Content uploaded by Eric Darrouzet
Author content
All content in this area was uploaded by Eric Darrouzet on Sep 20, 2016
Content may be subject to copyright.
Accepted Article
This article has been accepted for publication and undergone full peer review but has not
been through the copyediting, typesetting, pagination and proofreading process, which may
lead to differences between this version and the Version of Record. Please cite this article as
doi: 10.1111/1365-2664.12724
This article is protected by copyright. All rights reserved.
Received Date : 27-Nov-2015
Revised Date : 07-Apr-2016
Accepted Date : 16-Jun-2016
Article type : Standard Paper
Rapid spread of the invasive yellow-legged hornet in France:
the role of human-mediated dispersal and the effects of control measures
Christelle ROBINET1, Christelle SUPPO2, Eric DARROUZET2
1INRA, UR633 Zoologie Forestière, F-45075 Orléans, France
2Institut de Recherche sur la Biologie de l’Insecte, UMR 7261, CNRS – Université François-Rabelais de
Tours, F-37200 Tours, France
*Corresponding author:
Christelle Robinet
INRA, UR633 Zoologie Forestière, 2163 avenue de la Pomme de Pin, CS 40001 Ardon, F-45075
Orléans cedex 2, France
Email: christelle.robinet@orleans.inra.fr
Tel.: +33 2 38 41 80 38; Fax: +33 2 38 41 78 79
Running title: Rapid spread of invasive yellow-legged hornet
Accepted Article
This article is protected by copyright. All rights reserved.
Summary
1. The invasive yellow-legged hornet was first discovered in Europe, in south-western France, in
2004. It has since spread very rapidly and has caused significant mortality among honey bees and
native entomofauna. It also poses a risk to humans because its sting provokes allergic reactions. The
objectives of this study were the following: (i) to disentangle the roles played by human-mediated
dispersal and self-mediated dispersal in the species’ rapid range expansion and (ii) to estimate the
intensity of control measures in France and determine what needs to be done to slow the hornet’s
spread and dramatically reduce its population densities.
2. A mathematical model was developed to describe the hornet’s potential spread. This model
included parameters describing the population growth rate, carrying capacity, self-mediated
dispersal, human-mediated dispersal and the efficacy of control measures (i.e. the destruction of
detected nests). Model parameters were estimated using 2004–2009 occurrence data for France and
the model was then validated using 2013 occurrence data. Several scenarios were tested: human-
mediated dispersal was present or absent and control intensity varied. Then, the species’ spread in
coming years was simulated (from 2013 to 2020).
3. Despite some uncertainty on the value of the parameters, this model is relatively robust. Human-
mediated dispersal may not be necessarily responsible for the hornet’s rapid range expansion; the
species could spread rapidly on its own. It is likely that, to date, an average of 30–40% of detected
nests have been destroyed each year.
4. Increasing the percentage of destroyed nests from 30 to 60% could reduce the species’ spread by
17% and its nest density by 29%. If 95% of nests are destroyed, the species’ spread and nest density
could decline by 43% and 53%, respectively.
5. Synthesis and applications. The mathematical model developed in this study shows that human-
mediated dispersal of the invasive yellow-legged hornet may not be the only factor explaining the
hornet’s rapid range expansion and that controlling this invasive pest is still possible. Therefore,
there is an urgent need to reduce self-mediated dispersal and to intensify and improve control
measures to diminish the species’ impact and prevent its further spread. Control measures could
combine the mechanical removal and destruction of individuals or infested materials with biological
control techniques.
Accepted Article
This article is protected by copyright. All rights reserved.
Key-words:
biological invasion, CLIMEX, insect, dispersal, long distance, nest, reaction-diffusion model, pest
management, spread model, Vespa velutina
Introduction
The number of biological invasions has grown considerably over recent decades, mainly due to the
increased frequency and facility of human travel and international trade (e.g., Levine & D’Antonio
2003; Evans & Oszako 2007). Terrestrial invertebrate species account for a large proportion of the
invasions that have occurred in Europe (Hulme et al. 2009). Most such species come from Asia and
are largely found in France and Italy (Roques et al. 2009). Social insects are highly invasive because
they can easily adapt their reproductive modes and social structures to match new environments
(Beggs et al. 2011; Arca et al. 2015). For instance, there have been a large number of introductions
of Vespidae species across the world (n = 34) but only some (n = 9), all of them eusocial, have been
found to have negative ecological effects (Beggs et al. 2011).
The yellow-legged hornet, Vespa velutina, is the first exotic Vespidae species to successfully
arrive and establish itself in Europe (Beggs et al. 2011). It was first discovered in southwestern
France in the summer of 2004, and the first individual hornet was caught and identified in 2005
(Haxaire et al. 2006; Villemant et al. 2006). It was probably introduced via bonsai pots imported
from China and, according to genetic analyses, likely came from the eastern Chinese provinces of
Jiangsu and Zhejiang (Arca et al. 2015). It has since very rapidly expanded its range in France and is
now spreading to neighboring European countries. It is found throughout most of France (more than
2/3 of the country has been colonized; Rome et al. 2013); it is also present in Spain (López et al.
2011), Portugal (Grosso-Silva & Maia 2012), Italy (Federazione Apicoltori Italiani 2013), and Germany
(Rome et al. 2015). While a single male was observed in Belgium in 2011, very close to the French
border (Rome et al. 2012), the species does not seem to have established itself in that country yet
(pers. com., Dr. J.C. Grégoire, Université Libre de Bruxelles, Belgium, in 2014). A large proportion of
Europe, including France, seems to be suitable for the species’ establishment (Villemant et al.
2011a), and an even larger area could become suitable in the future given various climate change
scenarios (Barbet-Massin et al. 2013). To more accurately assess the potential spread of an invasive
species, it is important to consider the species’ dispersal abilities. Along with environmental
heterogeneity, dispersal ability can shape a species’ distribution and affect population growth
(Roques et al. 2008). In France, the yellow-legged hornet spreads at a rate of around 100 km/year,
Accepted Article
This article is protected by copyright. All rights reserved.
and several nests have been found more than 200 km from the invasion front (Villemant et al.
2011a; Rome et al. 2012). Three locations at which infestations have been observed are thought to
have been colonized as a result of long-distance dispersal events (Villemant et al. 2011a). It is
common to find pioneer colonies of invasive species at such large distances from the invasion front
because of accidental transport by humans. Indeed, in models of pest spread, human population
density is a key factor that can explain such jumps (e.g., see the European gypsy moth, Sharov &
Liebhold 1998; the horse-chestnut leaf miner, Gilbert et al. 2004; the emerald ash borer, Muirhead
et al. 2006; the pine wood nematode, Robinet et al. 2009; the pine processionary moth, Robinet et
al. 2012). However, in case of the yellow-legged hornet, distinguishing between dispersal patterns
caused by self-mediated flight versus those caused by human-mediated transport is challenging
because the hornet can fly very long distances. Gynes—females destined to become queens and
found new colonies—can fly an average of 18 km/day and can cover up to 200 km over 10 days in
flight mill experiments (pers. com., Dr. Daniel Sauvard, INRA, France). Although yellow-legged hornet
gynes might not have the same flight capabilities in nature, they may cover very long distances over
their extensive flight periods (i.e., before and after hibernation).
At present, France is clearly the most infested country in Europe. Although the full impact of
the yellow-legged hornet is not yet known, some concerns are hornet predation of domestic honey
bees (Apis mellifera) and local entomofauna and the allergic reactions caused by the hornet’s sting
(Monceau et al. 2014). As soon as the yellow-legged hornet was detected in France, beekeepers,
among others, became worried about the possible impact the species could have on honey bees and
tried to eliminate it by removing and destroying its nests. However, the overall control intensity and
efficiency (e.g., the effects on population size over time) of such control efforts remain unclear. Due
to the hornet’s rapid range expansion in France and the threat it poses, there is an increasing
demand for effective control measures. Even if traps and other control methods are available, they
remain inefficient and do not exclusively target the yellow-legged hornet. Mass trapping using food
baits could have negative impacts on native entomofauna, and it is uncertain how much it would
truly affect the yellow-legged hornet (Rome et al. 2011; EEA 2012).
Consequently, the objectives of this study were the following: 1) to disentangle the roles
played by human-mediated dispersal and self-mediated dispersal in the rapid range expansion of the
yellow-legged hornet in France, and 2) to estimate the intensity of control measures in France and
determine what needs to be done to slow the hornet’s spread and dramatically reduce its
population densities. To this end, a spread model was developed and various scenarios were tested.
Accepted Article
This article is protected by copyright. All rights reserved.
Materials and methods
STUDY SPECIES
The yellow-legged hornet produces one generation a year. After hibernation, fertilized queens found
new colonies. They build relatively large nests that are around 60–80 cm in diameter and 60–100 cm
in length and that can be easily detected (Darrouzet 2013; Rome et al. 2015). They are usually
located at the tops of trees but can also be found in bushes, in open buildings, or on walls. Since an
object of 1 cm x 1 cm can easily be seen from 1 m away and given that nests are generally at least 50
cm x 50 cm, according to the intersect theorem, it should be possible to detect the hornet’s nests
from up to 50 m away. This detection distance is also the one used when conducting surveys for
animals such as rabbits (Mutze et al. 1998).
Yellow-legged hornets catch a large range of insect species, but especially domestic honey
bees (37%), to feed their larvae (Villemant et al. 2011b). Adults mainly consume the sweet liquids
and proteinaceous juice produced by the larvae. Workers also gather a variety of suitable materials
for nest building and nest maintenance (i.e., plants and water). Because yellow-legged hornets can
disperse over long distances, exploit a wide range of prey species, and build/maintain their nests
with readily available materials, they likely face few limiting factors as they spread. In addition, they
compete very little with other Vespa species (Arca et al. 2015).
SPECIES OCCURRENCE DATASETS USED IN PARAMETER ESTIMATION AND MODEL VALIDATION
It is difficult to monitor the rapid spread of an invasive species across broad spatial scales. In the
case of the invasive yellow-legged hornet in France, this is made in the form of a citizen science
project organized by the French Museum of Natural History (MNHN, Paris). General information on
the yellow-legged hornet and also a guide that can be used to distinguish it from other species is
provided on a special webpage created by the museum (http://frelonasiatique.mnhn.fr/). Anyone
can report a species sighting by filling out a form indicating the precise location of the sighting,
describing what they observed, and sending either a photograph (of the nest or insect) or sending a
dried insect by regular mail. The MNHN then confirms the species’ identity, and the record is
entered into a database.
The MNHN provided the 2004–2009 occurrence data used in this study (e.g., number of
nests, nest spatial coordinates, and the year of first observation; see maps in the Supplementary
Material of Villemant et al. 2011a). A total of 4,802 nests were observed between 2004 and 2009 (2
Accepted Article
This article is protected by copyright. All rights reserved.
in 2004, 4 in 2005, 257 in 2006, 1,670 in 2007, 1,233 in 2008, and 1,636 in 2009). This dataset was
used to estimate the parameters of the spread model (see below).
The map of the species’ occurrence in France in 2013 was retrieved from the webpage
mentioned above; a total of 1,003 grid cells, each representing an area of 10 km x 10 km, were
marked as having been invaded. Since occurrence data for the period between 2010 and 2012 were
not available, this second dataset was used to independently validate the results of the model for
2013.
LOCAL DISPERSAL AND POPULATION GROWTH
A reaction-diffusion model was used to simulate local spread of the yellow-legged hornet. This
model is commonly used to describe the spatial spread of invasive species (e.g., Shigesada &
Kawasaki 1997). The reaction-diffusion model describes not only population dispersal but also
population growth. The model can be expressed using the Fisher equation:
=
² +
²+1−
Equation 1
where N is nest density (km-²) and depends on time t and spatial location (x,y); D is the diffusion
coefficient (km² year-1); r is growth rate (year
-1); and K is carrying capacity (km-²). This model
generates a travelling wave that has a constant speed defined by:
=2
√ (km year-1) Equation 2
The parameters c, K, and r were estimated and D was then calculated using those values and
Equation 2. Some calculations (see below) were performed at the departmental level: in France, the
department is an administrative division that corresponds to the European classification level NUTS3
(http://ec.europa.eu/eurostat/web/nuts/overview).
First, the local rate of spread, c, was estimated by measuring the radial distances between
the range edge observed in consecutive years from 2004 to 2009 in southwestern France; the three
locations at which hornet presence has been attributed to human-mediated dispersal were excluded
(Villemant et al. 2011a; Rome et al. 2012).
Then, carrying capacity, K, was given by the highest nest density in the most infested
department of France, which was where the yellow-legged hornet was initially discovered, in the
southwest. This value was assumed to represent what the environment can sustain. Since plant
Accepted Article
This article is protected by copyright. All rights reserved.
resources used in nest building (Monceau et al. 2014) and prey are found throughout France, K was
considered to be homogeneous across the country. Although nests are usually found close to rivers,
this factor was not incorporated into the model because its level of spatial resolution was too coarse
(i.e., 10 km x 10 km; see the model description below).
Finally, growth rate, r, was estimated. Population growth (based on survival and
reproduction) may not be homogeneous across France because it usually depends on climatic
conditions, and more specifically temperature. Growth index values (GI, varying from 0 to 100) were
extracted from a CLIMEX model previously developed for the yellow-legged hornet (Sutherst et al.
2007; Ibáñez-Justicia & Loomans 2011). This index describes the growth potential of a population
during favorable conditions and depends on temperature, moisture, and solar radiation. To obtain a
spatial estimate of r, we performed a linear regression between GI and r. For each infested
department and for each year i, we calculated the annual growth rate:
=
Equation 3
where is the number of nests found in year i in the department. Then, we took the mean value of
these growth rates over the exponential section of the population growth curve (i.e., before it
reaches K/2). For each department, we calculated the mean of GI and performed the regression.
Then, we used the regression equation and the GI values to estimate r for all of France. These values
were then used in the logistic growth (see Equation 1).
The diffusion coefficient, D, was assumed to be homogeneous across France because of the
fast spread rate (and thus likely few local barriers) and no evidence of a spatial heterogeneity in
terms of invaded areas per unit of time. However, since r and c likely vary, D was estimated using the
values of r and c for southwestern France (the Aquitaine region). First, we calculated the mean value
of r using the r values that had been estimated for each department in southwestern France.
Second, we calculated D for each department using Equation 2 and the previously determined value
of c. Finally, we calculated the mean value of D across all these departments.
Based on the results of bioclimatic envelope models (Ibáñez-Justicia & Loomans 2011;
Villemant et al. 2011a), most of France is suitable for the establishment of the yellow-legged hornet
and the least suitable areas in Europe seem to be mainly located in mountainous regions (Péré &
Kenis 2011). Therefore, we defined an elevation limit E above which the yellow-legged hornet should
not be able to establish itself; we used the highest elevation at which it was found between 2004
and 2009 (dataset used for parameter estimates). Climate change could possibly affect the
Accepted Article
This article is protected by copyright. All rights reserved.
elevational threshold. However, in this study, we did not explore this possibility as the hornet’s
spread is relatively fast compared to the effects of climate change.
LONG-DISTANCE DISPERSAL
Points where hornets occurred at distances greater than the local dispersal distance from their
nearest neighbors were identified. In addition, a subset of points with the highest dispersal distances
which fell within the same range than the three points thought to result from long-distance dispersal
(according to Villemant et al. 2011a) was also identified. For all these locations, the effects of human
population density (retrieved from CIESIN & CIAT 2005) on long-distance dispersal were analyzed.
We obtained estimates of human density at the points and compared those values with the values
associated with 100 randomly chosen points found both within habitat favorable to hornets (where
elevation < E) and outside the short-distance dispersal range. Since the data were not normally
distributed, a Wilcoxon test was used to compare human population densities between the two sets
of points (Millot 2011). A threshold value for human population density (H) above which long-
distance dispersal events could be observed was defined. This threshold distinguished the long-
distance dispersal points from the randomly chosen points.
Then, to simulate human-mediated dispersal, we defined a yearly probability of observing a
long-distance dispersal event, denoted as Pldd, which was based on the number of occurrences
observed every year between 2004 and 2009 that likely resulted from human-mediated dispersal.
Then, for each event, we randomly chose a nest founding point within the area where human
population density and elevation were suitable (above H and below E). Since the model was
stochastic, we ran 100 replicate simulations to obtain result convergence (see Appendix S1 in
Supplementary Information).
CONTROL MEASURES
Control measures (e.g., nest removal and destruction) may affect the range expansion of the yellow-
legged hornet. Their effects in the form of nest density reduction were included in the model for two
reasons: 1) to determine the percentage of nests removed and destroyed in the past in France and
2) to explore the possible effects in the future if control efforts increased. Since some control
measures have been applied since the species was detected in France, the model had to account for
them both in the model calibration and validation.
Accepted Article
This article is protected by copyright. All rights reserved.
Control measures are likely not uniformly applied across France because their application
largely depends on the likelihood of encountering nests. The nests can be detected by chance or
when yellow-legged hornets are reported and then the nests are searched for destruction.
Beekeepers face some dramatic problems in rural areas but anyone living in urban areas can also
face important problems because the yellow-legged hornet also occurs there, it can be aggressive
toward humans and its colony size is relatively large (Choi et al. 2012). Since there is a high number
of potential observers and a high number of people likely to get stung in urban areas, we assumed
that the probability of applying control measures is greater in these areas.
In the model, we assumed that people can detect a yellow-legged hornet nest from within a
50-m radius (d = 0.05 km). We then defined the parameter ρ as the proportion of a given area that
one person can observe, where ρ = (π × d²)/A. For a cell of 10 km x 10 km (A = 100 km²), the area
observed is π × d² = 7.85 10-3 km². Therefore ρ = 7.85 10-5 for that cell. The intensity of control
measures, C(h), is given by (Figure 1):
ℎ=
Equation 4
where h is the number of humans associated with the cell (i.e., human population density) and α is
the maximal control intensity if all nests were detected (0 ≤
α
≤ 1, where α = 0 means that no nest is
destroyed and α = 1 means that all the detected nests are destroyed). As human population
densities increase, C(h) approaches α. Control intensity describes the proportion of nests that are
destroyed. Although actual control efforts are more complex, this simple function based on the
detection radius and human population density likely captures the main pattern that explains
heterogeneity in attempts to control the yellow-legged hornet.
MODEL DESCRIPTION, VALIDATION, UNCERTAINTY ANALYSIS AND PREDICTIONS
We combined estimates of local spread and growth with estimates of long-distance dispersal and
control intensity. We applied these models to a grid that covered France with a spatial resolution of
10 km x 10 km; implementation took place in R (R Development Core Team 2014; Robinet et al.
2016; see also Appendix S1). The simulations began with the two nests that were present in 2004 at
the location where the hornet first appeared in France (Figure 2).
To validate the model and identify the most likely dispersal and control scenarios, the
species’ spread between 2004 and 2013 was simulated with and without long-distance dispersal
Accepted Article
This article is protected by copyright. All rights reserved.
events (representing human-mediated dispersal). Different levels of control intensity were also used;
α ranged from 0 to 1 and varied by increments of 0.1. The simulated results obtained for 2013 were
compared with actual observations from that year (Figure 2). We calculated the percentage of
departments for which hornet presence/absence was correctly predicted.
Due to uncertainty on the value of some parameters, we did an uncertainty analysis. First,
we identified the possible range of the parameters’ values for the local spread rate (c), the carrying
capacity (K), the growth rate (r) and the detection radius (d). Then, we simulated the potential
spread of the yellow-legged hornet between 2004 and 2013 using these values to obtain the
possible range of the model outputs. We considered no long-distance dispersal and α = 0.30, and we
calculated the percentage of departments correctly predicted.
To predict the effects of future control efforts, we took into account the simulated spread in
2013 (as initial conditions) and the most likely dispersal scenario (spread without human-mediated
dispersal; see results). We then simulated the species’ potential spread through 2020. The effects of
control intensity were tested (at α = 0.3, 0.6, and 0.95, which represent the current control intensity,
a doubling of control efforts, and nearly maximal control efforts, respectively).
Results
LOCAL DISPERSAL AND POPULATION GROWTH
The yellow-legged hornet spread over a radius of 112 km during 2004-2005, 75 km during 2005-
2006, 89 km during 2006-2007, and 75 km during 2008-2009; a pioneer population spread by 30 km
during 2007-2008. Therefore, the rate of spread ranged between 75 and 112 km/year. Given that
the yellow-legged hornet was discovered in 2004 but had probably arrived before then and needed
time to build up and spread (i.e., spreading less rapidly, possibly by around 30 km for the first year,
like the pioneer population above), the local rate of spread was more likely between 75 and 82
km/year. In the simulations, we considered that c = 78 km/year in southwestern France (Table 1).
In the department with the greatest infestation levels, the highest number of nests was
about 330. Since the area of this department is 5,361 km², we estimated K at 0.06 nest/km².
There was a weak positive correlation between r and GI (R2 = 0.11; F1,19 = 2.347, P > 0.05)
(Figure 3). The growth rate was defined by the following formula:
Accepted Article
This article is protected by copyright. All rights reserved.
= −1.29593 + 0.09502 × Equation 5
D was estimated for each department in the Aquitaine region using their respective r values
(r = 1.35, 1.80, 2.39, 1.08, and 1.65). Given that c = 78 km/year, the mean value of D was 984 km²
year-1 (Table 1).
Between 2004 and 2009, the highest elevation where the yellow-legged hornet was found
was E = 791.5 m. Consequently, this was the threshold used in the simulations.
LONG-DISTANCE DISPERSAL
A total of 30 occurrence points were found further than 78 km (the mean radial distance) from their
nearest neighbors. Distances ranged almost continuously from 78 km to 327 km (Figure 4), including
the three furthest points generally attributed to human-mediated dispersal in the literature; they
are located in Côte d’Or (point number 1), Ille-Et-Vilaine (point number 2), and Seine-Saint-Denis
(point number 3) (Figure 4). Since the distance distribution was rather continuous, there was no
clear threshold separating short-distance events from long-distance events, which calls into question
previous assumptions regarding human-mediated dispersal.
At these 30 occurrence points, human population densities were higher than at the 100
randomly selected points (W = 2034, P < 0.001) (average of 322.3 and 86.9 inhabitants/km²,
respectively). There were a subset of two to five points located further than the others, at the
periphery of the species’ range (Figure 4), which were therefore more likely associated with long-
distance dispersal events. At these points, human population densities were the following: 126.3
(point 1), 566.1 (point 2), 1,490.4 (point 3), 228.9 (point 4), and 19.0 (point 5). We chose H = 125
inhabitants/km² for the threshold above which the yellow-legged hornet arrives via human-
mediated dispersal. Four of the five extreme points were above this distance threshold. Although
point 5 had a relatively low human population density, it was located very close to point 4. The
yellow-legged hornet was observed at points 4 and 5 the same year, in 2007, so it is possible that it
actually arrived before then and spread from one point to the other.
A stochastic approach was used to simulate human-mediated dispersal. As the number of
points discovered at very long distance and possibly resulting from human-mediated dispersal
events was possibly between two and five for a 6-year period (2004-2009) (Figure 4) and a large
proportion of France is suitable for species establishment (meaning that most human-mediated
dispersal events could result in establishment), we estimated that the yearly probability of hornets
Accepted Article
This article is protected by copyright. All rights reserved.
being accidently dispersed by humans in France, denoted as Pldd, ranged from 0.3 to 0.8. A Bernoulli
distribution with probability Pldd was used to simulate the occurrence of human-mediated dispersal.
In case of long-distance events, we randomly selected a nest founding location within suitable
habitat (where E ≤ 791.5 m) that was not yet invaded and where human population density was
above H.
MODEL VALIDATION: ROLE OF HUMAN-MEDIATED DISPERSAL AND CURRENT CONTROL INTENSITY
A higher percentage of departments had their statuses correctly predicted (i.e., hornet
presence/absence) in the simulations without human-mediated dispersal and when control intensity
was intermediate (α = 0.3-0.4; Figure 5). When human-mediated dispersal was present, the model
overestimated the hornet’s spread, especially in northern and eastern France (Figure 6; Figure S1 in
SM1). In simulations without human-mediated dispersal but with higher control intensity (α ≥ 0.6),
spread was underestimated (Figure 6; Figure S1 in SM1) and predictions were less accurate (Figure
5). These results suggest that past control efforts have been moderate. In addition, when control
intensity was maximal (α = 1), all the dispersal scenarios converged. This finding makes sense since,
in the model, colonies dispersed by humans arrive in urban areas and most are destroyed (C(h) ≈ α).
UNCERTAINTY ANALYSIS
Uncertainty on the local spread rate (c from 75 to 82 km/ year) and on the carrying capacity (K from
0.06 to 0.18 nests/km²) impacted the percentage of departments correctly predicted by -3.1 to +1.1
%. Uncertainty on the detection radius (d from 10 to 1000 m) impacted this percentage by 0 to -
5.2%, with a best fit for values ranging from 40 to 70 m, which includes the default value (50 m). The
model output was mostly sensitive to the population growth (r). To estimate this parameter, we
calculated the observed growth rate each year (before it reached half of the carrying capacity), and
then these values were averaged for each invaded department. In this uncertainty analysis, we
considered the lower and upper limit of the 95% confidence interval of this mean, and we
considered the following linear regressions:
= −1.45386 + 0.07018 × Equation 6
= −1.13801 + 0.11985 × Equation 7
Accepted Article
This article is protected by copyright. All rights reserved.
for the lower limit of the 95% confidence interval (rinf) and the upper limit of the 95% confidence
interval (rsup), respectively. The percentage of correctly predicted departments decreased by 7.3%
when considering rinf and by 31.5% when considering rsup. The spread model developed in this study
was relatively robust, even if there was some uncertainty associated with the estimations of certain
parameters (see Appendix S2 for more details).
MODEL PREDICTIONS: EFFECTS OF INCREASING CONTROL INTENSITY
Based on the simulations run for the period between 2013 and 2020 in which human-mediated
dispersal events were excluded, the yellow-legged hornet could continue to expand its range in the
future (Figure 7). When control intensity was increased from 30 to 60%, the hornet’s spread was
reduced by 17.1% (453,400 versus 375,800 km² of total invaded area) and nest density was reduced
by 27.9% (from 16,249 to 11,708 nests). When control intensity was raised even further, to 95%, the
hornet’s spread was reduced by 42.7% (to 259,900 km² of invaded area) and nest density was
reduced by 53.3% (to 7,587 nests).
Discussion
YELLOW-LEGGED HORNET INVASIVENESS IS TIED TO DISPERSAL AND ESTABLISHMENT ABILITIES
This study presents the first evidence that the rapid spread throughout France of the yellow-legged
hornet may not be necessarily mediated by humans. Although the model used was stochastic and
may not precisely predict the area to be invaded, it nonetheless performed better than a purely
random model (see Appendix S1). It was clear that including human-mediated dispersal in the model
resulted in less accurate predictions than when only dispersal over “short” distances was allowed. In
fact, such “short-distance” dispersal could represent self-mediated dispersal by the yellow-legged
hornet and occur over relatively large distances: the average was 78 km/year in this study. How can
this unexpected result be explained?
First, founder females are capable of flying very long distances (Beggs et al. 2011). They can
cover 200 km in several days in flight mill experiments (Pers. Com., Dr. Daniel Sauvard, INRA,
Orléans, France). Despite the possible bias of such experiments, the distances recorded fit with the
rate of spread observed for some species in the field. Indeed, species range expansion can result
from the long-distance dispersal of just a few individuals with extreme dispersal abilities (e.g.,
around 5 km/year for the pine processionary moth, Battisti et al. 2005; Robinet et al. 2012). The
overall flight period of yellow-legged hornet gynes is relatively long because they may fly before
Accepted Article
This article is protected by copyright. All rights reserved.
hibernation (i.e., while looking for a hibernation site) and after hibernation (i.e., while building new
nests and thus founding new colonies). This long flight period could therefore translate into long
dispersal distances.
Second, the yellow-legged hornet is extremely successful at establishing new colonies. The
invasion in France probably resulted from the introduction of no more than a few and perhaps even
just one mated female (Arca et al. 2015). If so, it means that if a mated female manages to fly a
hundred kilometers and arrives in favorable habitat, then there is a high probability that she will
succeed in founding a new population. The fact that a single individual can give rise to a whole
population makes the yellow-legged hornet a formidable invader (EEA 2012). Because of the species’
high degree of establishment success, only a few females need to disperse long distances to create
satellite colonies far from the species’ main range.
DISENTANGLING HUMAN-MEDIATED DISPERSAL FROM SELF-MEDIATED LONG-DISTANCE DISPERSAL
This study suggests that individual hornets would be able to disperse long distances on their own
and thus generate stochastic patterns of spread similar to those resulting from human-mediated
dispersal. If the patterns they produce are similar, it is difficult to discriminate between the two
potential scenarios. This study therefore differs from other studies of invasive species in which the
long-distance dispersal events associated with rapid range expansion are mainly attributed to
human activities. For instance, human-mediated dispersal was considered as the factor explaining
long-distance dispersal and the subsequent rapid spread of the following species: (i) the pine
processionary moth, Thaumetopoea pityocampa, whose spread across France was hypothesized to
result from the transport and planting of large potted trees whose soil was infested (Robinet et al.
2012); (ii) the horse-chestnut leafminer, Cameraria ohridella, whose spread across Europe was
attributed to the transport of infested leaves via car traffic (Gilbert et al. 2004); (iii) the pine wood
nematode, Bursaphelenchus xylophilus, which is thought to have spread across China because of the
transport of infested wood material (Robinet et al. 2009); and (iv) the emerald ash borer, Agrilus
planipennis, which may have spread across North America because of the transport of infested
firewood (Muirhead et al. 2006). The results of this study therefore indicate that greater caution is
needed when interpreting the mechanisms explaining long-distance dispersal events.
FUTURE RESEARCH: APPLYING THE MODEL TO OTHER COUNTRIES
Accepted Article
This article is protected by copyright. All rights reserved.
It is not known whether there have been other, more recent introductions of the yellow-legged
hornet from Asia to Europe. The model presented here was developed to describe the potential
spread of the yellow-legged hornet in France. However, due to the species’ ability to spread rapidly
and effectively, it is now necessary to assess its potential spread across Europe. This same model
could be used to predict the hornet’s spread in Spain, Portugal, and Italy, countries where the hornet
is already present. The estimates of the model parameters could be adjusted and the model
eventually refined with these additional datasets. This model can also be used as a companion to
genetic studies, which are commonly used to trace the origins of populations (e.g., Arca et al. 2015);
simulations can be used to test whether the yellow-legged hornet likely arrived from neighboring
countries (if the simulation results fit with observations) or following other introduction events (if
the simulation results do not fit with observations). It is also possible to determine the likelihood
that the species will spread to uninvaded neighboring countries with favorable climates, such as
Switzerland, Belgium, and the United Kingdom.
COMBINING CONTROL METHODS TO IMPROVE EFFICIENCY
Insect eradication may be possible when the invasion is confined to small areas or a sharply
delimited breeding habitat (Pluess et al. 2012). Since the yellow-legged hornet has already spread
throughout most of France, it is no longer possible to eradicate it. Instead, the best strategy would
be to slow its spread and reduce its population densities at large scale. Control measures could
combine the mechanical removal and destruction of individuals or infested materials with biological
control techniques. For example, such an approach was successfully used to control the spruce
beetle, Dendroctonus micans (Fielding & Evans 1997; Pauly & Meurisse 2007). In the United Kingdom
and France, clearcutting infested trees and releasing a natural predator, Rhizophagus grandis,
caused the population to decline; it ultimately reached a low-level equilibrium, particularly in
recently colonized areas and areas experiencing outbreaks. Combining control measures in this way
could also be useful when dealing with rapidly spreading invaders. For instance, removing leaves
infested by the horse-chestnut leafminer, Cameraria ohridella, and releasing one of the insect’s
generalist predators could cause populations to reach an economically acceptable threshold (Aniţa
et al. 2014).
Combining different approaches—such as (1) removing and destroying hornet nests, (2)
trapping hornets with more selective traps and more specific attractant, and (3) releasing natural
enemies—could be highly valuable in efforts to control the yellow-legged hornet in France and in
Accepted Article
This article is protected by copyright. All rights reserved.
Europe in general. Although none of these methods would be 100% effective, they might be more
efficient when combined, applied at larger scales, and coupled with certain phenomena, such as
inbreeding, that naturally reduce the population sizes of invaders (Darrouzet et al. 2015a). Nest
removal and destruction can be performed more frequently if they are required by a country’s
regulations. Trapping yellow-legged hornets is not yet efficient since traps and attractants are not
selective enough (Monceau et al. 2012; Goldarazena et al. 2015). Reducing nest density should be
possible if traps catch gynes. There is therefore an urgent need to develop selective traps. In
addition, choosing appropriate locations at which to set traps (e.g., not necessarily near hives) could
improve trapping efficiency (Monceau et al. 2012). Since gynes are trapped under certain climatic
conditions, removing or closing traps when these conditions are not present could help prevent the
capture of non-target species (Monceau et al. 2012). Another promising control method that merits
further study is the use of biological control agents. These agents include the endoparasitoid, Conops
vesicularis, which was recently observed in dead queens in France (Darrouzet et al. 2015b) and
parasitic nematodes such as Pheromermis vesparum, which were collected from adult hornets in
France (Villemant et al. 2015). Although these biological agents by themselves would hardly hamper
the invasion (Villemant et al. 2015), if more efficient natural enemies were found, especially in the
yellow-legged hornet’s native range, they could be used to weaken populations. Nevertheless, like
for any intentional release of natural enemies, caution is needed as these controlling agents
themselves could have unexpected impact on the environment (Hajek et al. 2016). Some infamous
examples are: the marine cane toad, Bufo marinus, introduced in Australia to control cane beetles,
which has rapidly spread for unclear reasons and caused severe problems (Shanmuganathan et al.
2010), and the Harlequin ladybird, Harmonia axyridis, an Asian coccinellid introduced in America,
Europe and Africa to control aphids, which became invasive after some admixture with bridgehead
populations (Lombaert et al. 2010).
Combining different control methods in these ways, whilst preventing potential side effects,
could be the best approach to reduce the yellow-legged hornet’s impact and slow its spread in
France and in Europe in general. To this end, species-specific management policies should be
adopted in France and Europe in addition to the recent EU regulations instituted regarding invasive
alien species (see Regulation 1143/2014;
http://ec.europa.eu/environment/nature/invasivealien/index_en.htm). European countries should
anticipate the potential spread of invaders, especially given climate warming (Barbet-Massin et al.
2013), such that appropriate regulations are in place prior to the arrival of invasive species. In
France, the yellow-legged hornet has been considered a noxious species since 2012, yet prevention,
monitoring, and/or management efforts are still not mandatory (Monceau et al. 2014). The results
Accepted Article
This article is protected by copyright. All rights reserved.
of this study (which underscore that it is possible to slow the spread and reduce the nest densities of
the yellow-legged hornet if stronger control measures are deployed), and the development of new
control techniques (such as selective traps) will encourage decision makers to adopt stronger pest
management policies to control the invasive yellow-legged hornet and thereby reduce its impact on
honey bees, native entomofauna, and humans.
Acknowledgments
We gratefully acknowledge Claire Villemant, Quentin Rome, and Franck Muller (Museum of Natural
History, Paris, France) for kindly providing the 2004–2009 hornet occurrence data for France; Daniel
Sauvard (INRA, URZF, Orléans, France) for providing preliminary flight mill data about the flight
capabilities of the yellow-legged hornet; Jean-Claude Grégoire (Université Libre de Bruxelles,
Belgium) for providing the latest information about the invasion in Belgium; Sylvie Augustin and
Annie Yart (INRA, URZF, Orléans, France) for describing examples of successful control efforts. This
study was supported by funding from the region Centre-Val de Loire, namely the awarding of a
French regional grant for our “FRELON” project (“Frelon asiatique: étude et lutte ciblée contre une
espèce invasive prédatrice des abeilles”; 2011-2014).
Data accessibility
The R script and the data needed to use this script are available at https://zenodo.org/record/49021
(Robinet et al. 2016). The R code of the core function and 2013 occurrence data are provided in the
Appendix S1of this manuscript. Occurrence data for 2004 to 2009 are provided in the Supplementary
Material in Villemant et al. (2011b).
References
Aniţa, S., Casas, J., Suppo, C. (2014) Impulsive spatial control of invading pests by generalist
predators. Mathematical Medicine and Biology, 31, 284-301.
Accepted Article
This article is protected by copyright. All rights reserved.
Arca, M., Mougel, F., Guillemaud, T., Dupas, S., Rome, Q., Perrard, A. et al. (2015) Reconstructing the
invasion and the demographic history of the yellow-legged hornet, Vespa velutina, in Europe.
Biological Invasions, DOI 10.1007/s10530-015-0880-9
Barbet-Massin, M., Rome, Q., Muller, F., Perrard, A., Villemant, C., Jiguet, F. (2013) Climate change
increases the risk of invasion by the yellow-legged hornet. Biological Conservation, 157, 4-10.
Battisti., A., Stastny, M., Netherer, S., Robinet, C., Schopf, A., Roques, A., Larsson, S. (2005)
Expansion of geographic range in the pine processionary moth caused by increased winter
temperatures. Ecological Applications, 15, 2084-2096.
Beggs, J.R., Brockerhoff, E.G., Corley, J.C., Kenis, M., Masciocchi, M., Muller, F., Rome, Q., Villemant,
C. (2011) Ecological effects and management of invasive alien Vespidae. BioControl, 56, 505-526.
Choi, M.B., Martin, S.J., Lee, J.W. (2012) Distribution, spread, and impact of the invasive hornet
Vespa velutina in South Korea. Journal of Asia-Pacific Entomology, 15, 473-477.
CIESIN & CIAT (2005) Gridded Population of the World, version 3 (GPWv3): Population Density Grid.
Palisades, NY: NASA SEDAC. http://dx.doi.org/10.7927/H4XK8CG2. Accessed 05 April 2016.
Darrouzet, E. (2013) Les insectes bâtisseurs. Nids de termites, de guêpes et de frelons. Editions
Connaissances et Savoirs, Paris, pp. 108.
Darrouzet, E., Gévar, J., Guignard, Q., Aron, S. (2015a) Production of Early Diploid Males by European
Colonies of the Invasive Hornet Vespa velutina nigrithorax. PLoS ONE 10(9): e0136680.
doi:10.1371/journal.pone.0136680
Darrouzet, E., Gévar, J., Dupont, S. (2015b) A scientific note about a parasitoid that can parasitize the
yellow-legged hornet, Vespa velutina nigrithorax, in Europe. Apidologie, 46, 130-132.
DOI:10.1007/s13592-014-0297-y
EEA (European Environment Agency) (2012) The impacts of invasive alien species in Europe.
Technical report 16/2012, doi:10.2800/65864, pp. 114.
http://www.eea.europa.eu/publications/impacts-of-invasive-alien-species
Evans, H. & Oszako, T. (2007) Alien Invasive Species and International Trade. Forest Research
Institute, Warsaw, pp. 179.
www.iufro.org/download/file/4695/4505/70312-jedlnia06_pdf/
Accepted Article
This article is protected by copyright. All rights reserved.
Federazione Apicoltori Italiani (2013) Vespa velutina: prima segnala’ione in Italia. E' allarme! (Vespa
velutina: first record in Italy. Alarm!). Rome, Italy: Federazione Apicoltori Italiani.
http://www.fede rapi.biz/index.php?option=com_content&task=view&id=1176&Itemid=1
Fielding, N. & Evans, H. (1997) Biological control of Dendroctonus micans (Scolytidae) in Great
Britain. Biocontrol News and Information, 18, 51-60.
Gilbert, M., Gregoire, J.-C., Freise, J.F., Heitland, W. (2004) Long distance dispersal and human
population density allow the prediction of invasive patterns in the horse chestnut leafminer
Cameraria ohridella. Journal of Animal Ecology, 73, 459–468.
Goldarazena, A., de Heredia, I.P, Romon, P., Iturrondobeitia, J.C., Gonzalez, M., Lopez, S. (2015)
Spread of the yellow-legged hornet Vespa velutina nigrithorax du Buysson (Hymenoptera:
Vespidae) across Northern Spain. EPPO Bulletin, 45, 133-138.
Grosso-Silva, J.M. & Maia, M. (2012) Vespa velutina Lepeletier, 1836 (Hymenoptera, Vespidae), new
species for Portugal. Arquivos Entomoloxicos, 6, 53-54.
Hajek, A.E., Hurley, B.P., Kenis, M., Garnas, J.R., Bush, S.J., Wingfield, M.J., van Lenteren, J.C., Cock,
M.J.W. (2016) Exotic biological control agents: a solution or contribution to arthropod invasions?
Biological Invasions, 18, 953-969.
Haxaire, J., Bouguet, J.P., Tamisier, J.-P. (2006) Vespa velutina Lepeletier, 1836, une redoutable
nouveauté pour la faune de France (Hym., Vespidae). Bulletin de la Société entomologique de
France, 111(2), 194.
Hulme, P.E., Roy, D.B., Cunha, T., Larsson, T.-B. (2009) A pan-European inventory of alien species:
rationale, implementation and implications for managing biological invasions. Handbook of alien
species in Europe, Springer Science+ Business Media BV, pp. 1-14.
Ibáñez-Justicia, A. & Loomans, A.J.M. (2011) Mapping the potential occurrence of an invasive species
by using CLIMEX: case of the Asian hornet (Vespa velutina nigrithorax) in The Netherlands.
Proceedings of the Netherlands Entomological Society Meeting, 22, 39-46.
Levine, J.M. & D’Antonio, C.M. (2003) Forecasting biological invasions with increasing international
trade. Conservation Biology, 17, 322-326.
Lombaert, E., Guillemaud, T., Cornuet, J.-M., Malausa, T., Facon, B., Estoup, A. (2010) Bridgehead
effect in the worldwide invasion of the biocontrol Harlequin ladybird. PLoS ONE, 5(3):e9743.
Doi:10.1371/journal.pone.0009743
Accepted Article
This article is protected by copyright. All rights reserved.
López, S., González, M., Goldarazena, A. (2011) Vespa velutina Lepeletier, 1836 (Hymenoptera,
Vespidae): first records in Iberian Peninsula. EPPO Bulletin / Bulletin OEPP, 41, 439-441.
Millot, G. (2011) Comprendre et réaliser les tests statistiques à l’aide de R. Groupe De Boeck s.a.,
Editions De Boeck Université, Bruxelles, pp. 767.
Monceau, K., Bonnard, O., Thiéry, D. (2012) Chasing the queens of the alien predator of honeybees:
a water drop in the invasiveness ocean. Open Journal of Ecology, 2, 183-191.
Monceau, K., Bonnard, O., Thiéry, D. (2014) Vespa velutina: a new invasive predator of honeybees in
Europe. Journal of Pest Science, 87, 1-16.
Muirhead, J.R., Leung, B., van Overdijk, C, Kelly, D.W., Nandakumar, K., Marchant, K.R., MacIsaac,
H.J. (2006) Modelling local and long-distance dispersal of invasive emerald ash borer Agrilus
planipennis (Coleoptera) in North America. Diversity and Distributions, 12, 71–79.
Mutze, G., Cooke, B., Alexander, P. (1998) The initial impact of rabbit hemorrhagic disease on
European rabbit populations in South Australia. Journal of Wildlife Diseases, 34, 221-227.
Pauly, H. & Meurisse, N. (2007) Le dendroctone de l’épicéa en France : situation et mesures de lutte
en cours. Revue Forestière Française, 6, 595-608.
Péré, C. & Kenis, M. (2011) Le frelon asiatique (Vespa velutina): état des connaissances et évaluation
du risque pour la Suisse. CABI report, pp. 16.
Pluess, T., Cannon, R., Jarošík, V., Pergl, J., Pyšek, P., Bacher, S. (2012) When are eradication
campaigns successful? A test of common assumptions. Biological Invasions, 14, 1365-1378.
R Development Core Team (2014). R: A language and environment for statistical computing. R
Foundation for Statistical Computing, Vienna, Austria. ISBN 3-900051-07-0, URL http://www.R-
project.org
Robinet, C., Roques, A., Pan, H.Y., Fang, G.F., Ye, J.R., Zhang, Y.Z., Sun, J.H. (2009) Role of human-
mediated dispersal in the spread of the pinewood nematode in China. PLoS One, 4(2), e4646.
doi:10.1371/journal.pone.0004646
Robinet, C., Imbert, C.-E., Rousselet, J., Sauvard, D., Garcia, J., Goussard, F., Roques, A. (2012)
Human-mediated long-distance jumps of the pine processionary moth in Europe. Biological
Invasions, 14, 1557-1569.
Accepted Article
This article is protected by copyright. All rights reserved.
Robinet, C., Suppo, C., Darrouzet, E. (2016) R script for the potential spread of Vespa velutina in
France. Doi:10.5281/zenodo.49021. Available at: https://zenodo.org/record/49021.
Rome, Q., Muller, F., Gargominy, O., Villemant, C. (2009) Bilan 2008 de l’invasion de Vespa velutina
Lepeletier en France (Hymenoptera, Vespidae). Bulletin de la Société entomologique de France,
114 (3), 297-302.
Rome, Q., Perrard, A., Muller, F., Villemant, C. (2011) Monitoring and control modalities of a
honeybee predator, the yellow-legged hornet Vespa velutina nigrithorax (Hymenoptera,
Vespidae). Aliens: the invasive species bulletin, 31, 7-15.
http://www.issg.org/pdf/aliens_newsletters/A31.pdf
Rome, Q., Muller, F., Villemant, C. (2012) Expansion en 2011 de Vespa velutina Lepeletier en Europe
(Hym., Vespidae). Bulletin de la Société entomologique de France, 117 (1), 114.
Rome, Q., Dambrine, L., Onate, C., Muller, F., Villemant, C., García Pérez, A.L., Maia, M., Carvalho
Esteves, P., Bruneau, E. (2013) Spread of the invasive hornet Vespa velutina Lepeletier, 1836, in
Europe in 2012 (Hym., Vespidae). Bulletin de la Société entomologique de France, 118, 21-22.
http://inpn.mnhn.fr/docs/Vespa_velutina/1643-1-Rome_et_al_2013_p.pdf
Rome, Q., Muller, F.J., Touret-Alby, A., Darrouzet, E., Perrard, A., Villemant C. (2015) Caste
differentiation and seasonal changes in Vespa velutina (Hym.: Vespidae) colonies in its introduced
range. Journal of Applied Entomology. DOI: 10.1111/jen.12210 (in press)
Roques, L., Roques, A., Berestycki, H., Kretzschmar, A. (2008) A population facing climate change:
joint influences of Allee effects and environmental boundary geometry. Population Ecology, 50,
215-225.
Roques, A., Rabitsch, W., Rasplus, J.-Y., Lopez-Vaamonde, C., Nentwig, W., Kenis, M. (2009) Chapter
5: Alien terrestrial invertebrates of Europe. Handbook of alien species in Europe, Springer
Science+ Business Media BV, pp. 63-79.
Shanmuganathan, T., Pallister, J., Doody, S., McCallum, H., Robinson, T., Sheppard, A., Hardy, C.,
Halliday, D., Venables, D., Voysey, R., Strive, T., Hinds, L., Hyatt, A. (2010) Biological control of the
cane toad in Australia: a review. Animal Conservation, 13 Suppl 1, 16-23.
Sharov, A.A., Liebhold, A.M. (1998) Model of slowing the spread of gypsy moth (Lepidoptera:
Lymantriidae) with a barrier zone. Ecological Applications, 8, 1170–1179.
Accepted Article
This article is protected by copyright. All rights reserved.
Shigesada, N. & Kawasaki, K. (1997) Biological Invasions: Theory and Practice. Oxford Series in
Ecology and Evolution, Oxford University Press, Oxford.
Sutherst, R.W., Maywald, G.F., Kriticos, D. (2007) CLIMEX Version 3. User’s Guide. Hearne Scientific
Software Pty Ltd. Available: http://www.Hearne.com.au
Villemant, C., Haxaire, J., Streito, J.-C. (2006) Premier bilan de l’invasion de Vespa velutina Lepeletier
en France (Hymenoptera, Vespidae). Bulletin de la Société entomologique de France, 111, 535-
538.
Villemant, C., Barbet-Massin, M., Perrard, A., Muller, F., Gargominy, O., Jiguet, F., Rome, Q. (2011a)
Predicting the invasion risk by the alien bee-hawking yellow-legged hornet Vespa velutina
nigrithorax across Europe and other continents with niche models. Biological Conservation,
144, 2142-2150.
Villemant, C., Muller, F., Haubois, S., Perrard, A., Darrouzet, E., Rome, Q. (2011b). Bilan des travaux
(MNHN et IRBI) sur l’invasion en France de Vespa velutina, le frelon asiatique prédateur
d’abeilles. In : Barbançon, J-M, L’Hostis, M (eds). Journée Scientifique Apicole. JSA, Arles, 11
février 2011. ONIRIS-FNOSAD, Nantes pp. 18-28.
Villemant, C., Zuccon, D., Rome, Q., Muller, F., Poinar Jr, G.O., Justine, J.-L. (2015) Can parasites halt
the invader? Mermithid nematodes parasitizing the yellow-legged Asian hornet in France.
PeerJ 3:e947; DOI 10.7717/peerj.947
TABLES
Table 1. Estimates of model parameters. (*) Since parameters c and r vary across France, the
estimates provided are those for southwestern France.
Parameter Symbol Estimate
Rate of spread
c
78 km year-1 (*)
Carrying capacity K0.06 nest km-²
Growth rate r1.66 (*)
Diffusion coefficient D984 km² year-1
Maximum elevation
above sea level where hornets are found
E
791.5 m
Annual probability of long-distance
dispersal
PLDD 0.3-0.8
Threshold of human population density H125 km-2
FIGURES
Accepted Article
This article is protected by copyright. All rights reserved.
Figure 1. Control intensity as a function of human population density (given by the number of
inhabitants within 10 km x 10 km cells) when α = 1. The map shows the corresponding control
intensity throughout France which goes from low (less than 10%) in light gray areas to high (more
than 90%) in black areas. The dark gray polygons delineate the borders of the French departments.
Figure 2. Distribution of the yellow-legged hornet in 2013 (source: INPN website). The data are
projected onto a grid with a resolution of 10 km x 10 km (on the left) and depicted per department
(on the right). The black point in southwestern France indicates where the yellow-legged hornet was
first found in 2004.
Figure 3. Growth rate. (a) Correlation between the observed growth rates (r) and the growth index
values estimated from the CLIMEX model (GI) (one point per department). (b) Predicted growth
rates estimated from the GI values across France (10-km grid resolution).
Figure 4. Long-distance dispersal. (A) Number of hornet occurrence points found more than 78 km
from their nearest neighbors. (B) Locations of these long-distance dispersal events (black dots). The
most extreme points (largest black dots) are numbered so their distances can be identified in (A). In
the literature, points 1, 2, and 3 are hypothesized to have resulted from long-distance dispersal
(Villemant et al. 2011a). The gray dots are points located less than 78 km from their nearest
neighbors.
Figure 5. Percentage of departments for which the model correctly predicted the presence/absence
of the yellow-legged hornet in 2013 at different control intensities (α from 0 to 1). Several dispersal
scenarios were tested—the yearly probability of human-mediated dispersal events varied (Pldd =
0.3, 0.5, and 0.8) and human-mediated dispersal events were also excluded. In gray are the 95%
confidence intervals for each human-mediated dispersal scenario.
Figure 6. Potential spread in 2013 for simulations under different control intensities (α) and different
human-mediated scenarios (without human-mediated dispersal “no LDD” and with different yearly
probability of human-mediated dispersal event, Pldd). The color gradient expresses differences in
nest density (yellow: 0.01–0.1 nests per cell; orange: 0.1– 1 nests per cell; red: 1–3 nests per cell;
and dark red: more than 3 nests per cell).
Figure 7. Potential spread in 2020 for different control intensities (α =0.3, 0.6, and 0.95) where long-
distance, human-mediated dispersal was excluded. The color gradient expresses the differences in
Accepted Article
This article is protected by copyright. All rights reserved.
hornet population density as estimated via the number of nests in each grid cell (cell size = 10 km x
10 km).
Supporting Information
Appendix S1. Technical description of the model and additional supporting results.
Appendix S2. Uncertainty analysis.
Accepted Article
This article is protected by copyright. All rights reserved.
Accepted Article
This article is protected by copyright. All rights reserved.
