Regional level risk factors associated with the occurrence of African swine fever in West and East Africa
Huang et al. Parasites & Vectors
Regional level risk factors associated with the occurrence of African swine fever in West and East Africa
Zheng Y. X. Huang 0 1
Frank van Langevelde 0
Karanina J. Honer 0 2
Marc Naguib 2
Willem F. de Boer 0
0 Resource Ecology Group, Wageningen University , 6708PB Wageningen , The Netherlands
1 College of Life Sciences, Nanjing Normal University , 210023 Nanjing , China
2 Behavioural Ecology, Wageningen University , 6708WD Wageningen , The Netherlands
Background: African swine fever (ASF) causes severe socio-economic impacts due to high mortality and trade restrictions. Many risk factors of ASF have been identified at farm level. However, understanding the risk factors, especially wild suid hosts, determining ASF transmission at regional level remains limited. Methods: Based on ASF outbreak data in domestic pigs during 2006-2014, we here tested, separately for West and East Africa, which risk factors were linked to ASF presence at a regional level, using generalized linear mixed models. Results: Our results show that ASF infections in the preceding year was an important predictor for ASF presence in both West and East Africa. Both pig density and human density were positively associated with ASF presence in West Africa. In East Africa, ASF outbreaks in domestic pigs were also correlated with higher percentages of areas occupied by giant forest hogs and by high-tick-risk areas. Conclusions: Our results suggest that regional ASF risk in East Africa and in West Africa were associated with different sets of risk factors. Regional ASF risk in West Africa mainly followed the domestic cycle, whereas the sylvatic cycle may influence regional ASF risk in East Africa. With these findings, we contribute to the better understanding of the risk factors of ASF occurrence at regional scales that may aid the implementation of effective control measures.
Domestic cycle; Sylvatic cycle; Wild suid; Giant forest hog; Habitat fragmentation; Ornithodoros moubata
African swine fever (ASF), caused by a DNA virus of the
genus Asfivirus, is a highly contagious disease for
domestic pigs. It can have severe socio-economic
impacts due to high mortality and trade restrictions .
ASF is considered the major limiting factor for pig
production in many sub-Saharan African countries where a
significant commercial pig industry exists [1, 2]. ASF has
also spread outside Africa, including recently to the
Caucasus and Russia, posing considerable threats to the
global pig industry [3–6]. Because no vaccine has been
developed, a full understanding of the ecology and
epidemiology of ASF is fundamental to implement
effective control measures [2, 6, 7].
ASF outbreaks have been reported for many
subSaharan African countries, and the epidemiology and
ecology of ASF are thought to vary between different
regions [8, 9]. Generally, it has been considered that ASF
virus can circulate through two cycles in sub-Saharan
Africa, the sylvatic cycle and the domestic cycle [2, 9,
10]. The sylvatic cycle has been documented in some
Eastern and Southern African countries [7, 9–12]. It
involves wild suids as the host, including the common
warthog (Phacochoerus africanus), and is spread via soft
ticks (Ornithodoros moubata) to domestic pigs making
the virus persistent . However, in the domestic cycle
ASF virus mainly spreads via direct contact between
domestic pigs or between pork products and pigs. A recent
study assessed the suitability of these two transmission
© The Author(s). 2017 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0
International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and
reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to
the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver
(http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
pathways in Africa and concluded that the domestic
cycle may occur throughout sub-Saharan Africa, whereas
the sylvatic cycle might be found mainly in East, Central
and Southern Africa, where warthogs are present .
A few studies have been conducted to investigate the
factors linked with ASF risk . For the domestic cycle,
the density or the herd size of domestic pigs has been
considered to be positively associated with ASF
outbreaks at farm level [9, 14]. Farms with traditional
freeranging husbandry systems typically experience higher
ASF risk, presumably due to low biosecurity .
Previous occurrence of the disease on a farm and infected
neighbouring farms has also been found to promote the
chance of ASF infection . In addition to these risk
factors, ASF spread in the domestic cycle has been
linked to trade-related factors [16, 17], such as the
density of the road network and water bodies . In fact,
the movements of infected domestic pigs might be the
most important factor for ASF spread in some regions,
especially in West Africa where the domestic cycle is
considered the only transmission pathway [7, 10, 18].
For the sylvatic cycle, the common warthog has been
considered the most important vertebrate host for the
circulation of ASF virus [13, 19]. Interspecific
transmission between warthogs and domestic pigs may occur
indirectly through Ornithodoros ticks, especially when pigs
that share grazing areas with warthogs are bitten by
infected ticks, or during human interference when
warthog carcasses with infected ticks are transported .
Bushpigs (Potamochoerus larvatus) can also spread ASF
to domestic pigs via ticks, but they are generally
assumed less important than warthogs due to their lower
population densities, nocturnal habits, and limited
contact with pigs and soft ticks [13, 20]. Finally, the red river
hog (Potamochoerus porcus) and the giant forest hog
(Hylochoerus meinertzhangeni) can also be infected .
However, neither species seems to play an important
role in ASF transmission due to their limited/restricted
To date, risk factors have predominantly been
identified at farm level, whereas the influence of risk factors
such as the role of wild suids and tick distribution on
the dynamics of ASF transmission at larger spatial scales
is not well understood [8, 10]. Understanding regional
risk factors of infectious diseases is necessary for
revealing the underlying epidemiological processes that might
lead to important implications for control at regional
scales [1, 21, 22]. Therefore, the present study aims to
identify the risk factors that are associated with ASF
occurrence at a regional scale in Africa. Due to the
potentially different ASF transmission pathways, we compare
the risk factors between West Africa and East Africa.
We expect that the presence of wild suids and tick
distribution is positively associated with the probability of
ASF occurrence in East Africa, whereas domestic pig
density and indicators of trade activities have a positive
effect in West Africa.
African swine fever data
Data on ASF presence/absence in domestic pigs for
some countries in the sub-Saharan Africa from 2005 to
2014 is available in the World Animal Health
Wahidhome/Home) from the World Organization for
Animal Health (OIE) , which provides data on ASF
outbreaks at a temporal resolution of 6 months. Some
African countries reported ASF outbreaks only at
country level, while other countries specified ASF outbreaks
at a smaller administrative level. The smallest
administrative level of reporting was used as the level of analyses
in this study, whilst those countries reporting ASF
outbreaks only at country level were excluded from the
study. Only the countries with “ASF history” (at least
one outbreak was reported during the study period)
were included in the analyses, as this excluded those
countries which had possibly ASF outbreaks but never
reported them, so that we avoided some false absence
data. Thus, we assumed that no report from a
country/administrative unit that had previously reported ASF is
indicative of the absence of disease for that year. In addition, to
determine the effect of previous infection status, we only
involved the administrative areas from the countries that
reporting ASF outbreaks in more than two consecutive
years. Areas without domestic pigs were excluded. The
data set yielded from OIE included eleven countries
(Angola, Benin, Burkina Faso, Cameroon, Ghana,
Mozambique, Malawi, Nigeria, Rwanda, Togo, and
Zambia). As we compared East Africa and West Africa, we
then, according to the definition of Statistics Division of
the United Nations (http://millenniumindicators.un.org/
unsd/methods/m49/m49.htm), excluded Angola and
divided the remaining ten countries into East Africa and
West Africa. The final dataset (Additional file 1: Table S1,
Figure S1) for the analyses of West Africa consisted of
1287 cases of ASF presence/absence data covering six
countries and 177 administration areas, and the final
dataset for East Africa included 394 cases of presence/absence
data covering four countries and 51 administration areas.
Data of the predictors
As the infection status of the previous year (PreInf ) has
been linked to ASF risk in earlier studies, we tested its
effect to determine the dependency of ASF status
between years. We also tested for the effect of the
domestic pig density (Pig) on the risk of ASF occurrence. As
indicators of trade activities, road density (Road) and
human population density (Human) were calculated for
each administrative area (Table 1).
Since ASF risk might be influenced by the occurrence
of some wild suid species, we calculated the percentage
in each administrative area that is occupied by the
common warthog, the bushpig, the red river hog, and the
giant forest hog and tested whether or not the wild suid
species occurrences were correlated with ASF
occurrence. The distributions of these species were obtained
from the African Mammal Databank (AMD), a
GISbased databank of the distribution of medium to large
mammals in Africa [24, 25]. For each species, AMD
includes two polygon coverage files respectively describing
the distribution of suitable habitat and the distribution
of species occurrence at a 1 × 1 km resolution [24, 25].
The intersection of these two distribution maps was
calculated as the ‘actual distribution’ for each species
(Boitani et al. ). The percentage of protected area
within an administrative area (PDA) was also included
in the analysis since it was expected to hold a higher
density of wild suids, increasing ASF risk. Habitat
structure can influence ASF transmission by affecting host
distribution and movements [13, 26]. Therefore, some
habitat related predictors were incorporated in the
analyses. For each administrative area, we first calculated
the percentage of suitable habitat (Habitat) based on a
detailed (300 m) global land cover dataset (GlobCover
by European Space Agency). Twelve out of 23 land-use
types were combined to calculate the areas that can
sustain grazers (see Additional file 1: Table S2), i.e.
grassland and open woodland that are used by livestock and
several of the most important wild hosts. Based on the
map of grazing area, we calculated the Mean Nearest
Neighbour (MNN) and Mean Proximity Index (MPI) to
test the effect of fragmentation and isolation on the risk
of ASF occurrence. MNN, as a measure of patch
isolation, represents the average shortest distance of patches
that can sustain grazers in an administrative area to the
closest similar patch. For each patch, the size and mean
distance to all neighbouring patches of the same type
was calculated to provide the index of MPI, which
measures the degree of isolation and fragmentation .
In addition, climate can affect ASF transmission
dynamics by influencing the biology of the host,
Table 1 Description and summary (mean ± standard deviation, SD) of predictors used in the analyses for West and East Africa, with
abbreviation and unit
Description of data
Giant forest hog
Human population density
Area with high-tick-risk
Area with high-tick-risk
Area with high-tick-risk
Habitat related variables
Percentage of habitat area
Measure of the degree of isolation and fragmentation
Mean nearest neighbour
Percentage of protected area
Annual mean temperature
Mean temperature in preceding year
Annual mean precipitation
Mean precipitation in preceding year PreRainMean mm 97.4 ± 32.0
a There is no bush pig in West Africa and no red river hog in East Africa, tick risk variables were not included in the analyses of West Africa
pathogen, and vector [28, 29]. Precipitation and
temperature can affect ASF transmission by influencing
the movements and habitat use of wild suid species. For
example, in the dry season and under high temperature
conditions, wild suids aggregate at small water ponds,
which might facilitation ASF transmission .
Therefore, the annual mean temperature and mean
precipitation (Table 1) were calculated from the Climate
Research Unit (CRU) datasets , a time-series dataset
that yields month-by-month variations in climate from
1900 to 2010 with a grid cell of 0.5 × 0.5 degree. Because
of the adaptive process of hosts and ticks, the climatic
variables might show lag-effects on pathogen
transmission [32, 33]. We thus also tested the effects of the
temperature and precipitation conditions in the
Map of tick risk
To investigate the role of the sylvatic cycle, we also
tested the effect of the distribution of the tick vector O.
moubata on ASF spread at the regional scale. O.
moubata is considered to be an important tick vector for
ASF . The presence data of O. moubata (more than
200 records) was accessed through the VectorMap data
portal (http://www.vectormap.org). Then we applied
species distribution modelling with a machine learning
technique, namely maximum entropy modeling (Maxent
v.3.3.3 k)  to generate the distribution map of O.
moubata, using 19 data layers as predictors (Additional
file 1: Table S3) containing bioclimatic variables at 30
arc-seconds spatial resolution, acquired from the
WorldClim dataset (http://worldclim.org). The model with 10
subsampling runs using a random test percentage of
20% was constructed with default settings in Maxent
. We included all variables as a full model and did
not remove highly correlated predictors, because we
focused on the explanatory power of the model and
multicollinearity is known to have little effect on model
fit . The average AUC score (the area under the
Receiver Operating Characteristic curve) for the full
model was 0.928 ± 0.010. Based on this model, we
generated a map representing the probabilities of tick
presence (Additional file 2: Figure S1).
With the tick distribution map, we averaged the
probabilities of tick presence for each administrative area
(TickRiskMean). We also arbitrarily set three probability
values (0.2, 0.4 and 0.6) as the thresholds of
high-tickrisk, and consequently calculated the percentages of area
with high-tick-risk (areas with the the probability of tick
presence higher than 0.2, 0.4, and 0.6, respectively) for
each administrative area, and got another three
measures of tick risk (TickRisk2, TickRisk4, TickRisk6). As
O. moubata was predicted to be present in very few
administrative areas in West Africa (Additional file 2:
Figure S1), the tick risk variables were only included
in the analyses of East Africa. In addition, to test the
effect of tick-suid interactions, we also included the
interaction terms between tick risk variables and wild
The data for all predictor variables were acquired or
generated from existing databases (Additional file 1:
Table S4). All data pre-processing was carried out in
Generalized linear mixed models (GLMM) with a binary
response (logit link) were used to examine the effects of
predictors on the probability of ASF occurrence.
Country was included in the models as a random factor to
control for possible differences between countries,
thereby correcting for the effect of differences in
veterinary service efficiency and used control measures. Only
the ASF presence/absence data from 2006 to 2014 were
used as the dependent variable, because the earliest year
in the dataset was 2005, which was used to calculate the
previous infection status (PreInf ) for 2006. Before
performing the GLMMs, we log-transformed Pig, Human,
Road, MPI and MNN, making these variables closer to
normal distribution. All continuous variables were
rescaled to have a mean of zero and a standard deviation
Using GLMMs, we first performed univariate analyses
to identify the potential risk factors. The area of the unit
(Area) was retained in the model as an obligate variable
to correct for the effect of area size. Variables with a
Pvalue of less than 0.15 were identified as potential risk
factors, which were used to construct multiple
regression models. We then assessed multi-collinearity by
examining the variance inflation factor (VIF) of the
candidate variables. For highly correlated independent
variables, only the one with the smallest P-value in the
univariate analyses was maintained in fitting the multiple
regression models to avoid multi-collinearity. To
construct the final multiple model, we used both backward
and forward selection, where the likelihood ratio test
was applied to test for difference in the fit of the nested
models. For both West and East Africa, backward and
forward selection generated same models. We then
included interaction terms after including all main factors.
Main terms were maintained in the model if they were
included in a significant interaction term. We tested for
the spatial autocorrelation of the residuals (of the final
multiple model) using Moran’s I index and found little
evidence of spatial autocorrelation (Additional file 2:
Table S1). The AUC scores for the final models were
also reported to assess the goodness-of-fit. The whole
statistical processes were conducted in R 2.15.1 with
Over the entire study period, 25.0% of the administrative
areas in West Africa reported ASF occurrence, whereas
in East Africa this was 21.0% (Table 2).
Univariate analyses of risk factors
Twenty potential risk factors were individually tested
using univariate regression analyses (Table 3). Previous
infection status (PreInf ) was positively associated with
ASF occurrence in domestic pigs both in West and East
Africa. Besides, five biotic factors (Pig, Human, Road,
WarthogC, and RiverhogC), four climatic factors and
one related to grazing area (Habitat) were significantly
correlated with ASF occurrence in West Africa. In East
Africa, besides PreInf, only three biotic variables
(ForesthogC, WarthogC and TickRisk4) and two climatic
variables (TemMean and PreTemMean) were significantly
associated with ASF occurrence in domestic pigs.
Multiple regression models
After checking for collinearity, nine variables (PreInf,
Pig, WarthogC, RiverhogC, Human, Road, Habitat,
PDA, PreTemMean and RainMean) were retained to
construct the final multiple regression models for West
Africa, and five (PreInf, WarthogC, ForesthogC,
TickRisk4, and PDA) were retained for East Africa. For West
Africa, the results of stepwise model selection (Table 4)
showed that ASF occurrence in domestic pigs was
positively associated with previous infection status (PreInf,
OR = 4.63, 95% CI: 3.92–5.46, P < 0.001), pig density
(Pig, OR = 1.32, 95% CI: 1.19–1.46, P = 0.007) and
human population density (Human, OR = 1.36, 95% CI:
1.23–1.51, P = 0.002), though many variables were
significantly correlated with ASF presence in the univariate
analysis. The area under the ROC curve (AUC) for the
final multiple model for West Africa was 0.785. For East
Africa, previous infection (PreInf, OR = 2.40, 95% CI:
1.84–3.14, P = 0.001), percentage of the area occupied by
forest hog (ForesthogC, OR = 2.14, 95% CI: 1.63–2.81,
P = 0.005), and the percentage of area with
high-tickrisk (TickRisk4, OR = 6.98, 95% CI: 1.22–39.85, P =
0.029) had positive relationships with ASF occurrence
in domestic pigs. No interaction terms was
significantly correlated with ASF presence in East Africa.
The AUC for the final multiple model in East Africa
Many previous studies have been conducted to
investigate risk factors for ASF transmission at farm level [9,
10, 14, 16]. However, understanding of the factors
affecting regional disease risk remains limited. Here, we tested
which factors were correlated with the probability of
ASF presence in domestic pigs at regional scale in
Africa. Our study identified different sets of risk factors in
West Africa and East Africa. Previous infection status
was positively correlated with ASF occurrence in
domestic pigs in both West and East Africa, suggesting that for
both West and East Africa greater effort should be made
to control ASF in those areas that have experienced ASF
outbreaks in the past. Pig density and human density were
also positively associated with regional ASF occurrence in
West Africa, where ASF transmission follows the
domestic cycle. In East Africa, the percentage of the area where
giant forest hog occurred, and the high-tick-risk area, were
positively linked to ASF occurrence in domestic pigs,
suggesting that the sylvatic cycle is of epidemiologic
In line with previous studies [10, 16], the effect of the
previous infection status (PreInf ) on ASF presence in
domestic pigs in West Africa indicates that ASF occurs
Table 2 The number of infected areas, regional prevalence of ASF occurrence in West and East Africa
Table 3 Summary statistics (standardized regression coefficient b,
Z-value and P-value) for the potential predictors correlated with
the occurrence of African swine fever in univariate analyses for
West and East Africa. For explanation of the variables, see Table 2
West Africa (n = 727)
East Africa (n = 217)
< 0.001*** 1.02
< 0.001*** -0.026 -0.15
< 0.001*** 0.42
< 0.001*** 0.522 1.12
aThere is no bush pig in West Africa and no red river hog in East Africa. Tick
risk was not included in the analyses in West Africa
*P < 0.05; **P < 0.01; ***P < 0.001
repeatedly in the same area. In West Africa, where the
domestic cycle dominates, the endemicity of ASF might
be caused by the long persistence of the ASF virus
which, shed by infected pigs or asymptomatic carriers,
can remain in the environment for a long time . In
addition, the significant relationship of previous
infection status might be caused by those factors that were
not considered in the analyses but associated with the
previous ASF occurrence, such as within-country variation in
the quality of veterinary service or used control measures.
We showed that both pig density and human density
were positively correlated with regional ASF occurrence
in West Africa, which was also in line with previous
studies , suggesting that the domestic cycle is indeed
dominant in West Africa. A higher pig density implies a
higher probability for susceptible pigs to contact infected
pigs or pork products , contributing to the spread
and persistence of ASF virus [9, 40]. Thus, the
probability of ASF presence is higher with increasing pig density.
Higher human population density, which was used as
the surrogate of trade activity [41, 42], may also facilitate
ASF spread. In fact, pig movement and trade activity
have been considered as important factors for ASF
spread, especially in West Africa where the domestic
cycle is dominated [10, 11, 17]. As expected, no wild
suids showed a significant relationship with ASF
occurrence, suggesting that the sylvatic cycle may not play an
important role in ASF transmission at regional scales in
Our results from East Africa also identified previous
infection status (PreInf ) as a predictor for ASF occurrence
in domestic pigs at regional scales. The endemicity of
ASF in East Africa might be stimulated by the sylvatic
cycle where infected wild suids and ticks can maintain
the virus, transmit the virus to domestic pigs, and
contribute to ASF persistence. Asymptomatic carriers of
wild suids may also facilitate the persistence of ASF in
East Africa . We also found a positive relationship
between ASF risk and the area with high risk of ticks,
indicating that the sylvatic cycle may play an importance
role in ASF spread in East Africa. Previous studies have
documented the importance of warthogs in the sylvatic
cycle in East Africa. However, we did not find a
significant effect of warthog distribution in East Africa. This
might be caused by the limitation of our warthog data.
Table 4 Summary statistics (standardized regression coefficient b ± standard error, SE; Odds Ratio, OR, 95% CI, and P-value) of the
model averaging analyses on the occurrence of African swine fever in West and East Africa. Variables are previous infection status
(PreInf), pig density (Pig), human population density (Human), the percentage of the area occupied by giant forest hog
(ForesthogC), and the percentage of area with high-tick-risk (TickRisk4)
*P < 0.05; **P < 0.01; ***P < 0.001
West Africa (n = 727)
East Africa (n = 217)
Due to the lack of the data of warthog densities, we
could only use the percentage of area covered by
warthog distribution (WarthogC) as a surrogate to test for
the effect of warthog on ASF occurrence. However, in
our analyses all administrative areas in East Africa had
an occurrence of warthog (WarthogC > 0). The small
variation of WarthogC (86 ± 15% SD, Table 1) precluded
it to be a good variable to test for the effect of warthogs.
It has been considered that the giant forest hog is
unlikely to play an important role in ASF
transmission as its distribution is restricted to areas of dense
forest where domestic pig production is not common
. However, here we show a significant positive
relationship between ASF presence and the occurrence
of giant forest hog (ForesthogC), indicating that giant
forest hog might promote ASF risk at a regional scale
in East Africa. This suggests the presence of the
sylvatic cycle in East Africa. At a regional scale, giant
forest hog may promote an increase in the density of
tick vectors or transmit ASF to other wild suid
species that share the same habitats, and thus increases
regional ASF risk. Certainly, the positive effect of
giant forest hog might be caused by some
unidentified factors that was correlated with giant forest hog.
For example, it could be that in areas where forest
hog occur there are high densities of warthog that
facilitate ASF transmission. In any case, more study is
needed to better understand the effect of giant forest
hog on ASF transmission at regional scale.
Conclusions are not easy to be drawn from
largespatial-scale, correlative studies with data from different
sources, due to the complexity of the natural
environment and the difficulty of controlling confounding
factors . Some variables, such as the quality of the
veterinary services or used control measures, could not
be taken into account because of lack or incompleteness
of data. However, the random factor country used in the
analyses, controlled, to some extent, for the variation
caused by these variables at country level. In addition,
the difficulties of getting harmonized and accurate data
of ASF presence/absence might influence the precision
of this analysis. Especially, the information from the OIE
system might under-estimated the true ASF outbreaks
because of several reasons, such as clandestine pig sales
or ineffective reporting systems due to lack of budget or
change in administration. This under-reporting issue can
limit the generality of our results. Despite these
limitations, our study, for the first time, tests for the effect of
biotic and climatic factors on ASF presence at regional
level in Africa.
Our study showed that the factors that play an
important role in ASF transmission at farm level, like previous
infection status and domestic pig density, also link to
ASF disease dynamics at regional level. We also
demonstrated that ASF occurrence in East and West Africa
was associated with different sets of predictors. For West
Africa, we found support for the importance of the
domestic cycle in ASF transmission where the ASF virus
mainly spreads via direct contact between domestic pigs
or between pork products and pigs. In East Africa, we
show that the distribution of the tick vector and the
giant forest hog were correlated to the probability of
ASF presence at regional level, but more efforts are still
needed to better understand the sylvatic cycle and the
role of wild suids in the epidemiology of ASF at this
level. Our results are relevant for developing more
effective control strategies for ASF spread as they
highlight variation in the regional correlates of ASF
AMD: African mammal databank; ASF: African swine fever; AUC: Area
under ROC curve; CI: Confidence interval; CRU: Climate Research Unit;
GLMM: Generalized linear mixed model; OIE: World Organization for
Animal Health; OR: Odds ratio; ROC curve: Receiver operating
characteristic curve; SD: Standard deviation; VIF: Variance inflation factor
We thank two anonymous reviewers for their valuable comments on the
manuscript. We thank Chi Xu from Nanjing University and Ge He from China
Academy of Urban Planning and Design Shenzhen for their help with the
data collection and GIS application.
This study was supported by the Natural Science Foundation of China
(31500383), the China Postdoctoral Science Foundation (2016 M590476)
and the Program of Natural Science Research of Jiangsu Higher Education
Institutions of China (16KJB180013). None of the authors have any
competing interests in the manuscript.
Availability of data and materials
The sources of all the data sets supporting the results of this article are
included in Additional file 1.
ZYXH, WFdB, FvL and MN designed the study. ZYXH collected the data.
ZYXH and KJH analyzed the data and wrote the first draft. All authors
contributed to revising the article. All authors read and approved the final
The authors declare that they have no competing interests.
Consent for publication
1. Costard S , Wieland B , de Glanville W , Jori F , Rowlands R , Vosloo W , et al. African swine fever: how can global spread be prevented? Philos T Roy Soc B. 2009 ; 364 ( 1530 ): 2683 - 96 .
2. Jori F , Vial L , Penrith ML , Perez-Sanchez R , Etter E , Albina E , et al. Review of the sylvatic cycle of African swine fever in sub-Saharan Africa and the Indian ocean . Virus Res . 2013 ; 173 ( 1 ): 212 - 27 .
3. Torre A , Bosch J , Iglesias I , Muñoz M , Mur L , Martínez‐López B , et al. Assessing the risk of African swine fever introduction into the European Union by wild boar . Transbound Emerg Dis . 2013 ; 62 : 272 - 9 .
4. Sanchez-Vizcaino JM , Mur L , Martinez-Lopez B. African swine fever (ASF): Five years around Europe . Vet Microbiol . 2013 ; 165 ( 1-2 ): 45 - 50 .
5. Mur L , Martínez-López B , Costard S , de la Torre A , Jones BA , Martínez M , et al. Modular framework to assess the risk of African swine fever virus entry into the European Union . BMC Vet Res . 2014 ; 10 ( 1 ): 1 .
6. Sánchez-Vizcaíno J , Mur L , Gomez-Villamandos J , Carrasco L. An update on the epidemiology and pathology of African swine fever . J Comp Pathol . 2015 ; 152 ( 1 ): 9 - 21 .
7. Sanchez-Vizcaino JM , Mur L , Martinez-Lopez B. African swine fever: an epidemiological update . Transbound Emerg Dis . 2012 ; 59 : 27 - 35 .
8. de Glanville WA , Vial L , Costard S , Wieland B , Pfeiffer DU . Spatial multicriteria decision analysis to predict suitability for African swine fever endemicity in Africa . BMC Vet Res . 2014 ; 10 :9.
9. Penrith ML , Vosloo W , Jori F , Bastos ADS . African swine fever virus eradication in Africa . Virus Res . 2013 ; 173 ( 1 ): 228 - 46 .
10. Costard S , Mur L , Lubroth J , Sanchez-Vizcaino JM , Pfeiffer DU . Epidemiology of African swine fever virus . Virus Res . 2013 ; 173 ( 1 ): 191 - 7 .
11. Kalenzi Atuhaire D , Ochwo S , Afayoa M , Norbert Mwiine F , Kokas I , Arinaitwe E , et al. Epidemiological overview of African swine fever in Uganda (2001- 2012) . J Vet Med . 2013 ; 2013 : 949638 .
12. Ravaomanana J , Michaud V , Jori F , Andriatsimahavandy A , Roger F , Albina E , et al. First detection of African swine fever virus in Ornithodoros porcinus in Madagascar and new insights into tick distribution and taxonomy . Parasit Vectors . 2010 ; 3 ( 1 ): 1 - 9 .
13. Jori F , Bastos ADS . Role of wild suids in the epidemiology of African swine fever . Ecohealth . 2009 ; 6 ( 2 ): 296 - 310 .
14. Gulenkin VM , Korennoy FI , Karaulov AK , Dudnikov SA . Cartographical analysis of African swine fever outbreaks in the territory of the Russian Federation and computer modeling of the basic reproduction ratio . Prev Vet Med . 2011 ; 102 ( 3 ): 167 - 74 .
15. Mannelli A , Sotgia S , Patta C , Sarria A , Madrau P , Sanna L , et al. Effect of husbandry methods on seropositivity to African swine fever virus in Sardinian swine herds . Prev Vet Med . 1997 ; 32 ( 3-4 ): 235 - 41 .
16. Fasina FO , Agbaje M , Ajani FL , Talabi OA , Lazarus DD , Gallardo C , et al. Risk factors for farm-level African swine fever infection in major pig-producing areas in Nigeria , 1997 - 2011 . Prev Vet Med . 2012 ; 107 ( 1-2 ): 65 - 75 .
17. Fasina FO , Shamaki D , Makinde AA , Lombin LH , Lazarus DD , Rufai SA , et al. Surveillance for African swine fever in Nigeria, 2006 - 2009 . Transbound Emerg Dis . 2010 ; 57 ( 4 ): 244 - 53 .
18. Olugasa BO , Ijagbone IF , Calistri P , Conte A , Freier J , Ward M. Pattern of spread of African swine fever in south-western Nigeria , 1997 - 2005 . Vet Ital. 2007 ; 43 ( 3 ): 621 - 8 .
19. Plowright W , Thomson G , Neser J. African swine fever . In: Coetzer J , Thomson G , Tustin R , editors. Infectious diseases of livestock , with special reference to southern Africa. Southern Africa: Oxford University Press ; 1994 . p. 567 - 99 .
20. Wilkinson PJ , Pegram RG , Perry BD , Lemche J , Schels HF . The distribution of African swine fever virus isolated from Ornithodoros moubata in Zambia . Epidemiol Infect . 1988 ; 101 ( 3 ): 547 - 64 .
21. Glavanakov S , White DJ , Caraco T , Lapenis A , Robinson GR , Szymanski BK , et al. Lyme disease in New York State: Spatial pattern at a regional scale . Am J Trop Med Hyg . 2001 ; 65 ( 5 ): 538 - 45 .
22. Huang ZYX , de Boer WF , van Langevelde F , Xu C , Ben Jebara K , Berlingieri F , et al. Dilution effect in bovine tuberculosis: risk factors for regional disease occurrence in Africa . Proc R Soc B . 2013 ; 280 ( 1765 ): 20130624 .
23. Jebara KB , Caceres P , Berlingieri F , Weber-Vintzel L. Ten years' work on the World Organisation for Animal Health (OIE) Worldwide Animal Disease Notification System . Prev Vet Med . 2012 ; 107 ( 3-4 ): 149 - 59 .
24. Biotani L. A databank for the conservation and management of the African mammals . Rome: Istituto di ecologia applicata; 1998.
25. Boitani L , Sinibaldi I , Corsi F , De Biase A , Carranza ID , Ravagli M , et al. Distribution of medium- to large-sized African mammals based on habitat suitability models . Biodivers Conserv . 2008 ; 17 ( 3 ): 605 - 21 .
26. Estrada-Peña A , Ostfeld RS , Peterson AT , Poulin R , de la Fuente J. Effects of environmental change on zoonotic disease risk: an ecological primer . Trends Parasitol . 2014 ; 30 ( 4 ): 205 - 14 .
27. Gustafson EJ , Parker GR. Using an index of habitat patch proximity for landscape design . Landscape Urban Plan . 1994 ; 29 ( 2 ): 117 - 30 .
28. Gale P , Drew T , Phipps LP , David G , Wooldridge M. The effect of climate change on the occurrence and prevalence of livestock diseases in Great Britain: a review . J Appl Microbiol . 2009 ; 106 ( 5 ): 1409 - 23 .
29. Gale P , Estrada‐Peña A , Martinez M , Ulrich R , Wilson A , Capelli G , et al. The feasibility of developing a risk assessment for the impact of climate change on the emergence of Crimean‐Congo haemorrhagic fever in livestock in Europe: a Review . J Appl Microbiol . 2010 ; 108 ( 6 ): 1859 - 70 .
30. de Boer WF , Vis MJP , de Knegt HJ , Rowles C , Kohi EM , van Langevelde F , et al. Spatial distribution of lion kills determined by the water dependency of prey species . J Mammal . 2010 ; 91 ( 5 ): 1280 - 6 .
31. Harris I , Jones P , Osborn T , Lister D. Updated high‐resolution grids of monthly climatic observations - the CRU TS3. 10 Dataset . Int J Climatol . 2014 ; 34 ( 3 ): 623 - 42 .
32. Barrios JM , Verstraeten WW , Maes P , Clement J , Aerts J-M , Farifteh J , et al. Remotely sensed vegetation moisture as explanatory variable of Lyme borreliosis incidence . Int J Appl Earth Obs Geoinf . 2012 ; 18 : 1 - 12 .
33. Sun W , Gong J , Zhou J , Zhao Y , Tan J , Ibrahim AN , et al. A spatial, social and environmental study of tuberculosis in China using statistical and GIS technology . Int J Environ Res Public Health . 2015 ; 12 ( 2 ): 1425 - 48 .
34. Phillips SJ , Anderson RP , Schapire RE . Maximum entropy modeling of species geographic distributions . Ecol Model . 2006 ; 190 ( 3 ): 231 - 59 .
35. Donaldson TG , de León AAP , Li AI , Castro-Arellano I , Wozniak E , Boyle WK , et al. Assessment of the geographic distribution of Ornithodoros turicata (Argasidae): climate variation and host diversity . PLoS Negl Trop Dis . 2016 ; 10 ( 2 ): e0004383 .
36. Cohen J , Cohen P , West SG , Aiken LS . Applied multiple regression/ correlation analysis for the behavioral sciences . 3rd ed. NJ: Lawrence Erlbaum Associates Inc.; 2013 .
37. Bates D , Maechler M , Bolker B , Walker S. lme4 : Linear mixed-effects models using Eigen and S4. R package version . 2013 ; 1 ( 4 ).
38. Beltrán-Alcrudo D , Guberti V , de Simone L , De Castro J , Rozstalnyy A , Dietze K , et al. African swine fever spread in the Russian Federation and the risk for the region . EMPRES Watch . 2009 : 1 - 9 .
39. Holt RD , Dobson AP , Begon M , Bowers RG , Schauber EM . Parasite establishment in host communities . Ecol Lett . 2003 ; 6 ( 9 ): 837 - 42 .
40. Lloyd-Smith JO , Cross PC , Briggs CJ , Daugherty M , Getz WM , Latto J , et al. Should we expect population thresholds for wildlife disease? Trends Ecol Evol . 2005 ; 20 ( 9 ): 511 - 9 .
41. Si Y , Wang T , Skidmore AK , de Boer WF , Li L , Prins HHT . Environmental factors influencing the spread of the highly pathogenic avian Influenza H5N1 virus in wild birds in Europe . Ecol Soc . 2010 ; 15 ( 3 ).
42. Gilbert M , Xiao X , Pfeiffer DU , Epprecht M , Boles S , Czarnecki C , Chaitaweesub P , Kalpravidh W , Minh PQ , Otte MJ , et al. Mapping H5N1 highly pathogenic avian influenza risk in Southeast Asia . Proc Natl Acad Sci USA . 2008 ; 105 ( 12 ): 4769 - 74 .
43. Ostfeld RS , Glass GE , Keesing F. Spatial epidemiology: an emerging (or reemerging) discipline . Trends Ecol Evol . 2005 ; 20 ( 6 ): 328 - 36 .