Electromagnetic radiation of mobile telecommunication antennas affects the abundance and composition of wild pollinators

Abstract

The exponential increase of mobile telephony has led to a pronounced increase in electromagnetic fields in the environment that may affect pollinator communities and threaten pollination as a key ecosystem service. Previous studies conducted on model species under laboratory conditions have shown negative effects of electromagnetic radiation (EMR) on reproductive success, development, and navigation of insects. However, the potential effects that widespread mobile telecommunication antennas have on wild pollinator communities outside the laboratory microcosm are still unknown. Here we studied the effects of EMR from telecommunication antennas on key wild pollinator groups (wild bees, hoverflies, bee flies, remaining flies, beetles, butterflies, and wasps). We measured EMR at 4 distances (50, 100, 200 and 400 m) from 10 antennas (5 on Limnos Island and 5 on Lesvos Island, eastern Mediterranean, Greece), and correlated EMR values with insect abundance and richness (the latter only for wild bees and hoverflies). All pollinator groups except butterflies were affected by EMR. In both islands, beetle, wasp, and hoverfly abundance decreased with EMR, whereas the abundance of underground-nesting wild bees and bee flies unexpectedly increased with EMR. The effect of EMR on the abundance of remaining flies differed between islands. With respect to species richness, EMR only tended to have a negative effect on hoverflies in Limnos. As EMR affected the abundance of several insect guilds negatively, and changed the composition of wild pollinators in natural habitats, it might also have additional ecological and economic impacts on the maintenance of wild plant diversity, crop production and human welfare.

Full text

ORIGINAL PAPER
Electromagnetic radiation of mobile telecommunication antennas
affects the abundance and composition of wild pollinators
A. La
´zaro
1,2
•A. Chroni
1
•T. Tscheulin
1
•J. Devalez
1
•C. Matsoukas
3
•
T. Petanidou
1
Received: 9 October 2015 / Accepted: 17 April 2016
ÓSpringer International Publishing Switzerland 2016
Abstract The exponential increase of mobile telephony
has led to a pronounced increase in electromagnetic fields
in the environment that may affect pollinator communities
and threaten pollination as a key ecosystem service. Pre-
vious studies conducted on model species under laboratory
conditions have shown negative effects of electromagnetic
radiation (EMR) on reproductive success, development,
and navigation of insects. However, the potential effects
that widespread mobile telecommunication antennas have
on wild pollinator communities outside the laboratory
microcosm are still unknown. Here we studied the effects
of EMR from telecommunication antennas on key wild
pollinator groups (wild bees, hoverflies, bee flies, remain-
ing flies, beetles, butterflies, and wasps). We measured
EMR at 4 distances (50, 100, 200 and 400 m) from 10
antennas (5 on Limnos Island and 5 on Lesvos Island,
eastern Mediterranean, Greece), and correlated EMR val-
ues with insect abundance and richness (the latter only for
wild bees and hoverflies). All pollinator groups except
butterflies were affected by EMR. In both islands, beetle,
wasp, and hoverfly abundance decreased with EMR,
whereas the abundance of underground-nesting wild bees
and bee flies unexpectedly increased with EMR. The effect
of EMR on the abundance of remaining flies differed
between islands. With respect to species richness, EMR
only tended to have a negative effect on hoverflies in
Limnos. As EMR affected the abundance of several insect
guilds negatively, and changed the composition of wild
pollinators in natural habitats, it might also have additional
ecological and economic impacts on the maintenance of
wild plant diversity, crop production and human welfare.
Keywords Bee flies Beetles Butterflies Distance to the
antenna Electromagnetic smog EMR Hoverflies 
Species richness Wasps Wild bees
Introduction
Pollinators play an important functional role in most ter-
restrial ecosystems and provide a key ecosystem service
that is vital to the maintenance of wild plant communities
and agricultural productivity (Klein et al. 2007; Kremen
et al. 2007; Potts et al. 2010). Over the past decade, several
studies have warned about the decline in pollinators
(Biesmeijer et al. 2006; Pauw 2007; Goulson et al. 2008;
Burkle et al. 2013) and the serious consequences this may
have (e.g. Ashman et al. 2004; Burkle et al. 2013). This
pollinator loss has been related to anthropogenic distur-
bances such as alterations in land use, habitat loss and
climate change (Kearns et al. 1998; Aguilar et al. 2006;
Hegland et al. 2009; Potts et al. 2010). At the same time,
the use of mobile telephony has grown exponentially dur-
ing recent years, resulting in a pronounced increase in
electromagnetic fields in the environment. Detrimental
effects of electromagnetic exposure have been shown for a
Electronic supplementary material The online version of this
article (doi:10.1007/s10841-016-9868-8) contains supplementary
material, which is available to authorized users.
&A. La
´zaro
amparo.lazaro@imedea.uib-csic.es
1
Laboratory of Biogeography and Ecology, Department of
Geography, University of the Aegean, University Hill,
81100 Mytilene, Greece
2
Mediterranean Institute for Advanced Studies, C/Miquel
Marque
`s 21, 07190 Esporles, Balearic Islands, Spain
3
Department of Environment, University of the Aegean,
University Hill, 81100 Mytilene, Greece
123
J Insect Conserv
DOI 10.1007/s10841-016-9868-8

variety of living organisms, from vertebrates to inverte-
brates, plants and bacteria (see reviews in Cucurachi et al.
2013; Balmori 2015).
The majority of studies on the effects of electromagnetic
radiation on insects have been conducted on two model
species: the fruit fly (Drosophila melanogaster) and the
honeybee (Apis mellifera). The studies of electromagnetic
radiation on the fruit fly have mostly shown developmental
delays (Atli and U
¨nlu
¨2006) and negative effects on
reproductive success (e.g. Panagopoulos et al. 2004; Atli
and U
¨nlu
¨2006,2007; Panagopoulos and Margaritis 2010;
Panagopoulos et al. 2010; Chavdoula et al. 2010; but see
Weisbrot et al. 2003 for a positive effect, and Vijver et al.
2013 for no effect) due to DNA fragmentation and repro-
ductive cell death (Chavdoula et al. 2010; Panagopoulos
et al. 2007,2010). In honeybees, radiation decreases col-
ony strength and oviposition rate (Sharma and Kumar
2010; Sahib 2011), and induces worker piping, a behaviour
associated with swarming (Favre 2011). Electromagnetic
radiation also interferes with honeybee navigation, as
honeybees use compass mechanisms for orientation based
on magnetite structures in their bodies (Kirschvink et al.
2001; Wajnberg et al. 2010;Va
´lkova
´and Va
´cha 2012;
Balmori 2015). Due to electromagnetic smog, honeybees
are often unable to return to their hives (Harst et al. 2006;
Favre 2011; Sharma and Kumar 2010; Sahib 2011), and the
resulting massive loss of workers then leads to a colony
collapse (e.g. Harst et al. 2006; Sharma and Kumar 2010),
which is why electromagnetic radiation has been suggested
as one of the potential causes of the colony collapse dis-
order (CCD; e.g. Warnke 2009; Sahib 2011). Fewer studies
exist in the literature regarding other insects, but it is
known that radio frequency magnetic fields disrupt mag-
netoreception in the American cockroach (Va
´cha et al.
2009). In ants, electromagnetic radiation affects visual and
olfactory memory, influencing their ability to associate
cues to food (Cammaerts et al. 2012), as well as locomo-
tion and orientation (Cammaerts et al. 2014; Cammaerts
and Johansson 2014). Although most of these studies are
conducted on model organisms and under laboratory con-
ditions (Cucurachi et al. 2013), current evidence strongly
indicates that insects in natural and protected areas may be
negatively affected by the presence of base-station emitters
of electromagnetic radiation (Balmori 2015). It still
remains unknown, however, whether the presence of
telecommunication antennas can modify wild pollinator
communities by altering species abundance and/or species
richness, thus threatening the maintenance of pollination as
an ecosystem service.
In this study we investigate whether the electromagnetic
radiation emitted by mobile telecommunication antennas
affects the abundance and diversity of wild pollinators, as
suggested by a preliminary study (Tscheulin et al. 2010)
that indicated an effect of distance to the antenna on some
pollinator groups. To do this, we measured the intensity of
electromagnetic radiation at different distances from 10
telecommunication antennas in two Mediterranean islands,
and correlated these measurements with the abundance and
species richness of wild pollinators. Since the susceptibility
of insects to radiation may differ, we hypothesized that
electromagnetic radiation would modify the composition of
the insect community and affect species richness nega-
tively. Our specific questions were: (1) does electromag-
netic radiation affect the abundance and richness of
pollinating insects? And if so, (2) does the effect differ
among taxonomic groups and nesting behaviours?
Materials and methods
Study sites
The study was conducted on two Mediterranean islands
(Lesvos and Limnos) in the north-eastern Aegean (Greece),
characterized by a variety of natural and managed habitats
that support a great diversity of pollinators. In particular,
the bee-friendly habitats the study was carried out in were
phrygana (i.e. low scrub habitats with high pollinator
diversity and abundance; Petanidou and Ellis 1993; Nielsen
et al. 2011), which is dominant on Limnos and to a lesser
degree on Lesvos; and olive groves (i.e. semi-natural
habitats cultivated for centuries using non-intensive meth-
ods; Potts et al. 2006; Nielsen et al. 2011), which are co-
dominant on Lesvos. Both these habitats have been shown
to be equally rich in bee diversity and abundance (Potts
et al. 2006; Nielsen et al. 2011).
In each island, we selected 5 mobile telecommunication
antennas (antennas, hereafter; Fig. 1) as study sites, located
in either phrygranic habitats or cultivated olive groves
(Table 1). All study antennas were mobile telephony base
stations, using frequency bands between 800 and
2,600 MHz, and were located at altitudes below 350 m in
homogeneously flower-rich landscapes, and were separated
by a minimum distance of 5 km. We have no reason to
believe that antennas in the two islands differ in their
emitted frequencies as all providers in Greece use similar
frequency bands. Antennas located at high altitudes or in
coniferous habitats were excluded in order to avoid pro-
nounced differences in pollinator communities among
antennas and sampling points (see below). Each of the
study islands (and all the sites within them) is homoge-
neous in terms of climate and vegetation, and there are no
apparent differences in land use management among sites
(mostly light livestock grazing, beekeeping, and traditional
ploughing of the olive groves).
J Insect Conserv
123

Measurements of electromagnetic radiation
In April 2012, we measured the electrical field E compo-
nent of electromagnetic radiation (EMR, hereafter) at 4
distances (50, 100, 200 and 400 m; sampling points,
hereafter) from each of the 10 study antennas. These
sampling points were selected in order to achieve a large
range of EMR intensities at each study site, based on the
inverse-square law, which states that an isotropic physical
quantity or strength is inversely proportional to the square
of the distance from the source of that physical quantity
(Goldsmith 2005). However, the spatial structure of the
electric field around the base station can be quite complex,
and depend on factors such as topography, antenna vertical
tilt and emission lobes (Balanis 2005; Goldsmith 2005). At
each sampling point we also measured elevation with an
altimeter, to account for differences in elevation with the
distance to the antenna in the statistical analyses.
For the EMR measurements we used the device EMR-
Narda (Narda EMR-300 Broadband RF Survey Meter;
Flu
¨ge 2004) equipped with electric and magnetic probes.
The electric field E probe has a wider frequency spectrum
Fig. 1 Map of the study sites
on Lesvos and Limnos Island
(north-eastern Aegean, Greece)
Table 1 Study sites, respective habitat, coordinates, and mean EMR ±SD (V/m; see text for details)
Island Site Habitat Latitude Longitude Mean EMR
50 m 100 m 200 m 400 m
Lesvos Aghia Paraskevi Olive
groves
39°16021.4000N26°12053.3900E 0.376 ±0.042 0.313 ±0.054 0.670 ±0.015 0.271 ±0.058
Karava Phrygrana 39°15026.2600 N26°23039.0700E 0.385 ±0.012 0.341 ±0.037 0.611 ±0.030 0.632 ±0.041
Nees Kydonies Phrygrana 39°13056.7400N26°26046.0900E 0.516 ±0.121 0.350 ±0.049 0.069 ±0.056 0.010 ±0.004
Parakoila Phrygrana 39°11011.9500N26°8029.5600E 0.236 ±0.112 0.149 ±0.028 0.110 ±0.184 0.118 ±0.095
Pigi Olive
groves
39°10041.1800N26°24040.4600E 0.100 ±0.054 0.077 ±0.034 0.074 ±0.028 0.061 ±0.019
Limnos Aghios
Athanasios
Phrygrana 39°54019.5400N25°4044.6600E 0.512 ±0.134 0.556 ±0.057 0.169 ±0.035 0.358 ±0.085
Moudros I Phrygrana 39°50020.0600N25°18036.3000E 0.083 ±0.086 0.020 ±0.055 0.084 ±0.014 0.067 ±0.012
Moudros II Phrygrana 39°51052.5700N25°17011.6400E 0.139 ±0.015 0.066 ±0.043 0.010 ±0.013 0.057 ±0.030
Plaka Panaghia Phrygrana 39°59025.8000 N25°24049.3000E 0.278 ±0.007 0.288 ±0.014 0.386 ±0.012 0.196 ±0.023
Thanos Phrygrana 39°51034.1900N25°4041.5900E 0.327 ±0.099 0.232 ±0.247 0.132 ±0.014 0.309 ±0.078
Sampling points at 50 and 100 m were within the emission lobes of the antennas, except in Pigi (50 m ?100 m) and Moudros I (50 m)
J Insect Conserv
123

response than the magnetic field B probe, namely
100 kHz–3 GHz and 80 MHz–1 GHz, respectively. In
these frequency ranges, the EMR wavelength is much
smaller than the measurement distances from the antennas,
so one can safely assume that we are measuring in the far-
field and consequently either E or B measurements are
sufficient. We therefore chose to measure the E field, due to
its wider frequency spectrum. E measurements in units of
V/m are hereafter referred to simply as EMR measure-
ments. The aforementioned frequency ranges contain all
frequencies used by mobile telephony services. They also
include radio and television communications, which are not
serviced by the antennas at the locations monitored. In
order to record potential diurnal EMR patterns, EMR was
measured continuously during 24 h at the sites Karava,
Kydonies, and Pigi on Lesvos Island, and at Aghios
Athanasios, on Limnos Island. Following these measure-
ments of EMR, shorter-term measurements (2 h) were
carried out in all the remaining study sites. In total,
between 115 and 1,400 measurements were recorded at
each site and sampling point, each of these measurements
corresponding to a 6-min average of electrical field.
Insect abundance and richness
At each sampling point we collected insects three times
during the main flowering period (April, May and June) in
2012, using the pan trapping method (Westphal et al. 2008;
Nielsen et al. 2011). Samplings were carried out under
good weather conditions, i.e. temperature [15 °C, sun-
shine, low wind speed and relative humidity lower than
65 %. We arranged the pan traps in triplet units, with each
triplet comprising three pan traps of a different UV-bright
colour: blue, yellow, and white. These colours account for
different colour preferences of pollinating insects (West-
phal et al. 2008). Five triplets were set up at each sampling
point of an antenna complex, on the ground, with any two
triplets being separated by ca. 10 m. During each sampling
round, pan traps were filled with 400 ml of water con-
taining 1 drop of aroma-free kitchen detergent to break the
water surface tension, and left for 48 h covering the entire
flight activity period of diurnal and nocturnal pollinators.
Collected insects were transferred to small plastic zip-
lock bags and refrigerated if they were to be processed
within 24 h, otherwise deep frozen until pinning. After
processing in the laboratory, bees and hoverflies were
identified to species level (where necessary with the help of
European specialists), while the remaining insects were
assigned to one of the following taxonomic groups: beetles,
butterflies, wasps, bee flies, and remaining flies that
included all flies not belonging to hoverflies or bee flies
(i.e. flies mainly belonging to the families Anthomyiidae,
Muscidae, Calliphoridae, Nemestrinidae, Empididae,
Tachinidae, Stratiomyidae, Asilidae, Rhagionidae, and
Tabanidae). Honeybees were excluded from the analyses
because they are managed, and their abundance is therefore
biased by beekeepers’ decisions. All identified insects are
deposited in the Melissotheque of the Aegean (Petanidou
et al. 2013).
Pollinator abundance at each sampling point for each
group was estimated as the total number of individuals
collected at that sampling point, pooling the data of the
three sampling rounds together. Similarly, species richness
was calculated as the total number of species collected at
each sampling point during the three sampling rounds (e.g.
Westphal et al. 2008; Nielsen et al. 2011;La
´zaro et al.
2016).
Wild bees were categorized following their nesting
behaviour, i.e. underground versus aboveground nesting,
based on existing literature (Michener 2007;Mu
¨ller 2015),
and our own observations (see Table S1 for categories
used).
Flower cover
Flower cover measurements were conducted in ten
1m91 m squares selected at random at each sampling
point and round, within the area where insects were col-
lected. The total number of functional reproductive units,
i.e., flowers or inflorescences depending on the species
(‘flowers’ hereafter) for each plant species was counted
within these squares. For each sampling point we calcu-
lated flower abundance as total number of flowers recorded
in the ten squares, and flower richness as the total number
of flowering species recorded. Plant specimens were
deposited in the herbarium of the Laboratory of Biogeog-
raphy and Ecology at the University of the Aegean.
Statistical analyses
All the statistical analyses were carried out using general-
ized linear mixed models (GLMM, library lme4) conducted
in R 3.1.2 (R Development Core Team 2008), where site
was included as a random variable to avoid pseudorepli-
cation. To study how EMR varied with the distance to the
antenna, the elevation and the island, we ran a model
(Gamma distribution, log link function) with distance and
elevation as continuous variables, and island as a cate-
gorical predictor variable. Full models on the abundance
and species richness of insects were conducted separately
for each taxonomic group and included EMR, distance,
elevation (log transformed to improve model fitting), total
flower abundance, and flower richness (to control for
variations in the flowering community that could affect the
response) as continuous predictor variables, island as a
fixed predictor, and the interaction between EMR and
J Insect Conserv
123

island, to detect any difference between islands in the
response to EMR. For wild bee abundance and richness, we
ran additional models in which the nesting behaviour of
wild bees was included as a fixed categorical factor (un-
derground vs. aboveground nesting), to test whether this
factor affected the response to EMR. Due to the loss of
some pan traps during sampling, and to control for the
potential effects of these small changes in sampling effort
on the results, we included the number of pan traps (log-
transformed) as an offset in the models (Zuur et al. 2009).
Due to the nature of the data, we used Poisson distributions
and log link functions in these analyses. In some Poisson
models we included an observation-level random intercept
to cope with overdispersion (Zuur et al. 2013). Prior to
analyses, we ran variation inflation factor (VIF) analyses to
identify collinear predictor variables that should be
removed from further analyses (Zuur et al. 2009). VIF
values were smaller than 3 for all variables, thus none of
the predictors needed to be removed (Zuur et al. 2009).
Before the analyses, we also used Moran’s I (library ncf;
Bjørnstad and Falck 2001) to check for any spatial corre-
lation in the data or in the residuals of the models (without
random factor). Neither the data nor the residuals showed
any spatial correlation, indicating that our mixed model
approach was appropriate. Model selection was conducted
using the ‘dredge’ function (package MuMIn; Barton
2014), including EMR in each model, and setting the
maximum number of variables to 4 to avoid over-
parametrization. Means and estimates are accompanied by
their standard error throughout the text.
Results
EMR measurements
We found some diurnal-nocturnal patterns in EMR values
in the sites measured continuously for 24 h (see Figure S1
in Electronic Supplementary Material). However, the pat-
terns differed among sites, the effect of time was small
compared to differences among distances and sites
(Table 1; Figure S2 in Electronic Supplementary Material),
and EMR values did not differ significantly when measured
for only two or for 24 h (v
1
2
=2.53, p=0.112). Conse-
quently, in our models we used the average EMR values
measured per site and sampling point.
Overall, EMR values did not significantly differ between
islands (Lesvos: 0.27 ±0.05 V/m; Limnos:
0.21 ±0.04 V/m; v
3
2
=0.08, p=0.779) and did not
decrease with the distance to the antenna (v
1
2
=0.71,
p=0.398), supporting the results of other studies (e.g.
Vijver et al. 2013), possibly because some of the sampling
points located close to the antenna may have been outside
or at the edge of the emission lobes of the antenna. The
interaction between island and distance was also non-sig-
nificant (v
1
2
=0.95, p=0.329). Mean values (±SD)
recorded at each sampling point are shown in Table 1.
EMR effects on insects
On Lesvos, we collected 11,547 insects in total: 1,334 wild
bees (133 species), 41 hoverflies (9 species), 426 wasps, 75
bee flies, 2,857 other remaining flies, 6,758 beetles, and 84
butterflies. On Limnos, the total number of insects col-
lected amounted to 5,544: 2,467 wild bees (108 species),
155 hoverflies (6 species), 357 wasps, 11 bee flies, 2,263
other remaining flies, 252 beetles, and 131 butterflies.
Table S1 in the Supplementary Material shows the number
of different wild bee and hoverfly species found at each
study site.
EMR effects on the abundance of the different insect guilds
EMR had contrasting effects on the abundance of different
pollinator groups, affecting some positively, while others
negatively. In addition, for most pollinator groups the
effects found were consistent in both islands.
Overall wild bee abundance increased significantly with
EMR in both islands (Fig. 2a); however, the increase was
steeper on Limnos than on Lesvos (Table 2A; Fig. 2a).
When including the nesting behaviour of wild bees in the
analysis, the triple interaction Island 9Nesting beha-
viour 9EMR was also significant (v
1
2
=7.47, p=0.006),
showing that only underground-nesting wild bees were
positively related to EMR, while aboveground-nesting wild
bees were not affected on the study islands (Fig. 3). Again,
the response was steeper in the case of Limnos.
Beetle abundance decreased significantly with EMR in
both islands (Fig. 2b). However, as for wild bees, the
relationship was more pronounced on Limnos than on
Lesvos (Table 2B; Fig. 2b). There were significantly more
beetles on Lesvos than on Limnos (Lesvos: 337.9 ±62.8
cumulated insects per sampling point; Limnos: 12.6 ±6.3;
Table 2B).
The abundance of bee flies increased with EMR
(Table 2C; Fig. 2c), while EMR was negatively related to
the abundance of wasps (marginally non-significant rela-
tionship; Table 2D; Fig. 2d).
Wasp abundance also increased with flower richness
(estimate: 0.140 ±0.051; Table 2D) and decreased with
flower abundance (estimate: -0.008 ±0.004; Table 2D).
Hoverfly abundance was negatively related to EMR
(Table 2E; Fig. 2e) and tended to increase with flower
abundance (estimate: 0.009 ±0.006; Table 2E).
The effect of EMR on the abundance of the broad group
‘remaining flies’ differed drastically between islands. Thus,
J Insect Conserv
123

while EMR was positively related to the abundance of
remaining flies in Lesvos, it was negatively related in
Limnos (Table 2F; Fig. 2f).
Butterfly abundance (Table 2G) was not related to
EMR, but was significantly higher in Limnos than in
Lesvos (Limnos: 6.6 ±1.0 cumulated insects per sampling
point; Lesvos: 4.2 ±0.8; Table 2G).
EMR effects on species richness of wild bees and hoverflies
EMR was not related to overall wild bee species richness
(Table 2H), even when nesting behaviour was included in
the model (Best model: EMR: v
1
2
=0.14, p=0.71;
Nesting behaviour: v
1
2
=705.95, p\0.0001; 2.10 ±1.24
vs. 1.88 ±0.22 mean species/sampling point for under-
ground- and aboveground-nesting wild bees respectively).
The interaction between EMR and island had a mar-
ginally non-significant effect on hoverfly richness
(Table 2I). EMR tended to negatively affect hoverfly spe-
cies richness but only in Limnos, whereas in Lesvos there
was no significant relationship between these two variables
(Fig. 4).
Discussion
For the first time to our knowledge, we have shown that
electromagnetic radiation as typically emitted by
telecommunication antennas affects the abundance and
Fig. 2 Partial residual plots
showing the significant
relationships between EMR and
the abundance of pollinating
insects. When the interaction
between EMR and island was
significant, the relationships are
depicted separately for each
island. The lines represent the
estimate of the best model and
the circles represent partial
residuals. Note that different
graphs vary in y-axis scale
J Insect Conserv
123

composition of wild pollinators in natural habitats. In both
study islands, beetle, wasp and hoverfly abundance
decreased with EMR, whereas, to our surprise, wild bee
and bee fly abundance increased with EMR. The effect of
EMR on the abundance of remaining flies differed between
islands. These differences in the response of insect guilds
are likely due to differences in their susceptibility to
radiation.
We expected telecommunication antennas to affect the
abundance of wild pollinators because these animals rely
on their cognitive and orientation abilities to forage and
navigate (Chittka and Thomson 2001), and electromagnetic
smog is known to affect cognition and orientation in insects
(e.g. Cammaerts et al. 2012,2014; Sharma and Kumar
2010; Sahib 2011). Our communities were similar in terms
of habitat and human activities, and in our analyses we
controlled for the variations in flower abundance and
diversity among study sites (as these variables might
directly affect the pollinator community); therefore,
although we cannot totally discard the potential effect of an
unmeasured variable, we are confident that our results
reflect the effect of radiation on wild pollinator
Table 2 Results of the best models relating EMR to the abundance and richness of insects
Model Variables tested v
2
df p Estimate for relationship with EMR
(A) Wild bee abundance Island 1.79 1 0.181
EMR*Island 7.86 1 0.005 Lim (?), Les (?)
Flower richness 12.97 1 0.0003
(B) Beetle abundance Island 5.49 1 0.019
EMR*Island 85.88 1 \0.0001 Lim (-); Les (-)
(C) Bee fly abundance EMR 6.63 1 0.010 (?)
(D) Wasp abundance EMR 3.78 1 0.051 (-)
Flower richness 7.61 1 0.006
Flower abundance 5.05 1 0.025
(E) Hoverfly abundance EMR 3.98 1 0.046 (-)
Flower abundance 2.86 1 0.091
(F) Remaining flies abundance Island 0.52 1 0.469
EMR*Island 8.33 1 0.004 Lim (-); Les (?)
(G) Butterfly abundance Island 5.59 1 0.018
EMR 0.003 1 0.953 ns
(H) Wild bee richness EMR 0.10 1 0.753 ns
(I) Hoverfly richness Island 1.74 1 0.189
EMR*Island 3.13 1 0.077 Lim (-); Les (ns)
‘?’ a positive, ‘-’ a negative, and ns: a non-significant relationship. When there was a significant interaction between island and EMR the sign
of the estimate is given separately for Limnos (Lim) and Lesvos (Les)
Fig. 3 Partial residual plots
showing the relationship
between EMR and the
abundance of underground- and
aboveground-nesting wild bees.
As the triple interaction
EMR 9Island 9Nesting
behaviour was significant
(p=0.006), the relationships
are depicted separately for each
island. The lines represent the
estimate of the best model and
the circles represent partial
residuals
J Insect Conserv
123

communities, either directly or indirectly. As expected,
EMR modified the abundance of all the studied pollinator
groups, with the only exception of butterflies; however, pan
traps are not efficient tools for collecting butterflies (they
accounted for only 1.3 % of the total specimens collected),
and maybe a more adequate sampling method would have
led to a different result. Interestingly, the effects of EMR
on the abundance of pollinators were not always negative.
More specifically, we found that overall wild bee abun-
dance and bee fly abundance increased significantly with
EMR, while hoverfly, beetle, and wasp abundance
decreased with EMR in both islands. When looking closer
at wild bees, we found that only the abundance of under-
ground-nesting wild bees was positively related to EMR,
whereas wild bees nesting aboveground were not affected
by EMR. It is remarkable though, that the effects of EMR
on insect abundance were consistent in both islands for all
the pollinator groups except for the broad group ‘remaining
flies’, which might have been due to differences in the
composition of this group between the two islands. Previ-
ous studies have also shown that the effects of electro-
magnetic radiation vary for different animals. Within
mammals, negative effects of radiation have been found in
rats and mice (e.g. Maskey et al. 2010,2012; Narayanan
et al. 2015; Sahin et al. 2015), as well as in dogs, cats, and
cows (Marks et al. 1995;Lo
¨scher and Ka
¨s1998). Within
insects, the effects on different species were also variable.
For instance, negative effects of electromagnetic radiation
have been reported on cognition, locomotion, and orien-
tation in the ant Myrmica sabuleti (Cammaerts et al. 2012,
2014), and on worker piping and navigation in honeybees
(e.g. Favre 2011; Sharma and Kumar 2010; Sahib 2011),
and on development and reproductive success of the fruit
fly (e.g. Panagopoulos et al. 2004; Atli and U
¨nlu
¨2007).
Nevertheless, a field study on the effects of electromagnetic
exposure on the reproductive capacity of several insects
revealed no effects (Vijver et al. 2013). The authors sug-
gest that this lack of effects could be at least partially
attributed to short term exposures or to the length of waves
in relation to the small size of the animals (Vijver et al.
2013). Our study, however, revealed alterations in the
structure of the pollinator community. The negative rela-
tionship between EMR and the abundance of wasps, bee-
tles and hoverflies might indicate a high sensitivity of these
insects to electromagnetic radiation. In contrast, pollinators
potentially more tolerant to EMR, such as underground-
nesting wild bees and bee flies, may fill the vacant niches
left by less tolerant species and thus result in an increase of
their populations. One possible explanation for these
results is that EMR may have particularly detrimental
effects on the potentially more susceptible larval stages of
these flower visitors. If so, larvae developing aboveground
(many beetles, wasps, many hoverflies) may be more vul-
nerable than those developing underground (underground-
nesting wild bees, i.e. the majority of Apoidea), because
the former may be exposed to higher radiation levels. The
fact that bee flies are mostly represented in our study by the
genera Phthiria and Cyllenia, which are parasitoids of
prepupal or pupal stages of moths and sawflies developing
underground (Yeates and Greathead 1997), may also sup-
port our hypothesis. At any rate, these changes in the
composition of pollinator communities may have important
ecological consequences for the maintenance of pollination
services and biodiversity, but also economic impacts in
respect to potential effects on agricultural productivity
(Gallai et al. 2009; Potts et al. 2010).
Reduced pollinator diversity in a community can reflect
disturbances of the environment (Potts et al. 2010), and
thus, we expected EMR to negatively affect overall polli-
nator species richness. However, our results on pollinator
species richness were weak and not as conclusive as those
on abundance, because we found no significant effect of
EMR on wild bee species richness, and only a marginally
non-significant negative effect on hoverfly species richness
in Limnos, the study island where the overall effects of
EMR were stronger. At any rate, the tendencies were
always negative, and therefore, we cannot rule out that an
increase in sampling effort in future research could reveal
stronger effects of EMR on species richness.
Conclusions
Electromagnetic radiation from telecommunication anten-
nas affected the abundance and composition of wild pol-
linators in natural habitats. EMR had contrasting effects on
Fig. 4 Partial residual plots showing the significant relationships
between EMR and the hoverfly richness. Since the interaction between
EMR and island was marginally non-significant, the relationship is
depicted separately for each island. The lines represent the estimate of
the best model and the circles represent partial residuals
J Insect Conserv
123

the abundance of different pollinator groups (negative on
beetles, wasps, and hoverflies; positive on underground-
nesting wild bees and bee flies), which may be due to
different susceptibilities to radiation particularly of the
larval stages. Pollinators and their host plants constitute
pollination networks. Although the architecture of these
mutualistic networks can increase the capacity of pollinator
populations to persist under harsh conditions, once a tip-
ping point in human-induced environmental change is
reached, pollinator populations may collapse simultane-
ously (Lever et al. 2014). Therefore, these changes in the
composition of pollinator communities associated with
electromagnetic smog may have important ecological and
economic impacts on the pollination service that could
significantly affect the maintenance of wild plant diversity,
crop production and human welfare.
Acknowledgments The research has been co-financed by the Euro-
pean Union (European Social Fund—ESF) and Greek national funds
through the Operational Program ‘‘Education and Lifelong Learning’’
of the National Strategic Reference Framework (NSRF)—Research
Funding Program: THALES: Investing in knowledge society through
the European Social Fund. We would like to thank W. Arens, H.
Dathe, J. Dils, A. Ebmer, M. Kuhlmann, V. Mizerakis, A. Mueller, A.
Pauly, C. Praz, M. Quaranta, S. Risch, W. Schedl, E. Scheuchl, M.
Schwarz, M. Terzo and A. Vujic for insect identification.
References
Aguilar R, Ashworth L, Galetto L, Aizen M (2006) Plant reproductive
susceptibility to habitat fragmentation: review and synthesis
through a meta-analysis. Ecol Lett 9:968–980
Ashman T-L, Knight TM, Steets JA, Amarasekare P, Burd M,
Campbell DR, Dudash MR, Johnston MO, Mazer SJ, Mitchell
RJ, Morgan MT, Wilson WG (2004) Pollen limitation of plant
reproduction: ecological and evolutionary causes and conse-
quences. Ecology 85:2408–2421
Atli E, U
¨nlu
¨H (2006) The effects of microwave frequency
electromagnetic fields on the development of Drosophila
melanogaster. Int J Radiat Biol 82:435–441
Atli E, U
¨nlu
¨H (2007) The effects of microwave frequency
electromagnetic fields on the fecundity of Drosophila melano-
gaster. Turkish J Biol 31:1–5
Balanis C (2005) Antenna theory: analysis and design, 3rd edn.
Wiley, New York
Balmori A (2015) Anthropogenic radiofrequency electromagnetic
fields as an emerging threat to wildlife orientation. Sci Total
Environ 518–519:58–60
Barton K (2014) MuMIn: Multi-model inference. R package version
1.10.0. Retrieved May 14, 2014, from http://cran.r-project.org/
package=MuMIn
Biesmeijer JC, Roberts SPM, Reemer M, Ohlemu
¨ller R, Edwards M,
Peeters T, Schaffers AP, Potts SG, Kleukers R, Thomas CD,
Settele J, Kunin WE (2006) Parallel declines in pollinators and
insect-pollinated plants in Britain and the Netherlands. Science
313:351–354
Bjørnstad ON, Falck W (2001) Non-parametric spatial covariance
functions: estimation and testing. Environ Ecol Stat 8:53–70
Burkle LA, Marlin JC, Knight TM (2013) Plant-pollinator interac-
tions over 120 years: loss of species, co-occurrence and function.
Science 339:1611–1615
Cammaerts M-C, Johansson O (2014) Ants can be used as bio-
indicators to reveal biological effects of electromagnetic waves
from some wireless apparatus. Electromagn Biol Med
33:282–288
Cammaerts M-C, De Doncker P, Patris X, Bellens F, Rachidi Z,
Cammaerts D (2012) GSM 900 MHz radiation inhibits ants’
association between food sites and encountered cues. Electro-
magn Biol Med 31:151–165
Cammaerts M-C, Vandenbosch GAE, Volski V (2014) Effect of
short-term GSM radiation at representative levels in society on a
biological model: the ant Myrmica sabuleti. J Insect Behav
27:514–526
Chavdoula ED, Panagopoulus DJ, Margaritis LH (2010) Comparison
of biological effects between continuous and intermittent
exposure to GSM-900-MHz mobile phone radiation: detection
of apoptotic cell-death features. Mutat Res 700:51–61
Chittka L, Thomson JD (2001) Cognitive ecology of pollination.
Cambridge University Press, Cambridge
Cucurachi S, Tamis WLM, Vijver MG, Peijnenburg WJGM, Bolte
JFB, de Snoo GR (2013) A review of the ecological effects of
radiofrequency electromagnetic fields (RF-EMF). Environ Int
51:116–140
Dasdag S, Akdag MZ, Aksen F, Bashan M, Buyukbayram H (2004)
Does 900 MHZ GSM mobile phone exposure affect rat brain?
Electromagn Biol Med 23(3):201–214
Favre D (2011) Mobile phone-induced honeybee worker piping.
Apidologie 42:270–279
Flu
¨ge S (2004) EMR-TS PC Transfer Set Narda Safety Test Solutions
GmbH. BN 2244/90.36. Version 1.02 ff. http://www.narda-sts.
us/software/Narda_EMR.zip
Gallai N, Salles JM, Settele J, Vaissie
`re BE (2009) Economic
valuation of the vulnerability of world agriculture confronted
with pollinator decline. Ecol Econ 68:810–821
Goldsmith A (2005) Wireless communications. Cambridge University
Press, Cambridge
Goulson D, Lye GC, Darvill B (2008) Decline and conservation of
bumble bees. Annu Rev Entomol 53:191–208
Harst W, Kuhn J, Stever H (2006) Can electromagnetic exposure
cause a change in behaviour? Studying possible non-thermal
influences on honeybees—an approach within the framework of
educational informatics. Acta Syst IIAS Int J 6(1):1–6
Hegland SJ, Nielsen A, La
´zaro A, Bjerknes AL, Totland Ø (2009)
How does climate warming affect plant–pollinator interactions?
Ecol Lett 12:184–195
Kearns C, Inouye D, Waser N (1998) Endangered mutualisms: the
conservation of plant–pollinator interactions. Annu Rev Ecol
Syst 29:83–112
Kirschvink JL, Walker MM, Diebel C (2001) Magnetite-based
magnetoreception. Curr Opin Neurobiol 11:462–467
Klein AM, Vaissie
`re BE, Cane JH, Dewenter IS, Cunningham SA,
Kremen C, Tscharntke T (2007) Importance of pollinators in
changing landscapes for world crops. Proc R Soc Lond B Biol
Sci 274:303–313
Kremen C, Williams NM, Aizen MA, Gemmill-Herren B, LeBuhn G,
Minckley R, Packer L, Potts SG, Roulston T, Steffan-Dewenter
I, Va
´zquez DP, Winfree R, Adams L, Crone EE, Greenleaf SS,
Keitt TH, Klein AM, Regetz J, Ricketts TH (2007) Pollination
and other ecosystem services produced by mobile organisms: a
conceptual framework for the effects of land-use change. Ecol
Lett 10:299–314
La
´zaro A, Tscheulin T, Devalez J, Nakas G, Petanidou T (2016)
Effects of grazing intensity on pollinator abundance and
J Insect Conserv
123

diversity, and on pollination services. Ecol Entom. doi:10.1111/
een.12310
Lever JJ, van Nes EH, Scheffer M, Bascompte J (2014) The sudden
collapse of pollinator communities. Ecol Lett 17:350–359
Lo
¨scher W, Ka
¨s G (1998) Behavioural abnormalities in a dairy cow
herd near a TV and radio transmitting antenna. Prakt Tierarzt
79:437
Marks TA, Ratke CC, English WO (1995) Stray voltage and
developmental, reproductive and other toxicology problems in
dogs, cats and cows—a discussion. Vet Hum Toxicol
37:163–172
Maskey D, Kim M, Aryal B, Pradhan J, Choi I-Y, Park K-S, Son T,
Hong S-Y, Kim SB, Kim HG, Kim MJ (2010) Effect of
835 MHz radiofrequency radiation exposure on calcium binding
proteins in the hippocampus of the mouse brain. Brain Res
1313:232–241
Maskey D, Kim H-J, Kim HG, Kim MJ (2012) Calcium-binding
proteins and GFAP immunoreactivity alterations in murine
hippocampus after 1 month of exposure to 835 MHz radiofre-
quency at SAR values of 1.6 and 4.0 W/kg. Neurosci Lett
506:292–296
Michener CD (2007) The bees of the world, 2nd edn. John Hopkins
University Press, Baltimore
Minckley R, Packer L, Potts SG, Roulston T, Steffan-Dewenter I,
Va
´zquez DP, Winfree R, Adams L, Crone EE, Greenleaf SS,
Keitt TH, Klein A-M, Regetz J, Ricketts TH (2007) Pollination
and other ecosystem services produced by mobile organisms: a
conceptual framework for the effects of land-use change. Ecol
Lett 10:299–314
Mu
¨ller A (2015) Palaearctic osmiine bees. ETH Zu
¨rich. http://blogs.
ethz.ch/osmiini
Narayanan SN, Kumar RS, Karun KM, Nayak SB, Bhat PG (2015)
Possible cause for altered spatial cognition of prepubescent rats
exposed to chronic radiofrequency electromagnetic radiation.
Metab Brain Dis 30(5):1193–1206
Nielsen A, Steffan-Dewenter I, Westphal C, Messinger O, Potts SG,
Roberts SPM, Settele J, Szentgyorgyi H, Vaissie
`re BE, Vaitis M,
Woyciechowski M, Bazos I, Biesmeijer JC, Bommarco R, Kunin
WE, Tscheulin T, Lamborn E, Petanidou T (2011) Assessing bee
species richness in two Mediterranean communities: importance
of habitat type and sampling techniques. Ecol Res 26:969–983
Panagopoulos DJ, Margaritis LH (2010) The effect of exposure
duration on the biological activity of mobile telephony radiation.
Mutat Res 699:17–22
Panagopoulos DJ, Karabarbounis A, Margaritis LH (2004) Effect of
GSM 900-MHz mobile phone radiation on the reproductive
capacity of Drosophila melanogaster. Electromagn Biol Med
23:29–43
Panagopoulos DJ, Chavdoula ED, Nezis IP, Margaritis LH (2007)
Cell death induced by GSM 900-MHz and DCS 1800-MHz
mobile telephony radiation. Mutat Res 626:69–78
Panagopoulos DJ, Chavdoula ED, Margaritis LH (2010) Bioeffects of
mobile telephony radiation in relation to its intensity or distance
from the antenna. Int J Radiat Biol 86:345–357
Pauw A (2007) Collapse of a pollination web in small conservation
areas. Ecology 88:1759–1769
Petanidou T, Ellis WN (1993) Pollinating fauna of a phryganic
ecosystem: composition and diversity. Biodiv Lett 1:9–22
Petanidou T, Sta
˚hls G, Vujic
´A, Olesen JM, Rojo S, Thrasyvoulou A,
Sgardelis S, Kallimanis AS, Kokkini S, Tscheulin T (2013)
Investigating plant–pollinator relationships in the Aegean: the
approaches of the project POL-AEGIS (The Pollinators of the
Aegean Archipelago: Diversity and Threats). J Apicult Res
52:106–117
Potts SG, Petanidou T, Roberts S, O’Toole C, Hulbert A, Willmer P
(2006) Plant-pollinator biodiversity and pollination services in a
complex Mediterranean landscape. Biol Conserv 129:519–529
Potts SG, Biesmeijer JC, Kremen C, Neumann P, Schweiger O, Kunin
WE (2010) Global pollinator declines: trends, impacts and
drivers. Trends Ecol Evol 25:345–353
R Development Core Team (2008) R: a language and environment for
statistical computing. R Foundation for Statistical Computing,
Vienna, Austria. ISBN 3-900051-07-0. http://www.R-project.org
Sahib SS (2011) Impact of mobile phones on the density of
honeybees. J Public Adm Policy Res 3:131–137
Sahin A, Aslan A, Bas O, Ikinci A, O
¨zyilmaz C, So
¨nmez OF,
C¸ olakoglu OE (2015) Deleterious impacts of a 900-MHz
electromagnetic field on hippocampal pyramidal neurons of
8-week-old Sprague Dawley male rats. Brain Res 1624:232–238
Sharma VP, Kumar NR (2010) Changes in honeybee behaviour and
biology under the influence of cellphone radiations. Curr Sci
98:1376–1378
Tscheulin T, Spyropoulos A, Petanidou T (2010) Impacts of mobile
phone masts on the abundance of pollinators. In: 5th Conference
of the Hellenic Ecological Society, Patras, p 201
Va
´cha, M, Pu
˚z
ˇova
´T, Kvı
´c
ˇalova
´M (2009) Radio-frequency magnetic
fields disrupt magnetoreception in American cockroach. J Exp
Biol 212:3473–3477
Va
´lkova
´T, Va
´cha M (2012) How do honeybees use their magnetic
compas? Can they see the North? Bull Entomol Res
102:461–467
Vijver MG, Bolte JFB, Evans TR, Tamis WLM, Peijnenburg WJGM,
Musters CJM, de Snoo GR (2013) Investigating short-term
exposure to electromagnetic fields on reproductive capacity of
invertebrates in the field situation. Electromagn Biol Med
33(1):21–28
Wajnberg E, Acosta-Avalos D, Alves OC, de Oliviera JF, Srygley
RB, Esquivel DM (2010) Magnetoreception in eusocial insects:
an update. J R Soc Interface 7:S207–S225
Warnke U (2009) Bees, birds and mankind: destroying nature by
‘Electrosmog’ effects of wireless communication technologies.
A Brochure series by the competence initiative for the protection
of humanity, environment and democracy, Kempten, 1st edn,
November 2007, ISBN: 978-3-00-023124-7, English edn, March
2009, pp 14–33
Weisbrot D, Lin H, Ye L, Blank M, Goodman R (2003) Effects of
mobile phone radiation on reproduction and development in
Drosophila melanogaster. J Cell Biochem 89:48–55
Westphal C, Bommarco R, Carre G, Lamborn E, Morison N,
Petanidou T, Potts SG, Roberts SPM, Szentgyorgyi H, Tscheulin
T, Vaissie
`re BE, Woyciechowski M, Biesmeijer JC, Kunin WE,
Settele J, Steffan-Dewenter I (2008) Measuring bee biodiversity
in different European habitats and biogeographical regions. Ecol
Monogr 78:653–671
Yeates DK, Greathead DJ (1997) The evolutionary pattern of host use
in the Bombyliidae (Diptera): a diverse family of parasitoid flies.
Biol J Linn Soc 60:149–185
Zuur AF, Ieno EN, Walker NJ, Saveliev AA, Smith GM (2009)
Mixed effects models and extensions in ecology with R.
Springer, New York
Zuur AF, Hilbe JM, Ieno EN (2013) Beginner’s guide to GLM and
GLMM with R. Highland Statistics, Newburgh
J Insect Conserv
123