field conditions were simulated as closely as possible by transplanting cottonwoods into
the same 19-litre pots with fertilized potting soils and providing full sun and an ample
water supply (biweekly irrigation to ﬁeld capacity for open-top chambers and the same
drip irrigation for chamber experiments). The open-top chamber O
performed at the Boyce Thompson Institute Field Facility in Ithaca, New York, from 7 July
to 21 September 1996, in conjunction with an experiment that included four chambers per
treatment but showed no signiﬁcant chamber effects (N ¼ 5 plants per chamber)
Temperature and CO
growth chamber experiments (Conviron PGW36, Controlled
Environments, Inc.) simulated the temperatures and CO
concentrations of the NY
sites in year 2. Since cottonwoods showed twofold growth differences between sites in
the ﬁrst three weeks of the ﬁeld experiments
, chamber conditions were set to the average
conditions of the ﬁrst 21 days and experiments lasted three weeks. Control conditions
were: min. 18.8 8C/ max. 30.3 8C temperatures (ramped linearly between 6:00 and 15:00),
and 350 ^ 3 p.p.m. CO
concentrations. Treatments were asymmetrically elevated
temperatures (min. 21.1 8C/ max. 31.9 8C) and elevated CO
(400 ^ 3 p.p.m.). Days were set to 14.5 hours, relative humidity was 50%, and light was
maintained above 900
with dawn and dusk simulated by ramping
chamber lights in 25% increments at 15-min intervals. Each experiment was repeated
twice, switching the treatments between chambers (N ¼ 7 plants per chamber) with
treatment effects tested against the treatment by replicate interaction
. Further analyses
using individual plants as the experimental unit conﬁrmed the absence of temperature and
letters to nature
NATURE | VOL 424 | 10 JULY 2003 | www.nature.com/nature186
Received 21 February; accepted 28 April 2003; doi:10.1038/nature01728.
1. Lovett, G. M. et al. Atmospheric deposition to oak forests along an urban-rural gradient. Environ. Sci.
Technol. 34, 4294–4300 (2000).
2. Pouyat, R. V., McDonnell, M. J. & Pickett, S. T. A. Soil characteristics of oak stands along an urban-
rural land-use gradient. J. Environ. Qual. 24, 516–526 (1995).
3. Peterson, J. T. The Climate Of Cities: A Survey Of The Recent Literature. NAPCA Pub. No. AP-59 (US
Department of Health, Education and Welfare, 1969).
4. Idso, C. D., Idso, S. B. & Balling, R. C. Jr An intensive two-week study of an urban CO
Phoenix, Arizona, USA. Atmos. Environ. 35, 995–1000 (2001).
5. Gatz, D. F. Urban precipitation chemistry: A review and synthesis. Atmos. Environ. B Urban Atmos. 25,
6. Freedman, B. Environmental Ecology: The Ecological Effects Of Pollution, Disturbance, And Other
Stresses (Academic, San Diego, 1995).
7. US National Research Council. Committee on Tropospheric Ozone Formation and Measurement.
Rethinking The Ozone Problem In Urban And Regional Air Pollution (National Academy Press,
Washington DC, 1992).
8. Pickett, S. T. A. et al. Urban ecological systems: Linking terrestrial ecology, physical, and
socioeconomic components of metropolitan areas. Annu. Rev. Ecol. Syst. 32, 127–157 (2001).
9. Mooney, H. A., Winner, W. E., Pell, E. J. & Chu, E. (eds) Response Of Plants To Multiple Stresses
(Academic, San Diego, 1991).
10. Weinstein, L. H., Laurence, J. A., Mandl, R. H. & Walti, K. Use of Native And Cultivated Plants As
Bioindicators Of Pollution Damage: Plants For Toxic Assessment (eds Wang, W., Gorsuch, J. W. & Lower,
W. R.) 117–126 (Am. Soc. Testing and Materials, Philadelphia, 1990).
11. USDA Forest Service. Cottonwood: Populus deltoides: hybrid poplar. (Govt document no. A 13.42/
24:C 82, Government Printing Ofﬁce, Washington DC (1980).
12. Wait, D. A., Jones, C. G. & Schaedle, M. Controlling growth and chemical composition of Populus
deltoides saplings by iteratively matching nutrient supply to demand: a bootstrap fertilization
technique. Tree Phys. 16, 359–365 (1996).
13. Wait, D. A., Jones, C. G., Wynn, J. & Woodward, F. I. The fraction of expanding to expanded leaves
determines the biomass response of Populus deltoides to elevated CO
. Oecologia 121, 193–200 (1999).
14. Coleman, J. S. & Jones, C. G. Plant stress and insect performance: cottonwood, ozone and a leaf beetle.
Oecologia 76, 57–61 (1988).
15. Magnussen, S. Growth, transpiration, unit leaf rate, and water use efﬁciency of six poplar clones at
different light and temperature regimes. Forest Tree Improvement 0, 1–47 (1985).
16. New York State Department of Environmental Conservation, Division of Air Resources. NY State Air
Quality Report: Ambient Air Monitoring System. Report no. DAR-95-1 (NYS-DEC, Albany, New York,
17. Gregg, J. W. The Effect Of Urban Compared To Rural Environments On Plant Growth In The Vicinity Of
New York City: The Net Response And Relative Importance Of The Different Pollutants. Thesis, Cornell
18. Jones, H. G. Plants and microclimate: A quantitative approach to environmental plant physiology.
(Cambridge Univ. Press, 1996).
19. Jacovides, C. P., Timbios, F., Asimakopouolos, D. N. & Steven, M. D. Urban aerosol and clear sky spectra
for global and diffuse photosynthetically active radiation. Agric. Forest Meteorol. 87, 91–104 (1997).
20. Hesstvedt, E., Isaksen, I. S. A. & Hov, O. Ozone generation over rural areas. Environ. Sci. Technol. 12,
21. Lefohn, A. S., Knudsen, H. P., Logan, J. A., Simpson, J. & Bhumralkar, C. An evaluation of the kriging
method to predict 7-h seasonal mean ozone concentrations for estimating crop loss. J. Air Pollut.
Control Assoc. 37, 595–602 (1987).
22. USDA Forest Service FIA Ozone Biomonitoring Program. Field Methods For Phase 3 Measurements.
(USDA Forest Service, Amherst, Massachusetts, 2002).
23. Federal Land Managers’ Air Quality Related Values Workgroup (FLAG). Phase I Report. (US Forest
Service, National Park Service, and Fish and Wildlife Service, Fort Collins and Boulder, Colorado,
24. Appleton, B. & Koci, J. Trees For Problem Landscape Sites – Air Pollution. Virginia cooperative
extension Pub. no. 430-022 (Virginia Cooperative Extension, Blacksburg, Virginia, 2000).
25. Goldman, M. B., Groffman, P. M., Pouyat, R. V., McDonnell, M. J. & Pickett, S. T. A. CH
N availability in forest soils along an urban-rural gradient. Soil Biol. Biogeochem. 27, 281–286 (1995).
26. Lovett, G. M. & Lindberg, S. E. Atmospheric deposition and canopy interactions of nitrogen in forests.
Can. J. Forest Res. 23, 1603–1616 (1993).
27. Yun, S. C. & Laurence, J. A. The response of clones of Populus tremuloides differing in sensitivity to
ozone in the ﬁeld. New Phytol. 141, 411–421 (1999).
28. Potvin, C. & Tardif, S. Sources of variability and experimental designs in growth chambers. Funct. Ecol.
2, 123–130 (1988).
29. Institute of Ecosystem Studies. Environmental Monitoring Program (Institute of Ecosystem Studies,
Millbrook, NY, 2002).
Acknowledgements M. McDonnell, R. Pouyat, S. Pickett, A. Greller, G. Lovett, M. Geber and
P. Marks provided discussions at the outset of this research. C. Andersen, J. Compton, A. Hudak,
J. Laurence, H. Lee, D. Phillips, P. Rygiewicz, A. Solomon and D. Tingey provided discussions and
editorial feedback. H. Lee provided EPA O
data and statistical consultation. P. Dickerson created
the inverse-distance weighted O
map. M. Topa oversaw the open-top chamber experiment.
Organizations listed in the Methods provided site access, forest soils and technical assistance.
Financial support was provided to J.W.G. by the Edna Bailey Sussman Fund for Environmental
Internships, the New York State Heritage Foundation, a Cornell University Mellon Research
Grant, the Institute of Ecosystem Studies, Cornell’s Department of Ecology and Systematics, a
Mellon Foundation graduate training grant (to T.E.D.), the Cornell Center for the Environment,
Sigma Xi and a US EPA post doctoral fellowship.
Competing interests statement The authors declare that they have no competing ﬁnancial
Correspondence and requests for materials should be addressed to J.W.G. (firstname.lastname@example.org).
Strong population substructure is
correlated with morphology and
ecology in a migratory bat
Cassandra M. Miller-Butterworth*†‡, David S. Jacobs* & Eric H. Harley†
* Department of Zoology, University of Cape Town, Private Bag, Rondebosch,
7701, South Africa
† Wildlife Genetics Unit, Division of Chemical Pathology, University of Cape
Town, Observatory, 7925, South Africa
Examining patterns of inter-population genetic diversity can
provide valuable information about both historical and current
evolutionary processes affecting a species. Population genetic
studies of ﬂying and migratory species such as bats and birds
have traditionally shown minimal population substructure,
characterized by high levels of gene ﬂow between populations
In general, strongly substructured mammalian populations
either are separated by non-traversable barriers or belong to
terrestrial species with low dispersal abilities
. Species with
female philopatry (the tendency to remain in or consistently
return to the natal territory) might show strong substructure
when examined with maternally inherited mitochondrial DNA,
but this substructure generally disappears when biparentally
inherited markers are used, owing to male-mediated gene
. Male-biased dispersal is considered typical for mammals
and philopatry in both sexes is rare. Here we show strong
population substructure in a migratory bat species, and philo-
patry in both sexes, as indicated by concordance of nuclear and
mtDNA ﬁndings. Furthermore, the genetic structure correlates
with local biomes and differentiation in wing morphology. There
is therefore a close correlation of genetic and morphological
differentiation in sympatric subspeciﬁc populations of this
Schreibers’ long-ﬁngered bat, Miniopterus schreibersii natalensis
(Chiroptera, Verspertilionidae), migrates seasonally between win-
tering roosts (hibernacula) and summer maternity colonies in
South Africa. Ringing studies
indicate that ﬁdelity to both roost
types is well developed. Miniopterus schreibersii natalensis were
sampled (Fig. 1, Supplementary Table S1) from four South African
maternity roosts (Die Hel (DHL), Jozini Dam (JD), Peppercorn
(PC) and Sudwala (SW)), one hibernaculum (Steenkampskraal
(SKK)), four roosts that are occupied all year round but are used
primarily as summer roosts (De Hoop (DHP), Koegelbeen (KB),
Grahamstown (G) and Maitland Mines (MM)) and one transient
pre-maternity roost (Shongweni Dam (SHD)). An analysis of six
dinucleotide microsatellite loci indicated that the M. s. natalensis
population was genetically substructured into three major sub-
populations, occurring in the south (DHL and DHP), west (SKK
and KB) and northeast (G, MM, SHD, JD, PC and SW) regions of
the country (Fig. 2, Supplementary Table S2). Colonies within each
subpopulation were genetically similar and thus poorly differen-
tiated: r values ranged between 20.005 and 0.068 (P . 0.05 for all
comparisons). Genetic distances between these colonies were
also low: (
ranged between 0.084 and 0.446. However, colonies
from different subpopulations were strongly differentiated, both
when examined individually through pairwise comparisons (range
of r values 0.152–0.686, P , 0.01 for all comparisons; range of (
values 1.033–5.286) and when pooled into the three subpopulations
(range of r values 0.351–0.623, P # 0.0001 for all comparisons;
range of (
values 2.037–4.550). Each colony had sufﬁciently
‡ Present address: Laboratory of Genomic Diversity, National Cancer Institute, PO Box B, Frederick,
Maryland 21702, USA.
letters to nature
NATURE | VOL 424 | 10 JULY 2003 | www.nature.com/nature 187