ArticlePDF Available

A Landscape Ecology Approach Identifies Important Drivers of Urban Biodiversity

  • University of Zurich and Agroscope

Abstract and Figures

Cities are growing rapidly worldwide, yet a mechanistic understanding of the impact of urbanization on biodiversity is lacking. We assessed the impact of urbanization on arthropod diversity (species richness and evenness) and abundance in a study of six cities and nearby intensively managed agricultural areas. Within the urban ecosystem, we disentangled the relative importance of two key landscape factors affecting biodiversity, namely the amount of vegetated area and patch isolation. To do so, we a priori selected sites that independently varied in the amount of vegetated area in the surrounding landscape at the 500-m scale and patch isolation at the 100-m scale, and we hold local patch characteristics constant. As indicator groups, we used bugs, beetles, leafhoppers, and spiders. Compared to intensively managed agricultural ecosystems, urban ecosystems supported a higher abundance of most indicator groups, a higher number of bug species, and a lower evenness of bug and beetle species. Within cities, a high amount of vegetated area increased species richness and abundance of most arthropod groups, whereas evenness showed no clear pattern. Patch isolation played only a limited role in urban ecosystems, which contrasts findings from agro-ecological studies. Our results show that urban areas can harbor a similar arthropod diversity and abundance compared to intensively managed agricultural ecosystems. Further, negative consequences of urbanization on arthropod diversity can be mitigated by providing sufficient vegetated space in the urban area, while patch connectivity is less important in an urban context. This highlights the need for applying a landscape ecological approach to understand the mechanisms shaping urban biodiversity and underlines the potential of appropriate urban planning for mitigating biodiversity loss.
Content may be subject to copyright.
A landscape ecology approach identifies important
drivers of urban biodiversity
Department of Community Ecology, Institute of Ecology and Evolution, University of Bern, Baltzerstrasse 6, 3012, Bern,
Cities are growing rapidly worldwide, yet a mechanistic understanding of the impact of urbanization on biodiversity
is lacking. We assessed the impact of urbanization on arthropod diversity (species richness and evenness) and abun-
dance in a study of six cities and nearby intensively managed agricultural areas. Within the urban ecosystem, we dis-
entangled the relative importance of two key landscape factors affecting biodiversity, namely the amount of
vegetated area and patch isolation. To do so, we a priori selected sites that independently varied in the amount of veg-
etated area in the surrounding landscape at the 500-m scale and patch isolation at the 100-m scale, and we hold local
patch characteristics constant. As indicator groups, we used bugs, beetles, leafhoppers, and spiders. Compared to
intensively managed agricultural ecosystems, urban ecosystems supported a higher abundance of most indicator
groups, a higher number of bug species, and a lower evenness of bug and beetle species. Within cities, a high amount
of vegetated area increased species richness and abundance of most arthropod groups, whereas evenness showed no
clear pattern. Patch isolation played only a limited role in urban ecosystems, which contrasts findings from agro-
ecological studies. Our results show that urban areas can harbor a similar arthropod diversity and abundance com-
pared to intensively managed agricultural ecosystems. Further, negative consequences of urbanization on arthropod
diversity can be mitigated by providing sufficient vegetated space in the urban area, while patch connectivity is less
important in an urban context. This highlights the need for applying a landscape ecological approach to understand
the mechanisms shaping urban biodiversity and underlines the potential of appropriate urban planning for mitigat-
ing biodiversity loss.
Keywords: agricultural intensification, arthropods, biodiversity, habitat amount, isolation, landscape ecology, NDVI,
Received 15 September 2014 and accepted 31 October 2014
Over the last few decades, cities have grown rapidly in
size and density. This trend will continue as urban
areas are expected to absorb most of the global popula-
tion growth in the upcoming decades (United Nations,
2012), thereby tripling in area (Seto et al., 2012).
Although the conversion of natural habitat into built-
up area has become one of the major reasons for habitat
destruction worldwide, the effects of urbanization on
biodiversity remain ambiguous, and underlying pro-
cesses are poorly understood (reviewed in Faeth et al.,
2011; McDonnell & Hahs, 2008; McKinney, 2008). How-
ever, most studies found that species diversity
decreases due to urbanization, whereas species abun-
dance increases, probably due to species specialized on
urbanized areas, the so-called urban exploiters (Chace
& Walsh, 2006; McKinney, 2008; Møller, 2009; Faeth
et al., 2011; Pickett et al., 2011; Aronson et al., 2014).
One reason for the inconsistent pattern of effects of
urbanization on biodiversity might be that urbaniza-
tion occurs in various ecosystems and land-use types.
For example, in relatively pristine areas (e.g., forests),
the impact of urbanization might be different than in
intensively managed agricultural land. In addition,
most of our understanding of how urbanization
affects biodiversity is derived from studies along sin-
gle ruralurban gradients (e.g., Rickman & Connor,
2003; Cane et al., 2006; Hartley et al., 2007; Croci et al.,
2008; Ahrne et al., 2009; Bates et al., 2011; Bennett &
Gratton, 2012), which are not replicated at the ecosys-
tem scale and hence lack generality (but see Hedblom
om, 2010; Magura et al., 2010; Møller et al.,
2012). Most studies also did not take into account that
along ruralurban gradients not only the ecosystem,
but also local habitat characteristics change systemati-
cally. For example, that there are usually more iso-
lated and managed habitat patches toward the city
center (McKinney, 2008; Faeth et al., 2011). Conse-
quently, in these studies, changes of the ecosystem
are confounded with changes in local habitat charac-
Correspondence: Eva Knop, tel. +41 31 631 4539, fax +41 31 631
4888, e-mail:
1652 ©2015 John Wiley & Sons Ltd
Global Change Biology (2015) 21, 1652–1667, doi: 10.1111/gcb.12825
teristics. Here, we investigated the impact of urbani-
zation on biodiversity by assessing arboreal arthropod
diversity and abundance on standardized habitat
patches in six cities and six intensively managed agri-
cultural areas.
A major mechanism of how urbanization affects nat-
ural communities is the conversion of vegetated area to
built-up surface (McKinney, 2008; Goddard et al., 2010;
Pickett et al., 2011). As a consequence, primary produc-
tivity declines from rural to urban areas, except for
resource-limited regions (e.g., arid regions) where the
pattern might be the opposite (e.g., Milesi et al., 2003;
Imhoff et al., 2004; Yu et al., 2009; Lu et al., 2010). Due
to the positive relationship between primary productiv-
ity and species diversity (e.g., Waide et al., 1999; Mittel-
bach et al., 2001), biodiversity is expected to be reduced
in urbanized compared to rural areas (e.g., Lee et al.,
2004). In addition, other parameters change due to
urbanization and are expected to affect biodiversity,
such as the quality and the management of the vege-
tated area, or several physical parameters, such as tem-
perature (‘heat island effect’), air pollution, or soil
compaction (McKinney, 2008; Pickett et al., 2011). In
this study, we focused on the role of the amount of veg-
etated area as a major difference between urban and
rural ecosystems and held other parameters as much as
possible constant. In Switzerland, the predominant
share of rural land is agriculturally used, and approxi-
mately 88% of the habitat converted to built-up area
(0.69 m
) is former farmland (Bundesamt f
Statistik, 2014). As 89% of the Swiss farmland is inten-
sively managed (Bundesamt f
ur Statistik, 2014), we
focused on intensively managed farmland as a rural
reference to the urban ecosystem.
Within cities, the vegetated area is diverse and char-
acterized by small and isolated patches dispersed
between the built-up area. Therefore, for understanding
processes underlying biodiversity patterns, a landscape
perspective is especially pertinent within cities (Breuste
et al., 2008, Goddard et al., 2010). Nonetheless, princi-
ples of landscape ecology have rarely been applied in
an urban context (Goddard et al., 2010). Several studies
have investigated the impact of landscape variables on
urban biodiversity in a correlative way (e.g., Rickman
& Connor, 2003; Sadler et al., 2006;
Ockinger et al., 2009;
Lizee et al., 2012; Soga & Koike, 2012; Johnson et al.,
2013; Shwartz et al., 2013), that is, they assessed land-
scape variables (e.g., habitat amount and patch isola-
tion), but did not vary them independently. However,
as predicted by island biogeography and metapopula-
tion theory (MacArthur & Wilson, 1967; Hanski, 1994),
they generally found that species diversity increases
with patch size and connectivity.
To advance our understanding of the impact of
urbanization occurring in an intensively managed agri-
cultural region on biodiversity per se and to disentangle
the impact of different landscape variables on arboreal
arthropod diversity within cities, we asked the follow-
ing questions: (1) Does the urban ecosystem support a
lower arthropod diversity and abundance compared to
the intensively managed agro-ecosystem? (2) Does a
high amount of vegetated area increase arthropod
diversity and abundance within cities? (3) Does patch
isolation reduce local diversity and abundance of arbo-
real arthropods in urban areas? In contrast to previous
studies, we applied (1) a replication scheme across dif-
ferent cities (2) standardized focal patch size, type, and
management, and (3) a priori selected landscape sectors
that independently varied in the amount of vegetated
area and patch connectivity. Arthropod diversity was
investigated as the number of species and the evenness
in distribution of species’ abundances of four indicator
groups (bugs, beetles, leafhoppers, and spiders). The
number of species considers the presence of different
species, whereas species evenness reveals whether the
community is evenly assembled by different species or
dominated by just a few very abundant species (e.g.,
urban exploiters) (Purvis & Hector, 2000; Magurran,
2004). We expected that arthropod diversity and abun-
dance should be reduced in cities compared to rural
areas. Within cities, we expected that arthropod diver-
sity and abundance should be higher in landscape sec-
tors with a high amount of vegetated area compared to
landscape sectors with a low amount of vegetated area.
Further, arthropod diversity and abundance should
decrease with increasing patch isolation.
Materials and methods
Study area
The study was conducted in six cities of Switzerland,
namely Zurich, Basel, Geneva, Bern, Locarno, and Chur, and
in intensively managed agricultural areas adjacent to each of
the six cities (for geographic location of sites within Switzer-
land see the electronic supplementary material, Fig. S1a).
Zurich, Basel, and Geneva are the largest cities of Switzer-
land, and Bern, Locarno, and Chur are of intermediate size
(the number of inhabitants varies between 373 000 and
34 000 across the studied cities, for details, see the Table S1).
Switzerland has a temperate, middle-European climate;
mean annual temperatures in the studied regions range from
8.8 °C (Bern) to 12.4 °C (Locarno), and mean annual precipi-
tation ranges from 842 mm (Basel) to 1897 mm (Locarno)
(Bundesamt f
ur Meteorologie und Klimatologie Meteo-
Schweiz, 2014). At the locations of all cities, humans had set-
tlements already at Roman times, and all cities have
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
experienced a rapid growth over the last 100 years (Bunde-
samt f
ur Statistik, 2014).
Focal patches
To investigate the biodiversity of urban and agricultural
habitat patches, we focused on four arthropod indicator
groups inhabiting silver birch (Betula pendula) trees. Opting
for trees as focal patches allowed us to investigate the bio-
diversity effects that arise from differences in landscape fac-
tors rather than from differences in local management
practices or patch size, as we only included unmanaged
trees of similar size (mean SD height: 12.09 1.2 m). A
second reason for this choice was that scattered trees are
ubiquitous elements in cities, yet in an urban context they
have received little attention so far, as urban ecologists
have mainly studied herbaceous habitats (grasslands/
brownfields/wastelands, e.g., Hartley et al., 2007; Strauss &
Biedermann, 2006), gardens (e.g., Smith et al., 2006), and
woodlands (e.g., Sadler et al., 2006). Therefore, in this study,
we focused on trees that were not part of larger arrange-
ments of woody habitat or forest. We opted for birch trees
because birches (1) are known to harbor a high number of
arthropod species, especially also specialist herbivores
andle & Brandl, 2001), and (2) often colonize disturbed
areas (Atkinson, 1992), which is why they are common in
rural as well as in urban ecosystems.
To assure spatial independency of the studied trees, overlap
of the landscapes surrounding the trees in a 500 m radius was
allowed only when other criteria for site selection could other-
wise not be met (mean overlap with other 500 m buf-
fers SD: 5.1 9.4%, for details, see Table S2). Likewise,
distance to forest was maximized (mean distance to edges of
forest patches larger than 100 000 m
SD: 905.7 612.7 m,
calculation based on data provided by the Swiss Federal Office
of Topography swisstopo).
Study design
To investigate the differences at the ecosystem level (urban
versus agricultural), we sampled arthropods on a total of
sixty-three silver birch trees located in the six Swiss cities and
nearby intensively managed agricultural areas: In each of the
three larger cities, we selected twelve trees, and in each of the
remaining three cities, we selected three trees, summing up to
a total of forty-five urban trees studied. We further selected
three birch trees in each of the six intensively managed agri-
cultural areas adjacent to the cities (for distances between
urban and rural study sites see Table S1).
To assess the mechanisms shaping arthropod diversity and
abundance within urban areas, we did an in-depth study in
which we focused on the three larger cities (Zurich, Basel, and
Geneva). For investigating the importance of the amount of
vegetated area, we a priori selected landscape sectors that
provided either a high or a low amount of vegetated area
(see below). The focal trees were then selected in a way that
in every city six trees were surrounded by vegetation-rich
landscape sectors, and six trees were surrounded by
vegetation-poor landscape sectors within a 500-m radius. We
chose a landscape scale of 500 m because in agro-ecological
studies, this spatial scale had been proven to be relevant for
bugs (Torma & Cs
ar, 2013), beetles (Bat
ary et al., 2007),
other flying insects (Chaplin-Kramer et al., 2011), and spiders
(Clough et al., 2005). To disentangle the impact of patch con-
nectivity from the impact of the amount of vegetated area,
the trees were additionally either connected or isolated at the
100-m scale. We chose a 100-m scale because studies in agri-
cultural areas had found that 100 m isolation has a strong
effect on species diversity (e.g., Knop et al., 2011; Sch
et al., 2011). We were able to select trees that varied in their
100 m patch isolation independently of the larger-scale
(500 m) landscape context (see Fig. S1b) by defining isolation
of the focal tree as isolation from other birch trees and restrict-
ing our subsequent analysis of patch isolation effects to herbi-
vore birch specialist species, that is species for which trees
other than birches do not represent suitable habitat.
Amount of vegetated area in the landscape
The amount of vegetated area in the surrounding landscape of
all sixty-three studied trees was calculated within a 500-m buf-
fer and estimated based on the normalized difference vegeta-
tion index (NDVI) for the time period of JuneAugust 2011
using MODIS data (NASA Land Processes Distributed Active
Archive Center LP DAAC). The NDVI (for values see Table
S3) was used in two ways: While it entered the analysis com-
paring urban and rural ecosystems as a continuous variable,
in the within-city analysis, it was used to a priori classify land-
scape sectors (see above; vegetation rich: NDVI >0.5,
mean sd: 0.6 0.04, vegetation poor: NDVI <0.4,
mean SD: 0.3 0.06). The NDVI has been proven to be a
very useful tool for assessing primary productivity at larger
spatial scales (e.g., Milesi et al., 2003; Imhoff et al., 2004; Yu
et al., 2009; Lu et al., 2010) and relate it to species diversity pat-
terns (e.g., Pettorelli et al., 2011). Although we exclusively
sampled species that were present on trees, most nonspecialist
species included in this study also use other vegetation ele-
ments, for example for feeding or overwintering, which makes
an all vegetation index apposite.
Isolation metrics
Patch isolation was determined using two area-informed met-
rics because such metrics (1) had performed better than those
based exclusively on nearest neighbor distance in a previous
modeling study (Bender et al., 2003) and (2) were better suited
for capturing the actual study situation in which connectivity
was not established by proximity to a single large neighboring
patch, but proximity to several relatively small neighboring
patches (birch trees within a buffer of 100 m). First, we
selected focal birch trees in a way that, independently of the
amount of vegetated area within 500-m buffers, they were sur-
rounded by either a high (7, mean sd: 18.3 12.23) or a
low (6, mean SD: 2.6 1.73) number of other birch trees
within 100 m, and we hereafter refer to this two-level metric
as ‘isolation factor’.
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
Second, we calculated a connectivity index based on the for-
mula of Hanski CI
(Hanski, 1994). This
index combines the size of neighboring patches and their dis-
tance to the focal patch and performed best in a comparison of
three different connectivity metrics (Br
uckmann et al., 2010).
Hanski’s index was originally intended for single species with
bscaling for dispersal distance. As we studied a community of
species with mostly unknown dispersal distances, we set bto
1. The term Arepresents the size of neighboring habitat
patches. We visually assigned neighboring birches within 100-
m buffers to a scale ranging from one (very small) to five (very
large), which then entered the formula as A=0.5, 1, 10, 50,
and 100 m
, respectively. Distance d
was not used as an expo-
nential term to avoid giving too strong weighting to variances
in distance. This resulted in the formula CI
. All
landscape analyses were performed in ARCGIS (ArcMap, ver-
sion 10, ESRI, Redlands, CA, USA).
Nuisance variables
Given that urbanization changes abiotic conditions with major
effects on temperature (‘urban heat island’) and biogeochemi-
cal cycles (e.g., Pickett et al., 2011), we assessed the tempera-
ture at the study sites and the nitrogen content of the studied
trees. Temperature was recorded every four hours between
July 15, 2011 and August 15, 2011 with thermochron DS1922 i-
buttons (Dallas Semiconductor Corporation, TX, USA), and
the measurements were subsequently averaged. Between July
31 and August 22, we randomly clipped ten leaves per tree
and stored them at 20 °C. They were later dried for two days
at 70 °C and pulverized. The carbon and nitrogen contents
were measured following the standard protocol of the ana-
lyzer provider (Euro EA analyzer, Hekatech, Wegberg, Ger-
many), and CN ratios were calculated.
We further recorded the radius of the tree crown and the
density of the tree leaves to account for possible variation in
focal tree characteristics. To obtain the tree crown radius, the
distance from the trunk to the tip of the most projecting
branch was measured in every cardinal direction and then the
four measurements were averaged. Tree crown radius was
correlated with both the tree trunk diameter at breast height
(Pearson correlation coefficient PCC =0.72) and the tree
height (PCC =0.55). To estimate leaf density, we used a white
canvas 4 m in length and 1.7 m in height, which was stretched
perpendicular to the ground between two bars. One bar was
lengthwise attached to the trunk of the tree (i.e., parallel to the
trunk of the tree) and the other one was erected 4 m away,
with the upper border of the canvas being 4 m and the lower
border being 2.3 m aboveground. Using a photograph of the
canvas, we then calculated a continuous variable that ranged
from zero (whole canvas visible) to one (canvas fully cov-
ered by leaves) in Adobe Photoshop (CS4 Extended, version
Arthropod groups studied and sample collection
In a multitaxon approach, we studied Heteroptera (true bugs,
hereafter ‘bugs’), Auchenorrhyncha (plant- and leafhoppers;
as samples contained only one planthopper species, we here-
after refer to this group as ‘leafhoppers’), Coleoptera (beetles)
and Araneae (spiders). Except for spiders, these groups were
chosen because they include a relatively high number of birch
specialists, which were needed for analyzing the patch isola-
tion effects at 100 m. From a literature review, Br
andle &
Brandl (2001) concluded that in Germany 106 Coleoptera, 8
Heteroptera, and 20 Auchenorrhyncha species are associated
with birches. Spiders were included in the study because they
represent an exclusively predatory group. Experts identified
bugs, beetles, and leafhoppers to species level and spiders to
genus level. Hereafter, we use the term ‘taxa’ when referring
to the species and genus together.
In 2011, we visited each site four times. Sampling periods
were as follows: June 14 to July 11, July 31 to August 22, Sep-
tember 2 to 21, and November 19 to 27 (hereafter ‘sampling
period one’ to ‘sampling period four’). We held the order in
which we visited the different geographic regions constant
across sampling periods, but randomized the order of sites
within each region to avoid any time-of-the-day bias. During
sampling periods one to three, we used a suction sampler (Sti-
hl SH85, Waiblingen, Germany) to collect arthropods on days
with no rain and no heavy wind. Suction sampling has been
shown to be an effective method for catching our focal arthro-
pod groups in grasslands (Brook et al., 2008). We adapted the
method for sampling trees by extending the inflexible tube of
the sampler with a flexible tube 2.9 m in length and 0.1 m in
diameter, to which we attached a telescopic pole. This allowed
us to reach the branches up to 4.5 m above the ground. On
each tree, we took 120 suctions of different branches of leaves,
starting each sampling at another cardinal point and then
moving around the tree in clockwise direction.
To enlarge spider samples, we installed spider hides during
the first sampling period. They consisted of a roll of corru-
gated cardboard inserted into a plastic drosophila vial that
was attached to the tree trunk upside down approximately
2 m aboveground in a north-facing direction. Spider hides
were emptied during sampling periods two, three, and four.
All samples were stored in a freezer for later identification.
Data analysis
For every arthropod group, data from all sampling periods
were pooled. Two measures of diversity, namely species rich-
ness (number of taxa) and Pielou’s evenness (Oksanen et al.,
2013; online manual), and abundance (number of individuals)
were analyzed as response variables. Analyses were carried
out in R (R Core Team, 2012), using the packages ‘vegan’
(Oksanen et al., 2013) and ‘lme4’ (Bates et al., 2014).
To analyze the differences between agricultural and urban
ecosystems, we ran general linear mixed-effect models using
data from all sixty-three trees (further referred to as ‘ecosys-
tem analyses’) and geographic region as random intercept (six
levels: Zurich, Basel, Geneva, Bern, Locarno, and Chur). We
first fitted models with the variables we had a priori hypothe-
sized to influence our response variables, namely ecosystem
(two levels: urban versus agricultural) as a fixed factor, and
NDVI as well as the interaction ecosystem 9NDVI as other
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
explanatory variables (further referred to as ‘variables of inter-
est’). Effects of temperature, the CN ratio of the tree leaves,
the tree crown radius, and the tree leaf density were not the
focus of this study, but these variables were recorded to con-
trol for their potential impact on the response variables. They
are therefore further referred to as ‘nuisance variables’. To
avoid overfitting of the models, the nuisance variables were
only included into the full model in case they lowered the
model’s Akaike information criterion (AIC) by >2.
To investigate the factors driving arthropod diversity and
abundance within cities, we ran linear mixed-effect models
using the data from the thirty-six urban trees located in the
three cities of Zurich, Basel, and Geneva (further referred to as
‘within-city analyses’). Our random intercept was thus geo-
graphic region with three levels. Again, we first fitted models
with our variables of interest, namely the amount of vegetated
area (two levels: vegetation rich versus vegetation poor at the
scale of 500 m) and connectivity (two levels: connected versus
isolated at the scale of 100 m) as fixed factors, and CI
above) as a continuous explanatory variable. We further
included the interactions vegetated area 9connectivity and
vegetated area 9CI
. For spiders, we fitted models only with
the amount of vegetated area as we did not expect that the
connectivity to other birch trees would influence spider
assemblages. Regarding the nuisance variables, we proceeded
like described above for the ecosystem analyses.
To assess the impact of patch connectivity at the 100-m
scale, the within-city analyses were repeated using only the
data of the birch specialist species of bugs, leafhoppers, and
beetles, and analyzing only species richness and abundance
as response variables. As the number of birch specialist
beetle species was too low to analyze the species richness,
we fitted a generalized linear mixed-effect model with the
presence/absence data of specialists other than the overall
very abundant (see below) birch specialist species Trichapion
simile (Kirby). We assumed a model binomial distribution
and used the same fixed factors and random intercept as
described above, as well as the connectivity index CI
, but
excluded the interactions due to the lack of convergence of
the model including all parameters and interactions. To
check for overdispersion, we included an observation-level
random factor (as many levels as observations) into the full
model and tested it against the model without this factor
(Elman & Hill, 2009). This test was nonsignificant; therefore,
we used the model without the observation-level random
factor for model selection.
All general linear mixed-effect models were visually
checked for normal distribution of residuals and homoscedas-
ticity, and, if necessary, the dependent variable was trans-
formed to fulfill model assumptions (see Tables 2, 3, and 4 for
specifications). In all analyses, the minimal adequate model
was searched by AIC-based stepwise deletion of predictors
and interactions of the full model (AIC threshold =2). Main
effects were tested in models excluding the interactions they
were involved in. Given that model simplification approaches
have been criticized for increasing the risk of overestimating
effect sizes, we followed the recommendation of Forstmeier &
Schielzeth (2011) and additionally tested full models against
respective null models.
As the response of arthropod diversity and abundance to
temperature might be different in the rural compared to the
urban ecosystem, we ran a post hoc analysis after having
found the final model, testing whether there was a significant
interaction between temperature and ecosystem. To do so, we
compared the final model with and without temperature and
the interaction temperature 9ecosystem with likelihood-ratio
tests. The results of the post hoc analysis are presented in
Table S4.
We collected a total of 41 835 specimens belonging to
264 taxa (Table 1). Of these 264 taxa, fifteen were birch
specialist species, amounting to 35 742 individuals
(Table 1). The bug samples were dominated by the
very abundant birch specialist species Kleidocerys
resedae (Panzer), which constituted 89% of all bug
specimens collected (98% of bug birch specialist speci-
mens). The most abundant beetle species was the birch
specialist Trichapion simile, which constituted 60% of
all beetle specimens (98% of beetle birch specialist
Table 1 Numbers of individuals and taxa (genus of spiders, species of the remaining groups) sampled on sixty-three trees in
urban and intensively managed agricultural ecosystems. ‘bss.’ only herbivore birch specialists, ‘taxa’ number of species for bugs,
beetles, and leafhoppers, and number of genera for spiders. N.b.: Given the unequal number of study sites in urban and rural eco-
systems, these numbers do not directly reflect differences between ecosystems
Urban Rural Total
Individuals Taxa Individuals Taxa Individuals Taxa
All bss All bss All bss All bss All bss All bss
Bugs 27 605 25 321 50 4 6389 5597 36 4 33 994 30 918 55 4
Beetles 1751 1057 72 3 662 409 52 2 2413 1466 101 3
Leafhoppers 2844 2438 57 8 1014 920 35 8 3858 3358 67 8
Spiders 1217 39 353 27 1570 41
Total 33 417 28 816 218 15 8418 6926 150 14 41 835 35 742 264 15
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
Table 2 Results of likelihood-ratio tests comparing mixed-effect models with and without respective predictors at the ecosystem level (urban versus agricultural) for all arthro-
pods sampled on sixty-three trees in six Swiss cities and nearby intensively managed agricultural ecosystems. DAIC values always represent the calculation AIC
(more parsimonious
(less parsimonious model)
, that is positive values indicate that the respective variable lowered the model AIC. Nuisance variables were only included into the full model
if they lowered the model AIC by >2. Variables of interest were tested with AIC-based backwards model simplification
Richness (log) Abundance (log) Evenness
Bugs v. o. int. Ecosystem 12.32 0.000 10.32 0.50 0.14 8.09 0.004 6.09 1.36 0.46 5.06 0.024 3.06 0.20 0.09
NDVI 5.00 0.025 3.00 0.93 0.40 7.67 0.006 5.67 3.88 1.35 5.81 0.016 3.81 0.63 0.26
Ecos. 9NDVI 1.31 0.252 0.69 0.10 0.747 1.90 2.36 0.124 0.36
nuisance v. Temperature 2.52 0.113 0.52 0.75 0.386 1.25 3.57 0.059 1.57
CN ratio 0.01 0.903 1.99 0.65 0.418 1.35 0.11 0.742 1.89
Tree radius 0.01 0.937 1.99 0.46 0.498 1.54 1.17 0.280 0.83
L. dens. 4.53 0.033 2.53 3.27 0.070 1.27 0.04 0.850 1.96
L. dens. (fi.m.) 4.91 0.027 2.91 0.62 0.27 / / / / / /
fu.m. vs null m. 16.47 0.002 8.47 9.25 0.026 3.25 8.71 0.033 2.71
Richness (log) Abundance (log) Evenness (^2)
Beetles v. o. int. Ecosystem 0.11 0.744 1.89 6.43 0.011 4.43 0.84 0.32 8.87 0.003 6.87 0.26 0.08
NDVI 2.81 0.094 0.81 10.73 0.001 8.73 3.25 0.94 17.76 0.000 15.76 1.12 0.25
Ecos. 9NDVI 0.00 0.945 2.00 0.26 0.607 1.74 0.01 0.938 1.99
Nuisance v. Temperature 0.72 0.395 1.28 0.32 0.570 1.68 2.74 0.098 0.74
CN ratio 0.03 0.868 1.97 0.42 0.515 1.58 0.24 0.624 1.76
Tree radius 2.27 0.132 0.27 0.37 0.544 1.63 0.42 0.515 1.58
L. dens. 4.09 0.043 2.09 7.69 0.006 5.69 4.29 0.038 2.29
L. dens. (fi.m.) 6.62 0.010 4.62 0.74 0.28 7.49 0.006 5.49 1.80 0.63 4.32 0.038 2.32 0.36 0.17
fu.m. vs null m. 9.54 0.049 1.54 20.00 0.000 12.00 23.49 0.000 15.49
Richness (log) Abundance (log) Evenness
Leafhop pers v. o. int. Ecosystem 0.04 0.849 1.96 4.10 0.043 2.10 0.72 0.35 3.05 0.081 1.05 0.20 0.09
NDVI 3.25 0.071 1.25 5.39 0.020 3.39 2.43 1.02 7.82 0.005 5.82 0.63 0.26
Ecos. 9NDVI 0.25 0.617 1.75 0.51 0.475 1.49 4.51 0.034 2.51 1.61 1.04
nuisance v. Temperature 3.15 0.076 1.15 3.80 0.051 1.80 0.72 0.395 1.28
CN ratio 0.89 0.345 1.11 0.01 0.935 1.99 0.00 1.000 2.00
Tree radius 0.05 0.819 1.95 0.00 0.964 2.00 0.01 0.940 1.99
L. dens. 1.15 0.283 0.85 2.90 0.088 0.90 1.84 0.175 0.16
fu.m. vs null m. 3.54 0.316 2.46 6.23 0.101 0.23 15.39 0.002 9.39
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
Table 2 (continued)
Richness Abundance (log) Evenness
Richness Abundance (log) Evenness
Spiders v. o. int. Ecosystem 0.38 0.536 1.62 6.62 0.010 4.62 1.36 0.46 0.01 0.918 1.99
NDVI 1.75 0.186 0.25 4.37 0.036 2.37 3.88 1.35 0.01 0.924 1.99
Ecos. 9NDVI 1.54 0.214 0.46 0.32 0.570 1.68 0.02 0.902 1.98
nuisance v. Temperature 0.35 0.555 1.65 0.16 0.685 1.84 2.04 0.154 0.04
CN ratio 5.50 0.019 3.50 0.17 0.682 1.83 0.48 0.490 1.52
CN ratio (fi.m.) 11.98 0.001 9.98 0.21 0.16 / / / / / /
Tree radius 0.17 0.682 1.83 0.47 0.494 1.53 0.01 0.927 1.99
Leaf density 0.79 0.373 1.21 0.68 0.409 1.32 0.24 0.621 1.76
fu.m. vs null m. 15.66 0.004 7.66 7.04 0.071 1.04 0.03 0.998 5.97
Transformations of the response variables are indicated in brackets. Bold letters indicate significant tests (P<0.05). ‘v. o. int’., variables of interest; ‘nuisance v.’, nuisance vari-
ables; ‘ecos.’, ecosystem; ‘NDVI’, Normalized Difference Vegetation Index; ‘l. dens.’, leaf density; ‘fi. m.’, final model; ‘fu. m.’, full model; ‘null m.’, null model; LRT, likelihood-
ratio test statistic (Chi); ‘Est.’, estimate of the respective variable in the final model, estimate standard error. Means SE of the analyzed response variables are shown in Fig. 1a
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
specimens). The most abundant leafhopper species
was the birch specialist Kybos austriacus (Wagner), con-
stituting 60% of all leafhopper specimens (69% of leaf-
hopper birch specialist specimens). The most
abundant spider genus was Philodromus (36% of spider
specimens), followed by Clubiona (22%). For the mean
temperatures at study sites and the CN ratios of tree
leaves, see Table S3.
Ecosystem level
We present the results of the ecosystem analysis in
Table 2. Compared to intensively managed agricultural
ecosystems, urban ecosystems supported a significantly
higher species richness of bugs (Table 2, Fig. 1a), and a
significantly higher abundance of bugs, beetles, and
spiders, while the abundance of leafhoppers was
reduced in urban areas (Table 2, Fig. 1b). Evenness was
significantly lower in urban compared to intensively
managed agricultural ecosystems in bugs and beetles,
and there was also a statistical trend in the same direc-
tion for leafhoppers (Table 2, Fig. 1c). Apart from the
ecosystem effect, the amount of vegetated area within
500 m (NDVI) was positively associated with species
richness of bugs and abundance of bugs, beetles, and
spiders, and it was negatively associated with abun-
dance of leafhoppers (Table 2). Thus, on the ecosystem
level, we only observed an effect of the amount of vege-
tated area at the 500-m scale when the ecosystem effect
per se was controlled for. The interaction ecosys-
tem 9NDVI was significant for leafhopper evenness
(Table 2), with evenness being positively associated
with the NDVI in urban ecosystems (estimate SE =
0.89 1.011), but negatively correlated with the NDVI
in agricultural ecosystems (estimate SE =
1.41 0.98).
Including the nuisance variable improved the model
fit in only a few cases. The density of the tree leaves
positively covaried with the species richness of bugs
and the species richness and abundance of beetles, and
negatively covaried with the evenness of beetles
(Table 2). A high CN ratio was positively correlated
with spider genus richness, that is, nitrogen-rich sites
hosted a smaller number of spider genera than nitro-
gen-poor sites (Table 2).
Within-city level
Results of the within-city analyses are presented in
Table 3 (all species) and Table 4 (birch specialist herbi-
vores). Trees in landscapes with a high amount of vege-
tated area at the 500-m scale (NDVI >0.5) supported a
significantly higher species richness of bugs, leafhop-
pers, and spiders compared to landscapes with a low
amount of vegetated area (NDVI <0.4, Table 3,
Fig. 1d). Species richness of beetles was also enhanced
in landscapes supporting a high amount of vegetated
area, but only on birch trees that were poorly connected
to other birches (significant interaction 500 m vegetated
area amount 9CI
, Table 3). Abundance of bugs, bee-
tles, and spiders was also higher in landscape sectors
with a high versus low amount of vegetated area, while
for the abundance of leafhoppers, there was a trend in
the opposite direction (Table 3, Fig. 1e). Evenness of
bugs and beetles was significantly reduced in vegeta-
tion-rich landscapes compared to vegetation-poor land-
scapes, whereas the evenness of spiders did not differ
between landscape sectors with a high versus low
amount of vegetated area, and the evenness of leafhop-
pers was enhanced in vegetation-rich landscape sectors
(Table 3, Fig. 1f).
The richness of bug birch specialist species increased
with the Hanski connectivity index CI
(Table 4,
Fig. 1g), and the abundance of bug birch specialist spe-
cies was significantly higher on connected than on iso-
lated trees when connectivity was based on the number
of birches within a 100-m radius (connectivity factor
with two levels: connected versus isolated, Table 4,
Fig. 1h). Apart from these two relationships, there were
no significant associations between species richness
and abundance of birch specialist herbivores and the
connectivity measures (Table 4, Fig. 1h).
Again, the nuisance variables improved the model fit
in only a few cases and were rarely included in final
models. Higher temperature was significantly associ-
ated with higher abundance of bugs, both in the analy-
sis including the whole data set (Table 3) and when
only birch specialist bugs were included (Table 4), and
it was negatively associated with bug evenness
(Table 3). Furthermore, the radius of the tree crown
was positively correlated with the species richness of
beetles (Table 3), and the density of tree leaves was
positively related to the richness of leafhopper birch
specialist species (Table 4).
Urban ecosystems hosted a higher species richness of
bugs compared to intensively managed agricultural
ecosystems and had a similar species richness of the
other three indicator groups. These results suggest that
cities do not generally harbor a lower arboreal arthro-
pod diversity than intensively managed agro-ecosys-
tems. The generally poor status of biodiversity in the
latter might be the reason why we did not find a signifi-
cantly higher arboreal arthropod diversity in rural
areas compared to urban areas. The use of pesticides,
disturbance from farm machinery, or clear-cuts after
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
(g) (h)
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
harvesting are well known to have reduced biodiversity
over the past fifty years in the agro-ecosystem (Kleijn
et al., 2009; Bommarco et al., 2013), and these practices
are either absent in cities or occur rather in a patchy
than in an areawide pattern. Most likely, if we had cho-
sen more extensively managed agricultural areas as ref-
erence sites, rural arthropod diversity could have
exceeded urban diversity. This idea is further sup-
ported by the results that the expected positive associa-
tion between primary productivity (NDVI) and
arthropod species richness as well as abundance was
observed, but only when differences between the urban
and rural environments per se were controlled for. This
suggests that the expected positive relationship
between primary productivity and species diversity
might be disrupted by other factors, such as human
land use. Thus, the negative effect of intensive agricul-
tural land use on biodiversity may exceed the effect of
urbanization on biodiversity.
Alternatively, our results might be explained by
the fact that animals in urban areas could react dif-
ferently to novel disturbances as compared to con-
specifics in the rural ecosystem (Miranda et al., 2013)
or that species occurring in cities are preselected for
a high tolerance against noise (e.g., Paton et al.,
2012). Given that our sampling method was noisy
and unfamiliar to the arthropods on the trees, it
could have been that relatively fewer species escaped
our suction sampling in the urban system. However,
not much is known about adaptation or preselection
of urban arthropods (but see San Martin y Gomez &
Van Dyck, 2012), and therefore, this point would
require further investigation.
In three of four indicator groups, abundance was
significantly higher and evenness was significantly
lower in urban than in intensively managed agricul-
tural areas. This reflects that urbanization favors
specific species, probably those with a high tolerance to
urbanization and a high efficiency in exploiting urban
resources (e.g., Møller, 2009). An alternative explana-
tion for higher abundance of arboreal arthropods in
cities might be a weaker top-down control of herbivores
by predators in cities, as it has been suggested in a pre-
vious review (Raupp et al., 2010). However, our data
do not suggest that this is the case because we found
also a higher abundance of spiders in cities. We rather
believe that some of the species found in this study
were very efficient urban exploiters.
Within cities, net primary productivity was a strong
predictor of arthropod diversity and abundance. Land-
scape sectors with a high amount of vegetated area sup-
ported a higher abundance of bugs, beetles, and
spiders and a higher diversity of all indicator groups
(although for beetles this was only true in the case of
trees that were poorly connected to other birch trees at
the 100 m scale). However, evenness of bugs and bee-
tles was reduced in landscape sectors with a high
amount of vegetated area, indicating that not all species
can exploit urban vegetated areas equally successfully.
Interestingly, vegetation-rich urban landscapes even
supported higher total species numbers than sites in
intensively managed agricultural areas for bugs, leaf-
hoppers, and spiders. This underlines the importance
of vegetated areas in urban environments and indicates
that given appropriate urban planning, arthropod
diversity in cities has the potential to exceed the diver-
sity in intensively managed rural areas. Our findings
are consistent with the prediction that a higher amount
of vegetated area within the urban landscape should
increase species richness and abundance (Goddard
et al., 2010). They also confirm results from other eco-
systems where species diversity was positively corre-
lated with primary productivity measured as NDVI
(e.g., Bailey et al., 2004; Lassau & Hochuli, 2008; Pau
et al., 2012; Seto et al., 2004; but see Shochat et al., 2004).
As our results demonstrate the ability of urban vege-
tated areas to enhance arthropod diversity within cities,
they challenge a recent study which concluded that
urban planners should primarily focus on compacting
city development to minimize urban sprawl (Sushinsky
et al., 2013). The apparent trade-off between building
densely and providing sufficient urban vegetated areas
shows the need for developing sustainable urban plan-
ning solutions.
While the amount of vegetated area within 500 m
had a strong effect on arthropod diversity, patch con-
nectivity within 100 m played a limited role in the
urban context. Patch connectivity increased only spe-
cies richness and abundance of bug birch specialists
which reached high abundances due to the birch catkin
bug K. resedae. By contrast, birch specialists of beetles
and leafhoppers were unaffected by patch connectivity.
Fig. 1 (a) Species richness (number of species), (b) abundance (number of individuals), and (c) evenness of bugs, beetles, leafhoppers,
and spiders in urban (‘urb’) versus intensively managed agricultural (‘agri’) ecosystems, (d) species richness, (e) abundance, and (f)
evenness of arthropods in urban landscape sectors with a high (NDVI >0.5) versus low (NDVI <0.4) amount of vegetated area within
a 500-m radius, (g) species richness of bug birch specialists in relation to the modified Hanski connectivity index CIj within a 100-m
radius, and (h) abundance of insect herbivore birch specialists (‘birch special.’) on urban birches that were isolated (6 neighboring
birches) versus connected (7 neighboring birches) to other birches within 100 m. Means SE are shown. P<0.1; *P<0.05;
**P<0.01; ***P<0.001.
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
Table 3 Results of likelihood-ratio tests comparing mixed-effect models with and without respective predictors at the within-city level (urban landscape sectors with a high
versus low amount of vegetated area within 500 m) for all arthropods sampled on thirty-six trees in three Swiss cities. DAIC values always represent the calculation AIC
(more par-
simonious model)
(less parsimonious model)
, that is positive values indicate that the respective variable lowered the model AIC. Nuisance variables were only included if they low-
ered the model AIC by >2. Variables of interest were tested with AIC-based backwards model simplification. Transformations of the response variables are indicated in brackets.
Bold letters indicate significant tests (P<0.05)
Richness Abundance (sqrt) Evenness (sqrt)
Bugs v. o. int. 500 m v.a. 9.31 0.002 7.31 3.53 1.08 12.83 0.000 10.83 13.93 3.29 6.85 0.009 4.85 0.17 0.06
100 m c.f. 1.09 0.296 0.91 4.35 0.037 2.35 7.38 3.27 4.48 0.034 2.48 0.13 0.06
100 m CI
0.01 0.905 1.99 0.41 0.521 1.59 1.26 0.263 0.74
v.a. 9c. f. 0.01 0.920 1.99 0.11 0.744 1.89 0.07 0.791 1.93
v.a. 9CI
1.81 0.179 0.19 0.02 0.902 1.98 0.17 0.677 1.83
nuisance v. Temp. 1.57 0.211 0.43 5.64 0.018 3.64 5.04 0.025 3.04
Temp. (fi. m.) / / / 6.19 0.013 4.19 8.92 2.98 5.24 0.022 4.23 0.13 0.05
CN ratio 0.01 0.941 1.99 0.30 0.585 1.70 0.42 0.516 1.58
Tree radius 0.34 0.557 1.66 1.39 0.239 0.61 3.69 0.055 1.69
Leaf density 0.57 0.448 1.43 1.45 0.229 0.55 1.12 0.290 0.88
fu.m. vs. null m. 12.24 0.032 2.24 20.55 0.002 8.55 14.83 0.022 2.83
Richness (sqrt) Abundance (log) Evenness
Beetles v. o. int. 500 m v.a. 2.55 0.110 0.55 16.05 0.000 14.05 1.13 0.25 18.63 0.000 16.63 0.27 0.05
100 m c.f. 0.53 0.468 1.47 0.94 0.331 1.06 0.25 0.618 1.75
100 m CI
0.03 0.867 1.97 0.02 0.887 1.98 1.14 0.286 0.86
v.a. 9c. f. 0.18 0.669 1.82 0.00 0.979 2.00 1.97 0.161 0.03
v.a. 9CI
17.93 0.000 15.93 0.04 0.01 1.53 0.216 0.47 2.68 0.101 0.68
nuisance v. Temperature 0.89 0.346 1.11 1.07 0.300 0.93 0.41 0.523 1.59
CN ratio 0.19 0.665 1.81 0.23 0.628 1.77 4.27 0.039 2.27
CN ratio (fi. m.) / / / / / / 1.34 0.246 0.66
Tree radius 5.21 0.022 3.21 3.34 0.068 1.34 0.00 0.957 2.00
Tree radius (fi. m.) 4.60 0.000 2.60 0.00 0.00 / / / / / /
Leaf density 0.12 0.733 1.88 0.92 0.338 1.08 2.54 0.111 0.54
fu.m. vs. null m. 22.63 0.001 10.63 18.54 0.002 8.54 26.01 0.000 14.01
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
Richness (sqrt) Abundance (sqrt) Evenness
Leafhoppers v. o. int. 500 m v.a. 4.02 0.045 2.02 0.39 0.19 2.83 0.092 0.83 1.84 1.07 9.44 0.002 7.44 0.25 0.07
100 m c.f. 0.04 0.850 1.96 0.08 0.783 1.92 0.52 0.471 1.48
100 m CI
0.14 0.708 1.86 0.31 0.576 1.69 0.03 0.861 1.97
v.a. 9c. f. 0.38 0.535 1.62 0.18 0.674 1.82 0.87 0.350 1.13
v.a. 9CI
0.01 0.930 1.99 0.19 0.659 1.81 0.83 0.362 1.17
nuis. v. Temperature 1.29 0.256 0.71 0.13 0.717 1.87 0.04 0.848 1.96
CN ratio 0.96 0.327 1.04 0.87 0.350 1.13 0.29 0.589 1.71
Tree radius 0.82 0.365 1.18 0.05 0.817 1.95 0.17 0.676 1.83
Leaf density 0.73 0.392 1.27 0.46 0.497 1.54 0.49 0.485 1.51
fu.m. vs. null m. 4.59 0.468 5.41 3.59 0.610 6.41 11.70 0.039 1.70
Richness Abundance Evenness
Spiders v.i. 500 m v.a. 9.79 0.002 7.79 8.94 3.32 6.62 0.010 4.62 2.58 0.77 0.03 0.861 1.97
nuisance v. Temperature 0.03 0.857 1.97 2.48 0.115 0.48 2.30 0.129 0.30
CN ratio 2.46 0.117 0.46 0.08 0.771 1.92 0.59 0.441 1.41
CN ratio (fi. m.) / / / / / /
Tree radius 0.01 0.912 1.99 0.43 0.511 1.57 0.11 0.737 1.89
Leaf density 0.12 0.728 1.88 0.04 0.833 1.96 0.03 0.873 1.97
fu.m. vs. null m. 9.79 0.002 7.79 6.62 0.010 4.62 0.03 0.861 1.97
‘v. o. int’., variables of interest; ‘nuisance v.’, nuisance variables; ‘v.a.’, vegetated area amount; ‘c. f.’, connectivity factor; CI
,modified Hanski index; ‘temp.’, temperature; ‘fi.
m.’, final model; ‘fu. m.’, full model; ‘null m.’, null model; LRT, likelihood-ratio test statistic (Chi); ‘est.’, estimate of the respective variable in the final model ‘SE’, estimate stan-
dard error.
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
This is surprising as studies in agricultural areas found
that 100-m isolation has a strong effect on species diver-
sity (Knop et al., 2011; Sch
uepp et al., 2011) and
sometimes even outweighs the effect of 500-m habitat
amount (Farwig et al., 2009; Bailey et al., 2010). We
hypothesize that two reasons are responsible for this.
Table 4 Results of likelihood-ratio tests comparing mixed-effect models with and without respective predictors at the within-city
level (urban vegetation-rich versus urban vegetation-poor landscape sectors) for birch specialist insects sampled on thirty-six trees
in three Swiss cities. DAIC values always represent the calculation AIC
(more parsimonious model)
(less parsimonious model)
, that is
positive values indicate that the respective variable lowered the model AIC. Nuisance variables were only included if they lowered
the model AIC by >2. Variables of interest were tested with AIC-based backwards model simplification
Richness Abundance (sqrt)
Bugs v. o. int. 500 m v.a. 0.93 0.334 1.07 12.57 0.000 10.57 14.04 3.37
100 m c.f. 0.24 0.626 1.76 4.28 0.038 2.28 7.55 3.35
100 m CI
5.15 0.023 3.15 0.02 0.01 0.61 0.436 1.39
v.a. 9c. f. 0.02 0.898 1.98 0.16 0.692 1.84
v.a. 9CI
0.40 0.529 1.60 0.00 0.959 2.00
nuisance v. Temp. 0.46 0.497 1.54 5.85 0.016 3.85
Temp. (fi. m.) / / / 6.53 0.011 4.53 9.14 3.03
CN ratio 0.65 0.422 1.35 0.29 0.592 1.71
Tree radius 0.80 0.370 1.20 1.50 0.220 0.50
Leaf density 0.00 0.995 2.00 1.39 0.239 0.61
fu.m. vs. null m. 6.73 0.242 3.27 20.61 0.002 8.61
Richness (binomial, no transf.) Abundance (+1 log)
Beetles v. o. int. 500 m v.a. 4.25 0.039 2.25 2.36 1.33 17.94 0.000 15.94 4.02 0.83
100 m c.f. 0.27 0.603 1.73 0.25 0.616 1.75
100 m CI
0.01 0.908 1.99 0.88 0.347 1.12
v.a. 9c. f. / / / 0.09 0.766 1.91
v.a. 9CI
/ / / 2.29 0.130 0.29
nuis. v. Temp. 0.13 0.721 1.87 0.81 0.368 1.19
CN ratio 1.08 0.298 0.92 2.19 0.139 0.19
Tree radius 0.74 0.389 1.26 0.00 0.946 2.00
Leaf density 0.01 0.903 1.99 3.74 0.053 1.74
fu.m. vs. null m. 4.53 0.209 1.47 21.45 0.001 11.45
Richness Abundance (sqrt)
Leafhoppers v. o. int. 500 m v.a. 0.04 0.850 1.96 3.07 0.080 1.07 2.05 1.14
100 m c.f. 2.08 0.150 0.08 0.13 0.714 1.87
100 m CI
0.35 0.553 1.65 0.30 0.582 1.70
v.a. 9c. f. 0.07 0.795 1.93 0.09 0.769 1.91
v.a. 9CI
1.53 0.217 0.47 0.24 0.623 1.76
nuisance v. 0.33 0.566 1.67 0.03 0.863 1.97
CN ratio 0.03 0.860 1.97 0.80 0.370 1.20
Tree radius 2.59 0.108 0.59 0.79 0.376 1.21
Leaf density 5.26 0.022 3.26 0.56 0.455 1.44
Leaf density (fi. m.) 4.96 0.026 2.96 2.67 1.13 / / /
fu.m. vs. null m. 9.02 0.173 2.98 3.83 0.574 6.17
Transformations of the response variables are indicated in brackets. Bold letters indicate significant tests (P<0.05). ‘v. o. int’., vari-
ables of interest; ‘nuisance v.’, nuisance variables; ‘v.a.’, vegetated area amount; ‘c. f.’, connectivity factor; CI
,modified Hanski
index; ‘temp.’, temperature; ‘fi. m.’, final model; ‘fu. m.’, full model; ‘null m.’, null model; LRT, likelihood-ratio test statistic (Chi);
‘Est.’, estimate of the respective variable in the final model; ‘SE’, estimate standard error.
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
On the one hand, in cities, arthropods might adapt to
fragmentation by developing better dispersal abilities,
as has been shown for a grasshopper species (San Mar-
tin y Gomez & Van Dyck 2012), or cities might preselect
for species with good dispersal abilities. This idea has
mostly been studied for birds, but revealed ambiguous
results probably largely due to the difficulties to classi-
fying movement behavior (Møller, 2009; Evans et al.,
2011). On the other hand, the limited effects of isolation
at the 100-m scale might also reflect that the urban
matrix is less hostile compared to a typical matrix of
intensively managed agricultural land, where previous
studies on isolation effects were conducted. As woody
vegetation elements are typical in the urban mosaic
(Goddard et al., 2010), the urban matrix thus probably
offered more stepping-stones and dispersal corridors
for birch specialists than an average intensively man-
aged agricultural matrix. However, compared to many
cities worldwide, the cities studied here are relatively
small and thus might have provided more stepping-
stones than big metropolises. Nonetheless, our findings
indicate that the relative importance of potential habitat
amount and patch isolation differs between the urban
and rural ecosystem, with patch isolation having a
minor impact on species diversity and abundance of
arboreal arthropods in cities.
The abiotic local factors recorded as nuisance covari-
ables in this study (temperature and nitrogen content of
host plants) were overall weakly associated with
arthropod diversity and abundance. This is likely due
to the fact that variation between sites was generally
small, which was also intended as the focus of this
study was on larger-scale effects. Within the urban eco-
system, temperature was only important for bugs,
which exhibited higher abundance and reduced overall
evenness at warmer sites. Interestingly, our post hoc
analysis at the ecosystem level revealed that tempera-
ture negatively affected bug evenness in the urban as
well as rural ecosystem (no significant interaction). This
indicates that this indicator group contains single spe-
cies that particularly successfully profit from warmer
temperatures. We probably found small differences in
CN ratios between ecosystems because nitrogen depo-
sition is associated both with urbanization (Pickett
et al., 2011) and with intensive agricultural land use
(Kleijn et al., 2009).
The only arthropod diversity metric associated with
host tree nitrogen content was spider genus richness,
which was reduced at nitrogen-rich sites in the ecosys-
tem-level analysis. This is surprising, as spiders are pre-
dators and should hence not be directly influenced by
plant nitrogen content. The observed pattern, however,
might be a result of other factors associated with nitro-
gen deposition, namely air pollution in urban areas and
intensive management of the surrounding landscape in
agricultural areas. As members of a higher trophic
level, spiders could be more susceptible to toxic air pol-
lution (Butler & Trumble, 2008), or negative pollution
effects on herbivores could have been levelled out by
positive bottom-up effects of high host plant nitrogen
content (Tylianakis et al., 2008).
In this study, we show that urban areas can harbor a
similar (or even higher) arthropod diversity and abun-
dance compared to intensively managed agricultural
ecosystems. Further, we demonstrate that within cities
a high amount of vegetated area on a large (500 m)
scale per se enhances arthropod diversity and abun-
dance. In contrast to studies from agricultural areas, we
found little evidence for the importance of patch isola-
tion on species diversity and abundance of arboreal ar-
thropods in cities. While urbanization of rural areas
will always reduce the amount of vegetated area in the
landscape, the negative effects of this process may be
masked when intensively used agricultural land is
urbanized, given the well-known corrosion intensive
agricultural land use exerts on native communities.
Regarding processes within cities, our results highlight
the merits of incorporating landscape ecology methods
into urban ecology research, especially as the effects of
habitat isolation found in agro-ecosystems might not be
directly transferable to urban areas. They also indicate
that it is essential to provide a sufficient amount of veg-
etated areas in cities to maintain urban arthropod
diversity and underline the importance of appropriate
urban planning to promote the potential of cities to
become a refuge for species threatened by agricultural
We are grateful to the experts who identified species: Thomas
Friess and Rachel Korn (bugs), Rudolf Schuh (beetles), Gernot
Kunz (leafhoppers), Christian Komposch, Alexander Platz,
Marzena Sta
nska, and Izabela Hajdamowicz (spiders). We thank
Pius Wininger, Florian Krattinger, and Andrea Moser for assis-
tance in the field, Matthias Fries for providing his GIS expertise,
anzi Korner-Nievergelt for statistical advice, Peter H.W. Bie-
dermann, Myles H.M. Menz, Christof Sch
uepp, and two anony-
mous reviewers for their valuable comments on earlier versions
of this manuscript. This work would not have been possible
without the numerous private and public land owners who
allowed access to their properties. The MOD13Q1 data product
was obtained through the online Data Pool at the NASA Land
Processes Distributed Active Archive Center (LP DAAC),
USGS/Earth Resources Observation and Science (EROS) Center,
Sioux Falls, South Dakota (
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
Ahrne K, Bengtsson J, Elmqvist T (2009) Bumble bees (Bombus spp.) along a gradient
of increasing urbanization. PLoS One,4, e5574.
Aronson MFJ, La Sorte FA, Nilon CH et al. (2014) A global analysis of the impacts of
urbanization on bird and plant diversity reveals key anthropogenic drivers. Pro-
ceedings of the Royal Society B: Biological Sciences,281.
rspb.2013.3330 (accessed 12 February 2014).
Atkinson MD (1992) Betula pendula Roth (B. verrucosa Ehrh.) and B. pubescens Ehrh.
Journal of Ecology,80, 837870.
Bailey SA, Horner-Devine MC, Luck G et al. (2004) Primary productivity and species
richness: relationships among functional guilds, residency groups and vagility
classes at multiple spatial scales. Ecography,27, 207217.
Bailey D, Schmidt-Entling MH, Eberhart P, Herrmann JD, Hofer G, Kormann U, Her-
zog F (2010) Effects of habitat amount and isolation on biodiversity in fragmented
traditional orchards. Journal of Applied Ecology,47, 10031013.
ary P, B
aldi A, Sz
el G, Podluss
any A, Rozner I, Erd}
os S (2007) Responses of grass-
land specialist and generalist beetles to management and landscape complexity.
Diversity and Distributions,13, 196202.
Bates AJ, Sadler JP, Fairbrass AJ, Falk SJ, Hale JD, Matthews TJ (2011) Changing bee
and hoverfly pollinator assemblages along an urban-rural gradient. PLoS One,6,
Bates D, Maechler M, Bolker B and Walker S (2014) lme4: Linear mixed-effects models
using Eigen and S4. R package version 1.1-7. Available at: http://CRAN.R-projec- (accessed 1 September 2014).
Bender DJ, Tischendorf L, Fahrig L (2003) Using patch isolation metrics to predict ani-
mal movement in binary landscapes. Landscape Ecology,18,1739.
Bennett AB, Gratton C (2012) Local and landscape scale variables impact parasitoid
assemblages across an urbanization gradient. Landscape and Urban Planning,104,
Bommarco R, Kleijn D, Potts SG (2013) Ecological intensification: harnessing
ecosystem services for food security. Trends in Ecology & Evolution,28, 230
andle M, Brandl R (2001) Species richness of insects and mites on trees: expanding
Southwood. Journal of Animal Ecology,70, 491504.
Breuste J, Niemelae J, Snep RPH (2008) Applying landscape ecological principles in
urban environments. Landscape Ecology,23, 1139–1142.
Brook AJ, Woodcock BA, Sinka M, Vanbergen AJ (2008) Experimental verification of
suction sampler capture efficiency in grasslands of differing vegetation height and
structure. Journal of Applied Ecology,45, 13571363.
uckmann SV, Krauss J, Steffan-Dewenter I (2010) Butterfly and plant specialists suf-
fer from reduced connectivity in fragmented landscapes. Journal of Applied Ecology,
47, 799809.
Bundesamt f
ur Meteorologie und Klimatologie MeteoSchweiz (2014) Klimanormwerte,
Bundesamt f
ur Statistik (2014) Bodennutzung und Bodenbedeckung, Neuch^
atel, http://
Butler CD, Trumble JT (2008) Effects of pollutants on bottom-up and top-down pro-
cesses in insect-plant interactions. Environmental Pollution,156,110.
Cane JH, Minckley RL, Kervin LJ, Roulston TH, Williams NM (2006) Complex
responses within a desert bee guild (Hymenoptera: Apiformes) to urban habitat
fragmentation. Ecological Applications,16, 632644.
Chace JF, Walsh JJ (2006) Urban effects on native avifauna: a review. Landscape and
Urban Planning,74,4669.
Chaplin-Kramer R, O’rourke ME, Blitzer EJ, Kremen C (2011) A meta-analysis of crop
pest and natural enemy response to landscape complexity. Ecology Letters,14, 922
Clough Y, Kruess A, Kleijn D, Tscharntke T (2005) Spider diversity in cereal fields:
comparing factors at local, landscape and regional scales. Journal of Biogeography,
32, 20072014.
Croci S, Butet A, Georges A, Aguejdad R, Clergeau P (2008) Small urban woodlands
as biodiversity conservation hot-spot: a multi-taxon approach. Landscape Ecology,
23, 11711186.
Elman A, Hill J (2009) Data Analysis Using Regression and Multilevel/Hierarchical Mod-
els. Cambridge University Press, New York.
Evans KL, Chamberlain DE, Hatchwell BJ, Gregory RD, Gaston KJ (2011) What makes
an urban bird? Global Change Biology,17,3244.
Faeth SH, Bang C, Saari S (2011) Urban biodiversity: patterns and mechanisms.
Annals of the New York Academy of Sciences,1223,6981.
Farwig N, Bailey D, Bochud E et al. (2009) Isolation from forest reduces pollination,
seed predation and insect scavenging in Swiss farmland. Landscape Ecology,24,
Forstmeier W, Schielzeth H (2011) Cryptic multiple hypotheses testing in linear mod-
els: overestimated effect sizes and the winner’s curse. Behavioral Ecology and Socio-
Goddard MA, Dougill AJ, Benton TG (2010) Scaling up from gardens: biodiversity
conservation in urban environments. Trends in Ecology & Evolution,25,9098.
Hanski I (1994) A practical model of metapopulation dynamics. Journal of Animal Ecol-
ogy,63, 151162.
Hartley DJ, Koivula MJ, Spence JR, Pelletier R, Ball GE (2007) Effects of urbanization
on ground beetle assemblages (Coleoptera, Carabidae) of grassland habitats in
Western Canada. Ecography,30, 673684.
Hedblom M, S
om B (2010) Landscape effects on birds in urban woodlands: an
analysis of 34 Swedish cities. Journal of Biogeography,37, 13021316.
Imhoff ML, Bounoua L, Defries R, Lawrence WT, Stutzer D, Tucker CJ, Ricketts T
(2004) The consequences of urban land transformation on net primary productiv-
ity in the United States. Remote Sensing of Environment,89, 434443.
Johnson PTJ, Hoverman JT, Mckenzie VJ, Blaustein AR, Richgels KLD, Cadotte M
(2013) Urbanization and wetland communities: applying metacommunity theory
to understand the local and landscape effects. Journal of Applied Ecology,50,34
Kleijn D, Kohler F, Baldi A et al. (2009) On the relationship between farmland biodi-
versity and land-use intensity in Europe. Proceedings of the Royal Society B-Biological
Sciences,276, 903909.
Knop E, Herzog F, Schmid B (2011) Effect of connectivity between restoration mead-
ows on invertebrates with contrasting dispersal abilities. Restoration Ecology,19,
Lassau SA, Hochuli DF (2008) Testing predictions of beetle community patterns
derived empirically using remote sensing. Diversity and Distributions,14, 138147.
Lee PF, Ding TS, Hsu FH, Geng S (2004) Breeding bird species richness in Taiwan:
distribution on gradients of elevation, primary productivity and urbanization.
Journal of Biogeography,31, 307314.
Lizee M-H, Manel S, Mauffrey J-F, Tatoni T, Deschamps-Cottin M (2012) Matrix con-
figuration and patch isolation influences override the species-area relationship for
urban butterfly communities. Landscape Ecology,27, 159169.
Lu D, Xu X, Tian H, Moran E, Zhao M, Running S (2010) The effects of urbanization
on net primary productivity in Southeastern China. Environmental Management,46,
MacArthur RH, Wilson EO (1967) The Theory of Island Biogeography. Princeton Univer-
sity Press, Princeton, NJ.
Magura T, Lovei GL, Tothmeresz B (2010) Does urbanization decrease diversity in
ground beetle (Carabidae) assemblages? Global Ecology and Biogeography,19,1626.
Magurran AE (2004) Measuring Biological Diversity, Malden. Blackwell Science, MA.
McDonnell MJ, Hahs AK (2008) The use of gradient studies in advancing our under-
standing of the ecology of urbanizing landscapes; current status and future direc-
tions. Landscape Ecology,23, 11431155.
McKinney ML (2008) Effects of urbanization on species richness: a review of plants
and animals. Urban Ecosystems,11, 161176.
Milesi C, Elvidge CD, Nemani RR, Running SW (2003) Assessing the impact of urban
land development on net primary productivity in the southeastern United States.
Remote Sensing of Environment,86, 401410.
Miranda AC, Schielzeth H, Sonntag T, Partecke J (2013) Urbanization and its effects
on personality traits: a result of microevolution or phenotypic plasticity? Global
Change Biology,19, 26342644.
Mittelbach GG, Steiner CF, Scheiner SM et al. (2001) What is the observed relationship
between species richness and productivity? Ecology,82, 23812396.
Møller AP (2009) Successful city dwellers: a comparative study of the ecological char-
acteristics of urban birds in the Western Palearctic. Oecologia,159, 849858.
Møller AP, Diaz M, Flensted-Jensen E et al. (2012) High urban population density of
birds reflects their timing of urbanization. Oecologia,170, 867875.
Nasa Land Processes Distributed Active Archive Center Lp Daac MOD13Q1. USGS/
Earth Resources Observation and Science (EROS) Center. Sioux Falls, South Dakota.
Ockinger E, Dannestam
A, Smith HG (2009) The importance of fragmentation and
habitat quality of urban grasslands for butterfly diversity. Landscape and Urban
Oksanen J, Guillaume Blanchet F, Kindt R et al. (2013) Vegan: Community Ecology Pack-
age. R Package Version 2.0-10. Available at:
vegan. (accessed 1 September 2014).
Paton D, Romero F, Cuenca J, Carlos Escudero J (2012) Tolerance to noise in 91 bird
species from 27 urban gardens of Iberian Peninsula. Landscape and Urban Planning,
Pau S, Gillespie TW, Wolkovich EM (2012) Dissecting NDVI-species richness relation-
ships in Hawaiian dry forests. Journal of Biogeography,39, 16781686.
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
Pettorelli N, Ryan S, Mueller T, Bunnefeld N, Jedrzejewska B, Lima M, Kausrud K
(2011) The Normalized Difference Vegetation Index (NDVI): unforeseen successes
in animal ecology. Climate Research,46,1527.
Pickett ST, Cadenasso ML, Grove JM et al. (2011) Urban ecological systems: scientific
foundations and a decade of progress. Journal of Environmental Management,92,
Purvis A, Hector A (2000) Getting the measure of biodiversity. Nature,405, 212219.
R Core Team (2012) R 2.15.2: A Language and Environment for Statistical Computing.R
Foundation for Statistical Computing V, Austria, Vienna, Austria.
Raupp MJ, Shrewsbury PM, Herms DA (2010) Ecology of herbivorous arthropods in
urban landscapes. Annual Review of Entomology,55,1938.
Rickman JK, Connor EF (2003) The effect of urbanization on the quality of
remnant habitats for leaf-mining lepidoptera on Quercus agrifolia.Ecography,26,
Sadler JP, Small EC, Fiszpan H, Telfer MG, Niemela J (2006) Investigating
environmental variation and landscape characteristics of an urban-rural
gradient using woodland carabid assemblages. Journal of Biogeography,33, 1126
San Martin y, Gomez G, Van Dyck H (2012) Ecotypic differentiation between urban
and rural populations of the grasshopper Chorthippus brunneus relative to climate
and habitat fragmentation. Oecologia,169, 125133.
uepp C, Herrmann JD, Herzog F, Schmidt-Entling MH (2011) Differential effects
of habitat isolation and landscape composition on wasps, bees, and their enemies.
Oecologia,165, 713721.
Seto KC, Fleishman E, Fay JP, Betrus CJ (2004) Linking spatial patterns of bird and
butterfly species richness with Landsat TM derived NDVI. International Journal of
Remote Sensing,25, 43094324.
Seto KC, Guneralp B, Hutyra LR (2012) Global forecasts of urban expansion to 2030
and direct impacts on biodiversity and carbon pools. Proceedings of the National
Academy of Sciences of the United States of America,109, 1608316088.
Shochat E, Stefanov WL, Whitehouse MEA, Faeth SH (2004) Urbanization and spider
diversity: influences of human modification of habitat structure and productivity.
Ecological Applications,14, 268280.
Shwartz A, Muratet A, Simon L, Julliard R (2013) Local and management variables
outweigh landscape effects in enhancing the diversity of different taxa in a big
metropolis. Biological Conservation,157, 285292.
Smith RM, Warren PH, Thompson K, Gaston KJ (2006) Urban domestic gardens (VI):
environmental correlates of invertebrate species richness. Biodiversity and Conserva-
tion,15, 24152438.
Soga M, Koike S (2012) Relative importance of quantity, quality and isolation of
patches for butterfly diversity in fragmented urban forests. Ecological Research,27,
Strauss B, Biedermann R (2006) Urban brownfields as temporary habitats: driving
forces for the diversity of phytophagous insects. Ecography,29, 928940.
Sushinsky JR, Rhodes JR, Possingham HP, Gill TK, Fuller RA (2013) How should we
grow cities to minimize their biodiversity impacts? Global Change Biology,19, 401
Swiss Federal Office of Topography Swisstopo (2013) Vector25, Swiss Federal Office
of Topography Swisstopo, Wabern, Switzerland.
Torma A, Cs
ar P (2013) Species richness and composition patterns across trophic
levels of true bugs (Heteroptera) in the agricultural landscape of the lower reach
of the Tisza River Basin. Journal of Insect Conservation,17,3551.
Tylianakis JM, Didham RK, Bascompte J, Wardle DA (2008) Global change and spe-
cies interactions in terrestrial ecosystems. Ecology Letters,11, 13511363.
United Nations Department of Economic and Social Affairs, Population Division
(2012) World Urbanization Prospects: The 2011 Revision: Highlights. United Nations
Department of Economic and Social Affairs, Population Division, New York, Uni-
ted Nations.
Waide RB, Willig MR, Steiner CF et al. (1999) The relationship between productivity
and species richness. Annual Review of Ecology and Systematics,30, 257300.
Yu D, Shao H, Shi P, Zhu W, Pan Y (2009) How does the conversion of land cover to
urban use affect net primary productivity? A case study in Shenzhen city, China.
Agricultural and Forest Meteorology,149, 20542060.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Table S1. Study sites.
Table S2. Buffer overlap.
Table S3. Temperature-, CN ratio-, and NDVI values.
Table S4. Post hoc analyses.
Figure S1. Visual overview of (a) study locations and (b)
within-city design.
©2015 John Wiley & Sons Ltd, Global Change Biology,21, 1652–1667
... Functional connectivity, defined as the extent to which favoured areas are linked up, is positively related to biodiversity [4,10]. In particular, canopy cover is a central factor as it offers sites for habitat, food, shelter, nesting, etc.; thus, the species richness is greater the larger the canopy patches are [11,12]. The functional connectivity between canopy patches facilitates the movement of genes, individuals, populations, and species [13][14][15][16], and allows the recolonisation of empty habitat patches [17], the migration and persistence of metapopulations [18], and the ability of species to expand or alter their range in response to climate change [19][20][21]. ...
Full-text available
The connectivity of green infrastructure facilitating the movement of organisms is the key to strengthening biodiversity in cities. Brownfields are a valuable land resource, with their revitalisation as a Nature Based Solution high on the policy agenda. In supporting cities which simultaneously aim for densification and the maintenance or further development of greenery, this paper develops a model for identifying and prioritising the role of revitalised and prevailing brownfields for the connectivity of green infrastructure using the example of Leipzig, Germany. Comparing metrics between land use categories, brownfields have a central role as stepping stones, with a value of 13%, while revitalised brownfields substantially contribute to global connectivity, with a value of 87% being equally important, for example, with Leipzig’s central parks. This paper’s spatial-explicit network approach provides a complementary planning tool for prioritising brownfields and the added value of their renaturing by identifying (a) strategic functional corridors formed by brownfields, (b) the connectivity relevance and exposure of individual brownfields, and (c) how renatured brownfields would strengthen existing corridors and form alternative paths. This paper presents an approach using freely available software tools and high-resolution canopy data as a proxy for functional connectivity which serves as a standardised and comparable ex-ante evaluation of NBS strategies being implemented in other cities.
... These birds use buildings for roosting, foraging and nesting and are mostly unaffected by noise and human presence, such species can be termed 'Urban Exploiters' (Mohring et al., 2021). Avian species which were closely linked to humans like House Sparrow (Passer domesticus) and House Crow (Corvus splendens) were not recorded in the urban area which could be due to the reason like replacement of natural biotic cover with an artificial substrate like concrete, lawns and asphalt (Turrini et al., 2015). The other bird species like Black Kite (Milvus migrans), Indian Roller (Coracias benghalensis), Rose Ring Parakeet (Psittacula krameri) and Black Drongo (Dicrurus macrocercus) were recorded in adequate numbers throughout the city. ...
Full-text available
With rapid growth and changes in daily life, air pollution is also increasing at a high rate. Air pollution threats are not only restricted to urban areas but harm rural areas also. Apart from being harmful to human beings; air pollution possesses a negative impact on bird species also. This study was carried out to find out the adverse impact of air pollution on the diversity of the avian community. The study was executed at five different locations in Bilaspur city during the summer season (2022). Vehicular emissions, burning of fossil fuels, constructions etc. are the major source of pollution in the city. The point count and checklist method was adopted for the observation of bird species. The air quality and pollution monitoring had been carried out through the 'Smiledrive Air Quality Monitor Pollution Meter' which detects the concentration and level of PM 2.5, PM 10, TVOC and HCHO in the air. The diversity of bird species was calculated through total species richness and the Shannon-Wiener diversity index. It was observed that the site having minimum pollution levels have a large bird population with maximum diversity and the sites having high pollution levels have the least diversity of birds. It is the reason that many bird species avoid areas with high pollution concentrations. The study also revealed the remarkably high population of birds of the 'Sturnidae' and 'Columbidae families in polluted sites which validates that the birds of these families have adapted themselves well in the sites with high pollution levels.
... As cities do not only affect the built environment on which urban actors typically focus, but the character and function of their non-urban hinterland through the urban dependence of high-efficient land-, water-and air-based infrastructures for production, managing and transportation of resources, todays urban practice are effecting ecosystems in scales transgressing city borders [10,11]. Thus, urban biodiversity can indirectly and directly be beneficial to the planet, its global/local climate and to the diverse societies of organisms living in and around cities, including humans [6,12]. One might say that due to its multifaceted connections and effects, urban biodiversity IOP Publishing doi: 10.1088/1755-1315/1122/1/012008 2 could be playing a key role for the resilience of Earth, not least as a potential driver for changes in economic and political structures and societal norms indirectly impacting the comprehensive changes needed to bent the curve of biodiversity loss [2,3,13,14]. ...
Full-text available
The decline of biodiversity is progressing at an alarming scale and pace. A strengthening of relations between humans and nature may hold the key to a sustainable future, having the potential to increase human awareness of planetary well-being. Recently, urban actors across the world have seemingly accelerated the integration of Regenerative Thinking into strategies targeting biodiversity. Approaching a well-established urban planning and design practice with a new ideology, urban actors must tackle how to integrate this new arena of Regenerative Thinking in their work. This case study analyzes how urban actors such as architects and planners translate ideas of biodiversity into urban development in Copenhagen. The study identifies drivers and barriers related to the integration of biodiversity in Copenhagen and discuss potentials and challenges in this relation, including asking whether a new methodology such as a dialogue tool might have the potential to support urban actors working with biodiversity. The study concludes that urban actors are experiencing a range of drivers and challenges related to their work with biodiversity, and new methodologies seems to be needed. The study is an initial explorative component of a larger case study planned to be key content of a Ph.D. thesis.
... Other arthropod communities, such as ants, also show homogenization or filtering due to urban heat islands and climate warming [71,77]. By increasing or decreasing the survival and fitness of species, urban heat islands can lead to biotic homogenization as heat tolerant taxa become more abundant and others rare [78,79]. ...
Full-text available
Purpose of Review Our goal is to provide an overview of how urban heat islands affect forests and synthesize recent literature on that topic. We focused on direct effects of high temperatures from urban heat islands on forest trees and indirect effects via changes in soil moisture and pest density. We also focused on the effects of urban heat islands on arthropods with particular emphasis on tree pests. Recent Findings Urban heat islands can push trees and arthropods closer to their thermal limits with consequences for tree growth and arthropod fitness. Urban heat islands can alter the distribution of trees and arthropods allowing species to survive at higher altitudes or latitudes than they could otherwise. A primary risk for trees is that urban heat islands can increase pest density and damage. Summary Urban heat islands can increase forest air and soil temperature and reduce soil moisture especially when combined with greater climate change. Land managers should consider the surrounding urban density and forest size when trying to determine which plants and animals can persist in urban forests. As forests are fragmented or encroached upon by urbanization, the forest environment will change and become more hospitable for some species and less hospitable for others. Overall, there is insufficient research focused on urban-forest interfaces and the consequences of urbanization for plants and animals within forests. This research is not only important for urban forest conservation. Tree and arthropod responses to urban heat islands will help scientists and land managers predict responses to climate warming in rural areas as well.
... Heat extremes in cities are profoundly and detrimentally affecting urban residents and ecosystems, driving heat-related mortalities (Dousset et al., 2011;Scherer et al., 2013), respiratory and cardiovascular health issues (Kovats and Hajat, 2008), enhanced air pollution (Koken et al., 2003;Krug et al., 2019), higher energy use and carbon emissions (Santamouris and Kolokotsa, 2015;Tooke et al., 2014), and unforeseen shifts in urban biodiversity (Turrini and Knop, 2015). ...
Full-text available
The world’s expanding cities are increasingly at risk from the effects of climate and global change. Heat extremes pose one of the greatest threats to urban residents, causing mortalities, detrimental health effects, economic losses, increased energy consumption, and exacerbated air pollution. Various frameworks, such as the Sponge City, have been proposed to increase the resilience of cities to multiple environmental threats, mainly relying on the ecosystem services provided by urban greenery. Sustainability initiatives in cities need to be implemented at a neighborhood scale capturing socio-demographic vulnerability to heat stress. However, high-resolution, spatially distributed data to identify hotspots of heat exposure are largely unavailable. In particular, two major indicators of thermal comfort in cities - air temperature and evapotranspiration (ET) - urgently need to be characterized. However, their estimation remains challenging, as the urban environment is highly horizontally and vertically heterogeneous. The majority of existing process-based models cannot handle the urban complexity or be upscaled to the city-level. To address these challenges, state-of-the-art technologies, in particular, artificial intelligence (AI), remote sensing, and crowdsourcing, have emerged as a solution to producing spatially-explicit and accurate information on parameters relevant to climate change mitigation. Thus, the overall goal of this thesis was to characterize thermal comfort indicators in an urban environment at a high spatial resolution using remote sensing, open geodata, and AI. Berlin, Germany was selected as a test site due to its high risk of heat-related mortalities and established infrastructure of atmospheric measurements. This thesis had two main objectives: (1) investigate how accurately thermal comfort indicators in urban areas can be modeled at a high spatial resolution with remote sensing imagery, geodata, and meteorological data using machine learning (ML) regression algorithms; and (2) assess the spatial variation of the modeled ET and air temperature across Berlin to provide insights for urban planning and policy focused on heat risk mitigation. Air temperature, especially at night, is one of the best indicators of heat stress. However, high-resolution air temperature maps have until recently remained unavailable due to the limited number of weather stations in cities. Thus, the first part of this thesis focused on how open source remote sensing, GIS and crowdsourced weather data can be used to predict the spatial distribution of nocturnal air temperature one day in advance at a 30-m resolution for the entire city using ML. Three predictor scenarios were tested: (1) only remotely sensed predictors, (2) only the spatially interpolated crowdsourced air temperature from the previous day, and (3) both remotely sensed and crowdsourced data from the previous day. As crowdsourced weather data is associated with more uncertainties than traditional weather data, validation was conducted with both crowdsourced and traditional air temperature data using “leave-one-date-out” cross-validation. Using only remotely sensed predictors showed the highest overall accuracy when validating with conventional air temperature data, demonstrating that remote sensing data is essential to accurately predicting air temperature in the gaps between citizen weather stations. Accuracy was high when validating with crowdsourced data, but decreased when validating with independent data, likely due to the effect of citizen weather stations’ installation near building walls. However, accuracy was highest in highly sealed urban fabric, where heat risk is highest, demonstrating the utility of this approach for identifying dangerous hotspots within the city. ET provides an essential cooling service and approximates vegetation water use, making it a key indicator for urban sustainability. ET is defined as the water transported from the land surface to the atmosphere and consists mainly of soil evaporation, plant transpiration, and evaporation of intercepted precipitation in terrestrial ecosystems. To improve ET modeling in urban areas, a novel approach was developed which incorporated diverse land cover contributing to urban ET measurements into AI algorithms. Flux footprint modeling, which estimates the source area of eddy covariance (EC) measurements, was used to extract weighted averages from remote sensing and GIS data. Incorporating remotely sensed data extracted with flux footprints enhanced the predictive accuracy of models. In the next phase, this approach was extended to map ET for the entire city at a high spatial (10-m) and temporal (hourly) resolution for one year. Validation showed that this approach is reliable for mapping urban ET. Lastly, to support urban planning, the spatial variation of both modeled thermal comfort indicators was analyzed in relation to land cover. Ultimately, using remote sensing and AI to characterize indicators of thermal comfort can support urban planning aiming to reduce heat risks for an increasingly urban world population.
... On the other hand, cities may preselect for species with good dispersal abilities. This idea has mostly been studied for birds and invertebrates, but yielded ambiguous results, probably largely due to the difficulties in classifying movement behavior [60,61]. In addition, the model selection results of B. gargarizans revealed small differences at different spatial scales. ...
Full-text available
Cities are growing rapidly worldwide, with over half of the human population living in cities. Amphibians are the most threatened vertebrates on our planet and are particularly vulnerable to the effects of urbanization. While it is known that landscape features and scales are important for amphibians in urban areas, we do not adequately understand how the urban landscape affects diversity patterns, nor have we identified which spatial scale is most appropriate for evaluating how amphibians respond to urban environments. In this study, we examined the relationships between anuran abundance/richness and landscape features at four spatial scales in Shanghai, China. In order to determine the relative importance of landscape variables and the most appropriate spatial scale, a multi-model inference approach was used to evaluate and compare model weighted mean coefficients. Our results show that large spatial scales, i.e., 1500 m and 2000 m, best predicted relative anuran abundance and richness, while the total anuran abundance responded most strongly to landscape variables at smaller scales, i.e., 500 m and 1000 m. Patch richness and the interspersion and juxtaposition index play a large role in predicting the anuran species’ richness and abundance. The abundance of P. nigromaculatus, F. multistriata, and B. gargarizans increased with patch richness. Species richness and total abundance were most strongly related to the interspersion and juxtaposition index. Our research highlights the importance of identifying the most suitable spatial scale in urban environments because not all anuran respond to the same spatial scale. We found that the relationships between anuran relative abundance and species and urban habitat features are not consistent with the prediction of other landscapes (e.g., farmland, forest, and island). Additionally, constructing diverse habitat patches and more neighboring habitats may maintain or improve anuran communities in urbanizing landscapes.
... Cities in Ghana have increased both in population and in areal extent (Seto et al., 2012;Turrini & Knop, 2015). Rapid spatial urban expansion is a key characteristic of the major urbanised centres in Ghana, including Accra and Kumasi Nyamekye et al., 2020;Stemn & Agyapong, 2014;Wemegah et al., 2020). ...
Full-text available
Accra and Kumasi are the two major cities in Ghana. Spatial urban expansion has been experienced in transforming different non-urban Land Use Land Cover (LULC) types into urban/built-up areas with a potential direct relationship to temperature rise in the cities. Thus, this dissertation aims to establish the relationship between urban spatial expansion and temperature in Accra and Kumasi metropolis. Multi-source datasets such as remote sensing images, different GIS vector layers, reference maps and historical temperature datasets were used for this retrospective study. This research was grouped under three components: environmental science, environmental technology, as well as environmental management and planning. From the environmental science component, LULC maps were produced for different years to assess the trend of temporal change in the various LULC classes in the two metropolises. Remote sensing indices and land surface temperature were retrieved from the remote sensing images to determine their correlations. Temperature time series was analysed by calculating temperature indices and determining temporal trends to reveal changes in air temperature to detect urban warming and its impacts. From the environmental technology, the novel random forest algorithm was utilised to classify the satellite images of both cities since previous works have utilised other traditional classifiers. The satellite imageries were used for point-based estimation of temperature to determine Urban Heat Islands (UHIs) hotspots. For environmental management and planning, spatial urban expansion techniques were utilised to ascertain trends in urban/built-up areas, especially in both cities' sub-metropolitan zones. For prescient purposes, future LULC modelling was implemented to provide insights into the proportions of the various LULC changes in 2025. The analysis identified two salient findings: increased urban/built-up areas at the expense of agricultural and forestlands throughout the study period and the positive correlation between spatial urban expansion and temperature. This indicated warming up of urban temperature in both cities. The major findings in this dissertation provided evidence of how integrated datasets and research techniques can be utilised for LULC changes to determine the relationship between spatial urban expansion and temperature at local scales. Institutions such as metropolitan assemblies and policymakers may adopt the concepts demonstrated in this work to rapidly assess urban environments and investigate the relationship between spatial urban expansion and temperature.
... Biodiversity in urban landscapes is affected by habitat loss and fragmentation, reducing habitat quantity, quality, and connectivity between patches [1,2]. In cities, land use change through the conversion of natural landcover to various human uses tend to be dominated by impervious landcover types (roads, residential areas) [3,4]. To support growing human populations, urbanization is expected to intensify by 2050 globally, where up to 67.2% of the world's population is expected to live in cities [5]. ...
Full-text available
Urbanization adversely impacts biodiversity by reducing the quantity and quality of natural habitat areas. Additionally, the quality of natural habitat depends on its bio-physical characteristics (e.g., natural cover, impervious surfaces, urban tree canopy) as well as the functional traits of species inhabiting them (e.g., breeding/foraging habitat requirements). To better plan conservation of regional biodiversity in urbanized landscapes, it is therefore critical to assess the relationship between the landscape and the response of key Functional Trait Groups (FTGs) of species. To identify different FTGs of 116 avian species in the urbanized landscape of the Toronto region (Canada), we conducted a Functional Trait Analysis (FTA) using RLQ-fourth corner analysis. We focused on four species traits (diet, foraging, nesting, and territoriality) to identify the FTGs and their association with natural cover and landscape characteristics (landcover types, patch quality, habitat connectivity). Then, to predict FTG presence in relation to the landscape characteristics, we performed a Habitat Suitability Analysis (HSA). From this analysis, we found 21 avian FTGs with different habitat suitability values that correspond to forested patches and wetlands. The HSA for tree canopy, forest insectivore, and ground-nesting birds (or FTGs) have higher suitability values within forest patches, while aerial insectivores have higher suitability values in older residential neighborhoods indicating the value of the urban tree canopy. This methodological approach shows that by mapping habitat suitability by FTG one can identify strategic conservation areas that target multiple species, shifting efforts from a single species to a community-based functional focus. Our study highlights the conservation value of remnant and/or restored habitat patches in near urban and urban landscapes that help to maximize the persistence of regional avian biodiversity.
Full-text available
Here, the abundance of macro-invertebrates (Arthropoda and Gastropoda) of eight green roofs and their adjacent ground level habitats in the city of Antwerp, Belgium, is compared. All higher-level taxa found were present in both types of habitats without significant differences in their overall abundance between green roofs and ground level habitats. However, we found significant differences in abundances between the two types of habitats, when specific taxa were compared. Beetles (Coleoptera), isopods (Isopoda) and bees (Anthophila) were more abundant at ground level sites compared to green roofs, while for true bugs (Heteroptera) and cicadas (Auchenorrhyncha) the opposite was found. Our results support the idea that extensive green roofs in Belgium can provide a suitable habitat for different invertebrate taxa, but further research is needed to identify the true drivers behind differences in abundance between ground level and adjacent green roofs.
Urban environments are evolutionarily novel and differ from natural environments in many respects including food and/or water availability, predation, noise, light, air quality, pathogens, biodiversity, and temperature. The success of organisms in urban environments requires physiological plasticity and adjustments that have been described extensively, including in birds residing in geographically and climatically diverse regions. These studies have revealed a few relatively consistent differences between urban and non-urban conspecifics. For example, seasonally breeding urban birds often develop their reproductive system earlier than non-urban birds, perhaps in response to more abundant trophic resources. In most instances, however, analyses of existing data indicate no general pattern distinguishing urban and non-urban birds. It is, for instance, often hypothesized that urban environments are stressful, yet the activity of the hypothalamus-pituitary-adrenal axis does not differ consistently between urban and non-urban birds. A similar conclusion is reached by comparing blood indices of metabolism. The origin of these disparities remains poorly understood, partly because many studies are correlative rather than aiming at establishing causality, which effectively limits our ability to formulate specific hypotheses regarding the impacts of urbanization on wildlife. We suggest that future research will benefit from prioritizing mechanistic approaches to identify environmental factors that shape the phenotypic responses of organisms to urbanization and the neuroendocrine and metabolic bases of these responses. Further, it will be critical to elucidate whether factors affect these responses (a) cumulatively or synergistically; and (b) differentially as a function of age, sex, reproductive status, season, and mobility within the urban environment. Research to date has used various taxa that differ greatly not only phylogenetically, but also with regard to ecological requirements, social systems, propensity to consume anthropogenic food, and behavioral responses to human presence. Researchers may instead benefit from standardizing approaches to examine a small number of representative models with wide geographic distribution and that occupy diverse urban ecosystems.
Technical Report
Full-text available
Description Fit linear and generalized linear mixed-effects models. The models and their components are represented using S4 classes and methods. The core computational algorithms are implemented using the 'Eigen' C++ library for numerical linear algebra and 'RcppEigen' ``glue''.
In urban brownfields (derelict sites), we studied the influence of local factors (successional age, vegetation structure, soil) and landscape context (spatial arrangement of brownfields of different successional stages) on the diversity of phytophagous insects, grasshoppers and leafhoppers (Orthoptera and Hemiptera: Auchenorrhyncha). The study was conducted on a total of 246 plots in the cities of Bremen and Berlin, Germany. We used a habitat modelling approach, enabling us to predict the community from single species models (30 species in Bremen, 28 in Berlin). The results revealed that communities were predominantly determined by vegetation structure, followed by landscape context, soil parameters and site age. For most species, local factors were the most important. Only few species were strongly influenced by landscape context, even though some showed clear negative reactions to low proportions of brownfields in the surroundings. Along a successional gradient of vegetation structure, from scarce and low to dense and high vegetation, the insect community was not static. Even though species numbers remained comparatively constant, species composition changed considerably. Many species showed clear preferences for certain successional stages. Thus, maintaining the regional species pool of a city requires a mosaic of all successional stages.
Aim A growing body of research has used the normalized difference vegetation index (NDVI) as a proxy for productivity to predict species richness. Yet the mechanisms that produce the relationship between NDVI and species richness remain unclear because of correlated biotic and abiotic factors that influence NDVI. In this study we investigated different biotic and abiotic effects that potentially drive plant species richness–productivity relationships. Location Hawaiian Islands, USA. Methods We quantified woody plant species richness, structure (density, basal area and canopy height), and species composition along a precipitation gradient of 14 Hawaiian dry forest plots. We then used structural equation models combined with 10 years of satellite data to disentangle the effects of precipitation, structure and NDVI-estimated productivity on species richness. Results Underlying the simple correlation between NDVI and species richness was the indirect effect of precipitation and direct effect of forest structure. The best-fit model showed there was no direct effect of NDVI on species richness. Main conclusions Our results demonstrate that complex relationships drive simple correlations between species richness and productivity. Considering the mechanisms and underlying factors driving NDVI–species richness relationships could improve predictions of species diversity as satellite measures of productivity have an increasingly important role in habitat mapping, species distribution modelling and predictions for global change.