Corrected: Author correction

ARTICLE
https://doi.org/10.1038/s41467-019-12499-6 OPEN

Impacts of environmental and socio-economic

factors on emergence and epidemic potential
of Ebola in Africa
David W. Redding 1*, Peter M. Atkinson2, Andrew A. Cunningham 3, Gianni Lo Iacono 4,

Lina M. Moses 5, James L.N. Wood 6 & Kate E. Jones 1,3*

1234567890():,;

Recent outbreaks of animal-borne emerging infectious diseases have likely been precipitated
by a complex interplay of changing ecological, epidemiological and socio-economic factors.
Here, we develop modelling methods that capture elements of each of these factors, to
predict the risk of Ebola virus disease (EVD) across time and space. Our modelling results
match previously-observed outbreak patterns with high accuracy, and suggest further out-
breaks could occur across most of West and Central Africa. Trends in the underlying drivers
of EVD risk suggest a 1.75 to 3.2-fold increase in the endemic rate of animal-human viral spill-
overs in Africa by 2070, given current modes of healthcare intervention. Future global change
scenarios with higher human population growth and lower rates of socio-economic devel-
opment yield a 1.63-fold higher likelihood of epidemics occurring as a result of spill-over
events. Our modelling framework can be used to target interventions designed to reduce
epidemic risk for many zoonotic diseases.

1 Centre for Biodiversity and Environment Research, Department of Genetics, Evolution and Environment, University College London, Gower Street, London

WC1E 6BT, UK. 2 Lancaster Environment Centre, Lancaster University, Bailrigg Lancaster LA4 1YW, UK. 3 Institute of Zoology, Zoological Society of London,
Regent’s Park, London NW1 4RY, UK. 4 School of Veterinary Medicine, University of Surrey, Guildford, UK. 5 Department of Global Community Health and
Behavioral Sciences, Tulane University, New Orleans, LA, USA. 6 Department of Veterinary Medicine, Disease Dynamics Unit, University of Cambridge,
Cambridge, UK. *email: d.redding@ucl.ac.uk; kate.e.jones@ucl.ac.uk

NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6

L
ittle is known about how the majority of human infectious patterns underlying this outbreak, using case32 and genomic data30
diseases will be affected by predicted future global envir- to demonstrate that EVD spread can be successfully predicted by a
onmental changes (such as climate, land use, human societal dispersal model that is weighted by both geographic distance and
and demographic change)1–5. Importantly, two thirds of human human population density. Attempting to understand zoonotic
infectious diseases are animal-borne (zoonotic)6 and these dis- epidemic risk, however, using a human-only transmission model
eases form a major, global health and economic burden, dis- and without incorporating host ecology would inevitably lead to
proportionately impacting poor communities7,8. Many zoonotic areas with high human density and connectivity being identified as
diseases are poorly understood, and global health responses to the regions with the highest risk, despite some areas of these
them are chronically underfunded9. Our knowledge gaps and the lacking competent hosts. Therefore, to model both the spatial
need for improved forecasting of zoonotic disease risk were variation in spill-over risk and, concurrently, the likely progression
starkly illustrated by the 2013–2016 Ebola outbreak, which was of subsequent outbreaks in human populations, we need to take a
unprecedented in terms of size, financial cost, and geographical system-dynamics modelling approach1,33. Key non-linear feed-
location10,11. backs can also be captured, for example, the trade-off between
Ebola virus disease (EVD) was first identified in 1976, and increasing human populations and any potential loss of reservoir
since then there have been ~23 recognized outbreaks12, pre- host species through anthropogenic land-use conversion. We can
dominantly within central Africa. EVD is caused by any one of use this approach to create a methodological test-bed to design
four pathogenic strains of Ebola virus: Zaire (EBOV), Sudan adaptive vaccination and epidemic preparedness regimes across
(SUDV), Taï Forest (TAFV), and Bundibugyo (BDBV). It pre- Africa that are future-proof with respect to changing ecological and
sents as a non-specific febrile illness that can cause haemorrhagic human landscapes34.
fever, often with a high case fatality rate in diagnosed patients13. Here, using a disease system-dynamics framework (Fig. 1), we
Some Old World fruit bat species (Family Pteropodidae) have develop a discrete-time, stochastic epidemiological compartmental
been suggested as reservoir hosts14, however, while there is lim- model (Environmental-Mechanistic Model or EMM, Fig. 2)
ited direct evidence, they are strong candidates to play a key role incorporating spatial environmental variability to simulate
either as a reservoir or amplifying host15,16. In areas with EVD, present-day spill-over and subsequent human-to-human trans-
there are frequent direct and indirect human-bat interactions, mission using the example of Zaire Ebola virus (EBOV) (the strain
e.g., via bush meat hunting and during fruit harvesting17, pre- responsible for the 2013–2016 outbreak in West Africa). We run
senting numerous opportunities for bat-to-human pathogen spill- the model ~20,000 times with randomly-sampled starting condi-
overs to occur. Additionally, a third of known zoonotic spill-overs tions and critically examine the emergent patterns of infections to
have been connected to contact with great apes and duikers, discover which areas of the world are most at risk, and explore
although there is no evidence that these species act as reservoir currently unsampled effects of Ebola epidemiology in Africa, and
hosts10. It is clear, however, that once spill-over occurs human potential spread of EBOV at global scale based on flight data. As
social factors such as movement and healthcare responses greatly our model is flexible in terms of inputs, we can then examine the
influence the cumulative outcome of an outbreak18. For instance, possible trend of Ebola epidemiological patterns by projecting to a
previous work has highlighted the importance of family interac- future time-point, 207035–37 under a variety of integrated
tions19, funeral practices20 and differential transmission rates in global change scenarios38. We compare the model output from
hospitalized individuals18. three Representative Concentration Pathways (RCP) scenarios of
Many attempts to understand Ebola outbreak dynamics have increasing greenhouse gas concentrations: GCAM-RCP4.5 (‘High
focused on mechanistic modelling approaches of human-to-human Climate Mitigation’), AIM-RCP6.0 (‘Low Climate Mitigation’),
transmission post spill-over from animal hosts13,18,19,21–24. and MESSAGE-RCP8.5 (‘Business as Usual Emissions’)39, and
Mechanistic, or process-based, models are ideal for capturing three possible socio-economic development scenarios (Shared
epidemiological characteristics of diseases and, importantly, testing Socio-economic Pathways or SSP), ordered by increasing human
how disease outbreaks might be impacted by intervention efforts10. population density and reduced regional socio-economic coop-
One downside is that mechanistic models of zoonoses often do not eration: SSP1, SSP2 and SSP3 (termed respectively ‘Sustainable’,
incorporate spatially heterogeneous ecological and environmental ‘Middle of the Road’ and ‘Regional Rivalry’ Development). Using
information, such as the environmental differences leading to these scenarios as reasonable boundaries of possible future global
variation in host suitability25. In this context, correlative, or pat- change, we then identify the prevailing direction of changes to the
tern-based, models (e.g., MaxEnt, Boosted-Regression Trees) have spatial patterns of risk of EBOV outbreaks and epidemics across
been used to simultaneously capture the spatial risk of both zoo- Africa, and subsequent importations of disease cases across the
notic spill-over and subsequent human-to-human infection12. For world via airlines.
some spatially-explicit analyses of Ebola, there have been attempts
to incorporate spatial patterns of human populations and/or air or
other transportation networks26–28, but few studies have con- Results
sidered whole-system analyses for major epidemic zoonoses across Present day patterns. Our EMM simulation for present day
the whole endemic disease area. Like other rare or poorly-sampled EBOV-EVD risk correctly identified areas of observed outbreaks
diseases, Ebola suffers from limited data availability, meaning as high risk, such as Democratic Republic of Congo, Gabon and
approaches using pattern-finding, correlative analytical techniques the 2013–2016 outbreak in West Africa, but also identified some
to find robust models are at a disadvantage29. areas where EVD has not been reported, including most countries
In 2014, a spill-over in Gueckedou district, Guinea of Ebola- in West and Central Africa, especially Nigeria and Ghana
Zaire virus led to an EVD outbreak ~100 times larger than any of (Fig. 3a). Our model outputs predict the observed pattern of
the previous 21 known outbreaks30. Such epidemics have a dis- outbreaks with high accuracy (AUC = 0.83 ± 0.012–95% con-
proportionate impact on the affected societies. For example, the fidence interval (CI) from 10,000 repetitions of 1000 random
World Bank estimates a cost of US$2.2 billion to the three most pseudo-absences), which is significantly higher than 10,000 ran-
affected countries31 due to, amongst other drivers, widespread domised versions of the risk surface (AUC = 0.53 ± 0.028–95%
infrastructure breakdown, mass migration, crop abandonment and CI), a pattern which held irrespective of how the endemic area
a rise in endemic diseases due to overrun healthcare systems. was defined (Supplementary Fig. 1). Since the modelling exercise
Recent work has uncovered the human-to-human transmission was run in 2018, our model has successfully predicted subsequent

2 NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6 ARTICLE

a Physical environment
Climate, land-use g
Livestock density Non-endemic human
Crops, farming methods
Geology, soil characteristics
populations
Socio-
economic

Migration
drivers

b Host niche e Human Governance,

populations Stability,
Climatic niche

Importations
Corruption,
Population dynamics Healthcare
Biotic interactions Endemic human infrastructure,
populations Food availability
Poverty,
Preventative & security, health
c Infected host niche healthcare, policies,
education,
Population mixing d Pathogen immunity
transport,
Immunological variability
spill over religious beliefs
Co-infection
Localised f Infected
behaviour, populations
stochastic
Healthcare response,
processes
human-to-human inf.

Fig. 1 System-dynamics model of zoonotic disease transmission. Letters a–f indicate major system components, arrows showing links, and key sub-
components in smaller font. Within the global physical environment (a), both the host niche (b) and infected host niche (c) are nested subsets, which all
vary over a relatively slow time-scale. Endemic human populations are nested within the global human population (e), with human socio-economic factors
(g) affecting all human populations. Spill-overs happen in the fast-moving spatial and temporal interface of these two nested systems (d), where both
infected hosts, susceptible people and spill-over specific factors occur, resulting in infected human populations (f)

outbreaks (Equateur & Kivu provinces, Democratic Republic of (here we term ‘epidemics’) with high, to very high, numbers of
Congo) and indicate they occurred in areas at high risk of larger cases (1500+). This threshold of assigning an outbreak with
size (1500+ cases) outbreaks (Fig. 3c). Our predictions also greater than 1500 cases as an epidemic also corresponds to the
replicate the pattern where outbreaks in the Gabon hotspot are top 1 percentile of a log-normal distribution approximating the
likely to be smaller in terms of total case size, with those in the variation in pre-2016 observed outbreak sizes (~1538 cases per
east and west having the potential for larger outbreak sizes year). Of the ~2500 simulation runs for present day conditions,
(Fig. 3). Our modelling approach, which focuses on establishing epidemics (>1500) occurred approximately in 5.8% of the yearly
the fundamental conditions where EBOV could reasonably spill- simulations, with catastrophic epidemics (>2,000,000) occurring
over, without using case data, provides further support that the in around 2.3% of simulations, or once every 43.5 years given
at-risk area for EBOV-EVD spill-overs to occur (Fig. 3b; Guinea- current conditions. From the sensitivity testing, the key
Bissau, Liberia, Sierra Leone, Ghana, Togo, Benin, Nigeria, parameters that affected outbreak size were illness length and
Cameroon, Gabon, Equatorial Guinea, Congo, Central African R0, which positively increased case numbers (Supplementary
Republic, Ruanda, Burundi, Kenya and Tanzania) is much larger Fig. 5A), whereas the annual spill-over rate (Supplementary
than the area made up of just those countries that have reported Fig. 5B) was most impacted by the spill-over rate constant
disease outbreaks thus far. We note that our risk map also (strongly positive), shape of the poverty/spill-over curve (weakly
identified high-risk areas that are endemic for the other EVD positive), and by host movement distance (weakly negative).
strains, likely due to similar transmission pathways and reservoir
host characteristics (Fig. 3a). When comparing the mean number
of spill-overs across Africa per year, present day simulations were Future trajectories. Our future EMM simulations suggest that, in
higher (2.464 spill-overs per year 95% CI 2.361–2.567) than the most scenarios, given current projected patterns of global change,
mean historical number over the last 40 years: 0.75 (95% CI there will be a general, ongoing increase in Ebola incidence over
0.695–0.905), reflecting the increase in human population in these time. For example, we estimate an annual increase in the max-
regions during this time period. High risk of Ebola case impor- imum area impacted by the disease from 3.45 million km2 to
tation outside Africa, using the current network of airline flights, 3.8 million km2 under the worst-case scenario by 2070, with
was seen in China, Russia, India, the United States as well as increases in the maximum area seen under all future scenarios.
many high-income European countries (Fig. 4), a similar pattern The maximum area where just spill-overs could occur, however,
to the importations seen during the 2014–2016 outbreak (United increased by just 1% under the GCAM-RCP4.5 SSP1 (High Cli-
States, Spain, Italy, United Kingdom). (Fig. 4). mate Mitigation + Sustainable Development: 2.01 million km2),
Similar to historic data (Supplementary Fig. 4), the distribution when compared to present day (Fig. 3b: 1.99 million km2), but
of the final size of the simulated outbreaks was multimodal with increased by 14.7% from present day under the MESSAGE-
distinct peaks at very low numbers (less than 3 cases) and RCP8.5 SSP3 (Business as Usual Emissions + Regional Rivalry
medium outbreaks (3–1500 cases) (Supplementary Fig. 4). Under Development: 2.29 million km2) scenario. Conversely, the total
extensive simulation, the most common outbreaks were very area where epidemics could start decreased under the GCAM-
small, at odds with the observed data, with singleton or two- RCP4.5 SSP1 by 47% (0.444 million km2), when compared to
person outbreaks frequently observed. We are also able to explore present day (Fig. 3c: 0.836 million km2), but again increases under
the lower probability areas of the distribution effectively and, AIM-RCP6.0 SSP2 (Low Climate Mitigation + Middle of the
unique to the simulation data, there is a third peak of outbreaks Road Development) this time by 20.5%, and then by 34% under

NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6

a
c

Adjacent grid cells

b 0.30
0.25
Grid cells 0.20
0.15
0.10
0.05
Spill-over Present day 0.00
S E I F R

Host probability, H
∈fr ∈fr d
z = kSH γ1 - σ

 σ F
S E I F R

 = ISI + FSF

0.30
0.25
0.20
0.15
0.10
0.05
0.00
Future

Fig. 2 Environmental Mechanistic Model (EMM) EBOV Simulation Schematic. Within 0.0416° (5.6 km at equator) grid cells across the globe (a), we used
a SEIFR (Susceptible, Exposed, Infectious, Funeral and Removed) disease compartmental model (b), to estimate the number of people in each compartment.
S–E transmission rate was determined for each grid cell by calculating the force of zoonotic infection (between hosts and humans) λz, and within human
populations λ (see Methods). Travel of exposed or infectious individuals between grid cells occurred across existing road and flight transport networks,
with transmission rate εfr. Mean transition rates used as the starting parameters for simulations were as follows: α for E–I was calculated as the reciprocal
of incubation time in days (α = 1/7), γσ (I–F transition rate) was the product of the probability of the reciprocal of days infectious (γ = 1/9.6) and maximum
poverty-weighted case fatality rate (σ = 0.78), γ1-σ (I–R transition rate) was the product of the probability of the reciprocal of days infectious (γ = 1/9.6)
and probability of recovering (1-σ), and γF (F–R transition rate) was the reciprocal of the burial time of 2 days. Each simulation was run 2500 times for
365 days, only including grid cells containing a human population. The total number of people in each compartment per grid cell, per day from each
simulation was then used to calculate the total number of index and secondary cases and mapped spatially (c). Bioclimatic, land use and demographic
conditions were then changed to predicted values for 2070 to estimate changes to λ and λz, and the simulations repeated to investigate impacts of global
change on disease (d)

the MESSAGE-RCP8.5 SSP3 scenario. Country-level patterns GCAM-RCP4.5 SSP1 and GCAM-RCP4.5 SSP2 scenarios and a
showed a trend of a general decrease in EVD risk for the better- mean 1.475 spill over a year increase between AIM-RCP6.0 SSP2
case scenarios (e.g. High Climate Mitigation + Sustainable and AIM-RCP6.0-SSP3. In general, the probability of the index
Development) towards an increase in risk, under the worst sce- cases resulting in small outbreaks reduced in future environ-
narios (e.g. Business as Usual Emissions + Regional Rivalry ments, whereas the chance of epidemics increased (Fig. 6). For
Development), with this pattern especially pronounced in West instance, the proportion of epidemics per year (>1500 cases)
Africa (Fig. 5). decreased in the GCAM-RCP4.5 SSP1 to 3.43% (from 5.8%
The increases seen in the area affected is mirrored by greater in present day) but increased in all others, with AIM-RCP6.0
total numbers of spill-overs experienced in future scenarios, SSP3 gaining the greatest number, with epidemics in 9.5% of
with the greatest increase seen under the MESSAGE-RCP8.5 all simulations. The number of catastrophic epidemics
SSP3 scenario at 7.92 (CI 7.62–8.19) spill-overs per year. Spill- (>2,000,000), generally increased with both RCP and SSP values
over numbers increased with greenhouse gas concentrations up to 3.43% and 3.54% for the AIM-RCP6.0 and MESSAGE-
(represented here by the RCP value) with a mean 0.257 spill over RCP8.5 SSP3 scenarios respectively, but again saw a decrease
a year increase between the GCAM-RCP4.5 SSP2 (High Climate from the present-day level (2.3 %) to 1.19% in the ‘best case’
Mitigation + Middle of the Road Development) and AIM- future scenario (GCAM-RCP4.5 SSP1).
RCP6.0 SSP2 scenarios, and a mean 0.343 spill over a year
increase between the AIM-RCP6.0 SSP3 (Low Climate Mitiga-
tion + Regional Rivalry Development) and MESSAGE-RCP8.5 Discussion
SSP3 scenarios. Greater increases were seen, however, with SSP Our Ebola system-dynamics model, in the absence of case data, is
change, with a mean 1.297 spill over a year increase between able to recapitulate the known endemic area of EVD accurately,

4 NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6 ARTICLE

a b

c
0.30

0.25

0.20

0.15
10,000 cases
0.10

500 cases 0.05

Mean no.of cases
10 cases 0.00 per simulation

Fig. 3 Present day risk of EVD cases caused by Zaire Ebola virus (EBOV) from EMM simulations. Maps represent the mean number of EVD-EBOV cases
between zero (dark blue) and 0.3 (yellow) per grid cell (0.0416°—5.6 km at equator) across 2500, 365-day simulation runs for the present day, where (a)
shows all cases (both index and secondary), (b) index cases only, and (c) index cases from epidemics (1500+ cases). White crosses in (a) represent log
outbreak size. White symbols in (b) represent all locations of known EVD index cases from different viral strains, where circles represent Zaire (EBOV),
square Sudan (SUDV), triangles Taï Forest (TAFV), and tetrahedrons Bundibugyo (BDBV). Diagonal white cross in (c) represents location of index case
from 2014–2016 epidemics, white stars the locations of Ebola outbreaks that have occurred since the modelling was run

No. of importations
per simulation
0.4

0.3

0.2

0.1

0.0

Fig. 4 Most common country locations for importation of EBOV infected individuals. Map shows the number of importations per simulation that countries
outside Africa received via airline flights. Countries with the most EBOV infected individuals imported are represented in yellow with darker green, then
blue, coloured countries having proportional fewer importations and dark blue showing zero importations and the EVD endemic area. Data come from
2500 simulations of EVD outbreaks under present data climate, land-use, demographic and transportation conditions

but also suggests that other areas of the continent, such as West quantify the likelihood of the 2014–2016 West African outbreak
and East Africa are also potentially at risk. Despite there being reoccurring, the risk of an outbreak in this region should be
only two known spill-overs, there is a paucity of information to considered valid. These findings support previous work that has
discount EVD risk in West Africa, and given the lack of data to suggested that several countries in Africa could be at risk despite

NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6

RCP 8.5

0.06

0.04

0.02
RCP 6.0

0.00

–0.02

–0.04

–0.06
RCP 4.5

Mean change
in probability
of Ebola case

SSP1 SSP2 SSP3

Fig. 5 Change in future risk of EVD cases caused by Zaire Ebola virus (EBOV) for 2070. Maps represent mean change in per grid cell (0.0416°—5.6 km at
equator) EVD case probability from zero (yellow) to −0.06 (green) and 0.06 (red), aggregated at the country level with data from EMM simulations for
2070. Rows and columns show all reasonable combinations of the different scenarios of global change (GCAM-RCP4.5, AIM-RCP6.0, MESSAGE-RCP8.5
and SSP1 to 3)

having experienced no known cases16,40,41. That our results here We identify Nigeria (but also Ghana, Kenya and Rwanda) as
differ from previous case-driven analyses is unsurprising given not only a key area for epidemics to be initiated, but also an area
minimal case data available to build correlation-based models and with potential for many small outbreaks, somewhat contradicting
common problems associated with small data sets, such as model analyses based on current case data16,40,45. Our finding could
instability, where many differently specified models have similar indicate that our underlying model has not correctly balanced the
likelihoods, and biases caused by the limited geographic coverage impact of healthcare infrastructure on disease spread, that there
of case reports42. Indeed, the disagreement, for instance, we see are regional behavioural barriers to infection, or that there are
between the observed data and simulation outputs, in terms of the strong regional differences in effective contact rates between both
frequency of small infection events, could also be due to a humans and hosts. Ebola outbreaks are rare, but potentially high-
minimum detection threshold for outbreak size, especially given impact events, and our experience of endemic health care systems
very high numbers of fevers being misdiagnosed as malaria43. being challenged by EVD cases is, as yet, limited to a handful of
While vaccinations and exhaustive health care efforts44 have been data points, meaning it is unclear what the true relationship is
effective to contain recent outbreaks, the sporadic spatial and between health investment and EVD risk reduction. For Nigeria,
temporal nature of spill-over events has meant it is currently therefore, though we may be underestimating the role of the
unclear where preventative health care infrastructure should be health infrastructure46, given just a single replication of Ebola
best targeted. Our risk surfaces can be useful here, for instance, as importation it would be unwise to assume the same outcome
we show that some parts of the endemic area have a higher every time, especially given the thousands of different potential
chance of turning into very large outbreaks (Congo, Uganda, transport routes for infected people into Nigeria. An extreme
southern parts of West Africa). Also, the scale of our simulations discontinuity in the spatial distribution of Ebola virus ser-
is very flexible and, therefore, our framework could be employed oprevalence or pathogenicity of competing strains in Nigeria, and
adaptively in real outbreak situations. For each outbreak, surrounding countries, could explain the observed lower-risk
for instance, we can simulate the expected spread of the disease pattern, but it would be prudent to prove this is the case using
and as vaccinations occur it would be possible to update the extensive bio-molecular surveys. Certainly, the seroprevalence of
spatial inputs and rerun the model. Future work could, Ebola virus in wildlife species is a key missing dataset that could
therefore, focus on making our model an available tool for health substantially aid future understanding of this disease. Until these
care professionals. additional factors are explicitly tested, we believe the high human

6 NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6 ARTICLE

1500+ 3–1500 <3 there is potential to use spatial interpolation techniques to make
reasonable predictions in unsampled areas. Our model does not
2015 present incorporate this data or test its impact and, similarly due to lack
of data resources, we do not use information about local differ-
ences in funeral practices. Hospital compartments are thought to
be useful to understand quarantine and super-spreading events,
2070 RCP4.5 SSP1 but there is very limited data on the quality and geographic reach
of small and temporary-response health clinics. Some other
important behavioural trends are not captured in our model, such
as the post-outbreak behavioural reactions of human populations
2070 RCP4.5 SSP2 e.g. mass migration away from affected regions, or, avoidance of
clinics and treatment centres due to transmission fears. Recent
findings regarding the persistence of Ebola virus in semen of
2070 RCP6 SSP2
convalescent men may also help explain the intermittent spatio-
temporal patterns of infections in endemic areas48,49. Future
work incorporating such data, may further improve the spatial
resolution and accuracy of risk estimates.
2070 RCP6 SSP3 Looking to understand the possible current direction of
changes to Ebola burden over time, we project our models to
2070 and show that changes to SSP scenarios, which control levels
of poverty and human population size in our models, will likely
2070 RCP8.5 SSP3 have a greater impact than climate and land-use change (here
mediated via RCP scenarios). This is not surprising, as poverty
reduction increases the presumed EVD-EBOV healthcare
response in our simulations, and many of the countries in the
–20 –10 0 10 20
endemic region are expected to have substantial reductions in
Standarised residual
poverty levels by 207036. Similarly, contact rates in our simulation
Fig. 6 Comparison of 2070 EMM simulation scenarios by EVD-EBOV final (both between humans and between humans and wildlife) depend
outbreak size. Circles represents standardized residuals from a chi-squared linearly on human population growth, whereas climate change
test of association (χ = 466.27, df = 10, p < 0.001) between simulation increases EVD-EBOV cases through more complex interactions.
scenario and final outbreak size category. More orange/red colours show Species distribution models indicate that the presumed wildlife
a greater than expected (vs. randomly allocated) number of outbreaks for hosts prefer warm and wet conditions (Supplementary Figs. 2, 3),
any given combination of scenario and final outbreak size, with more blue which are expected to increase in these regions according to the
colours representing fewer than expected outbreaks. Size of circle also HADGem2-AO climate model37 (Supplementary Fig. 6). This
indicates the overall quantity different to expected, with large circles model, when compared to the predicted host change from 32
contributing more to the overall chi-value compare to smaller circles other models represents the ‘middle ground’ in terms of host
spatial occurrence change (Supplementary Fig. 7), but future
work should further examine how climate model choice impacts
density and known presence of putative wildlife hosts means that future predictions. This expansion of the optimal conditions for
all the at-risk areas we present here should remain as candidate presumed the wildlife host species effectively increases the at-risk
locations for outbreaks. As an illustrative example, we recover human population by including more of the northern, eastern
Guinea as being at high risk of large outbreaks but if our outputs and southern areas of Africa (Fig. 3a). Predicted future anthro-
were, for instance, validated against data prior to 2013 it would pogenic land-use changes, however, reduces the optimal wildlife
appear a false prediction. host habitat, thereby reducing human-wildlife interactions. In
There is a similar lack of information to fully understand the terms of global socio-economic development, EVD cases num-
true distribution of global EVD importation risk, as there have bers would appear to respond best to efforts in slowing the rate of
only been a handful of documented importations observed so far. population growth and increasing socio-economic development
The pattern we report is effectively a recreation of the shape of while mitigating climate change, such that global change most
airline network when flying from locations that commonly closely tracks the GCAM-RCP4.5 SSP1 scenario (High Climate
experience outbreaks. Any errors made in delineation of the Mitigation + Sustainable Development). Global binding com-
dominant areas that experience outbreaks will, therefore, be mitments to reducing climate change may act to slow the effects,
propagated into this prediction. Future EVD importation risk will but evidence50 suggests a wholesale change is unlikely. Efforts to
be an emergent property of the interactions between areas of decreases poverty in Central and Western Africa with a con-
Africa that are likely to experience the disease, the healthcare comitant increase in healthcare resources, therefore, would
response in terms of treatment and containment, and the shape appear to be the most realistic approach to reducing future EVD
and size of the airline network at that time. Future research disease risk globally.
could, using our approach, model the likelihood of index cases Our approach demonstrates not only an important framework
occurring in different parts of Africa managing to reach every for understanding Ebola but also for other diseases, given our
country given different sets of healthcare interventions. modelling framework is disease agnostic and flexible in terms of
To further improve our model, we would need to better compartment model structure, transmission location and scale.
understand the spatial variation in other key disease transmission Analysing diseases singly, and within just an epidemiological or
parameters. For instance, bush-meat hunting is an important socio-economic framework, cannot be an effective approach for
process by which human populations come into contact with policy making at a large geopolitical scale, particularly in regions
large bats47 and the spatial variation in bush-meat extraction is with multi-disease burden and limited healthcare resources. Net
likely a component of spill over variation. Little is known, how- disease risk patterns, when summed across a wide variety of
ever, about bush-meat hunting outside a few specific studies, but zoonoses, will be an emergent property of the distribution of very

NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6

different wildlife host species and their respective responses to single value representing the probability of any reservoir species being present and
increasing anthropogenic land-use conversion and climate a similar approach was taken for the non-reservoir host species. First, we assume
that complete mixing occurs within each grid cell and that if secondary host species
change. Any lack of data in the short-term does not reduce the are present in a cell they meet the presumed primary hosts. Then we calculate one
obvious importance of understanding future disease trends. minus the probability of each host presence, multiplied across all species, giving us
Interdisciplinary methods, such as ours, establish a first heuristic the probability of there being no species present in any given cell. One minus this
step on a pathway towards building tools to devise future-proof value gives the probability of at least one species being present. We assumed that
intervention measures aimed at reducing overall future disease bats were the default, fundamental component for transmission (i.e., acting as the
‘working’ reservoir) and, given complete mixing in each cell, apes and duikers were
burden. a secondary, less common route of human infection. Since roughly a third of index
cases have been attributed to ape/duiker host spill-overs10, in cells with both groups
we down-weighted the probability of the ape/duiker occurrence by two thirds and
Methods reservoir occurrence by one third when combining the two lots of layers together,
Analysis outline. Focusing on the Zaire Ebola virus (EBOV), we use a stochastic though, we note, due to the similar habitat requirements a very high majority of the
epidemiological compartmental model29, to simulate both pathogen spill-over and cells containing duikers and apes also had high probability for bats so the precise
subsequent human-to-human disease transmission (Fig. 2). Within grid cells value here will have a limited impact. The final resulting probability was bounded
(0.0416°—5.6 km at equator) covering continental Africa, we used a Susceptible, by zero and one. Additionally, as EBOV presence in non-reservoir host species is
Exposed, Infectious, Funeral and Removed (SEIFR) EVD-EBOV disease com- impossible without the presence of reservoir hosts, cells with a reservoir host
partmental model following13,19,23 to estimate the number of individuals per probability of zero were given a value of zero irrespective of the non-reservoir host
compartment, in each time step t, for present day bioclimatic, land use and score. For computational simplicity, we assume that all human individuals have
demographic conditions. Although some previous compartmental models for equal chance of exposure to infected host species. The initial value used for spill-
EBOV have included a Hospital compartment51, adding this complexity was not over rate κ, per time step t, was estimated from the number of historic outbreaks O
feasible over large and poorly known geographical areas. Without knowing more (defined here as distinct clusters of cases) taken from empirical EBOV outbreak
about the spatial variation in health seeking behaviour, exactly which grid cells data12, and the number of historically susceptible individuals Sh (inferred from
contain clinics, and the variation of healthcare resources in these clinics, adding in human population estimates from 1976 to 2015 from36 (see Supplementary
this compartment would not likely significantly improve our model’s ability to Equation 3). During each simulation run, κ was allowed to vary using the same
predict the progression of outbreaks. Furthermore, hospital interventions had the method as for the compartmental transmission parameters above.
least impact controlling EVD outbreaks in a recent meta-analysis (24). Grid cell (b) Force of Infection, λ. The force of infection for human-to-human
size was chosen as the highest resolution at which computation was feasible while transmission λ per time step t, was estimated as the product of the effective contact
being able to use a non-stochastic human movement model to approximate con- rate β, and the number of individuals that can transmit the disease in each relevant
tacts per cell (see details below). All mapping and analyses were carried out in R compartment (‘Infectious’ and ‘Funeral’) per grid cell (Supplementary Equation 4).
v.3.2.252. Each stage of the EMM simulation is discussed in more detail below: We assumed that β for the Infectious and Funeral compartments was equivalent,
due to the simplicity of the movement model used as we would not be able to
Stage 1: SEIFR compartmental model within grid cells. We used starting EBOV reasonably differentiate between the two compartments and maintain
transmission characteristics of incubation time = 7 days, onset of symptoms to computational efficiency. Indeed, with the contact rates of moving individuals from
resolution = 9.6 days, maximum case fatality rate for very low income countries the Infectious compartment being offset by aggregations of individuals at funerals
(CFR) σ = 0.78, and burial time = 2 days23 to parameterize the SEIFR compart- it is not clear if there would be a large difference when approximated using a gas
mental model to determine transition rates α (between Exposed to Infectious model. We estimated the effective contact rate β, as the basic reproduction number
compartments), γσ (Infectious to Funeral), γ1–σ (Infectious to Removed), and γF R0 divided by the product of the total number of individuals N, and infectious
(Funeral to Removed) (Fig. 2). To incorporate sensitivity around these transmis- duration D (the sum of Infectious and Funeral compartment time, 11 days taken
sion parameters, we allowed values to vary for each simulation run by sampling from23). As a starting value for R0 we used a value of 1.757 and this was allowed to
from a Gaussian distribution where the mean was their initial value and the vary per simulation run using the same method as for the compartmental
standard deviation was fifth of the mean, to give a reasonable spread of values. For transmission parameters above. As per previous research29, we incorporated spatial
each time step t, the number of individuals moving between all compartments was variance in contact rates among grid cells using a weighting factor m, whereby the
estimated by drawing randomly from a binomial distribution (Supplementary effective contact rate in grid cells with greater than expected contact rates was
Equation 1), parameterized using the respective compartmental transition rates. increased and decreased where fewer contacts were predicted (Supplementary
Transition rates for compartments were assumed to be the same in all grid cells Equation 5). We estimated m by creating an ideal free gas model of human
except for the transition between Susceptible to Exposed. The per grid cell Sus- movement within each grid cell and approximated collision frequency per person
ceptible to Exposed transition rates were determined by the force of zoonotic per day, using the following: the total individuals in each grid cell (estimated from
infection λz, and the force of infection λ (Fig. 2) and these were calculated as Gridded Population of the World v358, an individual interaction sphere of radius
follows: 0.5 m, and the per person, daily walking distances in metres vΔt, where v is walking
(a) Force of Zoonotic Infection, λz. The force of infection for zoonotic velocity, and Δt equals time period (Supplementary Equation 6). To capture
transmission λz, per time step t, was estimated as the product of the probability of geographic variation in human movement patterns, each grid cell was assigned a
host presence H, and spill-over rate κ (Supplementary Equation 2). Without any value for per person daily walking distance, based on the empirical relationship
evidence to the contrary15,53, we parameterized H by calculating the spatial between daily walking distances and per person per country Gross Domestic
probability of the presence of the most likely EBOV reservoir host species based on Product (measured as Purchasing Power Parity from36) (Supplementary Table 3).
available data (Old World fruit bat species Epomophorus gambianus gambianus, As the availability of mass transit as an alternative to walking tends to be centrally
Epomops franqueti, Hypsignathus monstrosus, and Rousettus aegyptiacus see controlled, we assumed that grid cells in each country had the same value.
Supplementary Table 1) within each grid cell across the African continent using Under observed conditions, the effective reproduction number Re decays over
species distribution models (SDMs)54 and assuming constant pathogen prevalence. time as both efforts are made to control disease spread and the pool of susceptible
We also calculated the spatial probability of the presence of other species which are reduces, which results in R0 being equal to Re only when time t is zero. Therefore,
known to provide an alternative route of infection, but likely do not act as to calculate effective contact rate β, we allowed Re to decay per time step t
reservoirs (Gorilla spp., Pan spp., and Cephalophus spp.)12. SDMs for each species (Supplementary Equations 7, 8 and 9). However, countries that can invest more in
were inferred using boosted regression trees (BRT) using distribution data from the health infrastructure (e.g., barrier nursing, surveillance) should see a more rapid
Global Biodiversity Information Facility (GBIF)55 and 11 present day bioclimatic reduction in Re over time compared to countries that do not have such
and land use variables (Supplementary Table 2). Data with coarse scale GBIF infrastructure and also a concomitantly, a decrease in CFR. Therefore, we derived
spatial coordinates (decimal degree coordinates with less than four decimal places) an empirical estimate of the relationship between wealth (measured using GDP-
were filtered out of the analysis. To reduce spatial autocorrelation and duplicate PPP per capita) and both the relative rate of decay of Re over time (Supplementary
records, any records that co-occurred in the same grid cell were removed. Lastly, Equation 10) and CFR (Supplementary Equation 11). Using a spatially
GBIF records older than 1990 were discarded to ensure samples more closely disaggregated poverty data layer59 we weighted the per grid cell per time step Re
matched the current landscapes. BRT tree complexity was set at 5 reflecting the reduction and CFR accordingly to the values in each grid cell. While we found the
suggested value and the learning rate was adjusted until >1000 trees were relationship between wealth and both Re and CFR reduction over time to be best
selected56. A total of 25 models were estimated for each species using four-fifths of described using curves with exponents of −0.08 and −0.02, respectively, this was
the distribution data as a training dataset and one fifth as a testing dataset, chosen inferred using relatively few data points (Supplementary Table 4). In our
randomly for each model. Those with the highest predictive ability (high area simulation runs, therefore, we allowed these exponents to vary similarly to the
under operating curve, AUC—all models AUC > 0.9) were selected as the best parameters above, to allow either more linear declines or deeper curves to best
model for each species (Supplementary Fig. 2). The most important spatial estimate the true impact of this relationship.
variables determining distributions across the different reservoir host species were
BIO7 Temperature Annual Range, BIO13 Precipitation of Wettest Month, BIO2
Mean Diurnal Temperature Range and Land Use-Land Cover (Supplementary Stage 2: SEIFR compartmental model between grid cells. We allowed those
Fig. 3). The outputs from all putative reservoir (bat) species were combined into a individuals that had contracted EBOV to travel between grid cells (specifically

8 NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6 ARTICLE

individuals in Exposed and Infectious but not Funeral compartments) (Fig. 2), but number, Re. We inferred human population estimates per grid cell for 2070 by
assumed for simplicity that the overall net movement of susceptible individuals using the Gridded Population of the World v458 for present day and multiplying
between cells was zero. As previously supported with empirical data, we employed each cell by the expected future proportional change over that time period
a movement model that was weighted by both geographic distance and human predicted by three Shared Socio-economic Pathways: SSP1, SSP2 and SSP3.
density30,32, and was also geographically constrained to known transportation Specifically, these pathways represent: (i) SSP1 (Sustainable Development)—
routes. The transmission rate ε, of individuals between target compartments of high regional cooperation, low population growth due high education and high
different grid cells was estimated by two different methods: between grid cells along GDP growth, (ii) SSP2 (Middle of the Road Development)—a ‘processes as
road networks εr, and along flight routes εf. We sampled randomly, from a bino- usual’ scenario with ongoing levels of population growth and wealth, with
mial distribution, the number of travellers per grid cell and time step t (Supple- medium estimates for both these by 2070, and (iii) SSP3 (Regional Rivalry
mentary Equation 1) with the probability of travel by road per day εr, being Development)—regional antagonism, high population growth, unsustainable
proportional to the distance to the nearest road using the Global Roads Open resource extraction and low economic growth. Future poverty estimates per
Access Data Set (Global Roads Open Access Data Set from60). Global roads dataset country were similarly inferred using a spatially-disaggregated GDP layer59
contains in total 585413 routes from tracks to multi-lane highways and has been multiplied by the expected change in per country GDP over the time period as
extensively validated for Africa60. We allowed travellers to move freely (agnostic to predicted by the SSP integrated scenario, with any future changes in wealth
any particular transportation method or country boundary) across the continent interacting with the Re parameter to affect disease epidemiology accordingly. We
up to 10 road junctions in any directionalong the road network starting from the note that as our travel probability is defined per person, increasing future
centroid of the target cell (Global Roads Open Access Data Set from60), giving a populations will see a proportional increase in the amount of both road and air
potential of up to 500 km of linear travel per time step. Each proposed travel end travel, though with unknown patterns of future trade and travel we kept the
point was given an individual probability from the daily distance travelled prob- travel network the same shape.
ability curve from Fig. 2f of ref. 61, which is derived from transport data and (c) Comparison of simulation runs. We reran the EMM simulations under
validated against mobile phone data. For air travel, we set the potential pool of 5 plausible combinations of 2070 future environmental and socio-economic
travellers as the individuals in grid cells containing airports across the world from scenarios of global change and greenhouse gas concentrations, matching greater
Open Flights Airport Database62, plus all the Exposed individuals in the 8 grid cells greenhouse gas emissions to less progressive and less cooperative socio-economic
surrounding each airport grid cell. We sampled randomly from a binomial dis- scenarios as follows: GCAM-RCP4.5/SSP1 (High Climate Mitigation + Sustainable
tribution the number of travellers per grid cell and time step t (Supplementary Development), GCAM-RCP4.5/SSP2 (High Climate Mitigation + Middle of the
Equation 1) with the probability of travel by air per day εf, being proportional to Road Development), AIM-RCP6.0/SSP2 (Low Climate Mitigation + Middle of the
the total number of flights per day divided by the population of that country36. We Road Development), AIM-RCP6.0/SSP3 (Low Climate Mitigation + Regional
allowed travellers to move up to 2 edges on the current airline routes from airport Rivalry Development), MESSAGE-RCP8.5/SSP3 (Business as Usual Emissions +
origin using62. This approximates a traveller taking either a one or two-legged Regional Rivalry Development)38,39,64,67. For each of the six scenarios we aimed for
journey. Final destinations were sampled at random, based on all potential air 2500 runs of 365 days, each day measuring the number of spill-overs, the number
routes having equal priority, but in most cases potential destinations were located of secondary cases associated with each spill-over, and the geographical areas
nearby, which by default meant that more distance travel was less likely than travel affected. This allowed us to measure likelihood of spill-overs leading to small,
to a nearby location. We decided to keep each edge on the network as equal medium and very large outbreaks, and also to determine the geographical areas
likelihood due to a lack of comprehensive and detailed information that we could with the highest risk of experiencing cases. We also noted the destination of any
find on passenger numbers at the time of modelling. For both road and air tra- flights out of Africa that contained infected people.
vellers, individuals were then added to the correct compartment of their final
destination in the new grid cell and removed from the same compartment of the
Reporting summary. Further information on research design is available in
original source grid cell.
the Nature Research Reporting Summary linked to this article.

Stage 3: Impact of future anthropogenic change. (a) Future force of zoonotic Data availability
infection λz. We recalculated values of the force of zoonotic infection λz, by esti- The datasets generated during and/or analysed during the current study are available in
mating the probability of EBOV host presence, H2070 under several different future the figshare repository, https://doi.org/10.6084/m9.figshare.1559959
integrated scenarios that incorporate projections of bioclimatic and land use
variables (Supplementary Table 2). Estimates of bioclimatic variables for 2070 were
based on the HADGem2-AO climate model37 under three Representative Con-
Code availability
Code used during and/or analysed during the current study are available from the
centration Pathways: CAM-RCP4.5, AIM-RCP6.0, and MESSAGE-RCP8.539,63,64.
corresponding author on reasonable request.
These different options were, specifically: (i) GCAM-RCP 4.5 (High Climate
Mitigation)—stabilization scenario in which total radiative forcing is stabilized
shortly after 2100, (ii) AIM-RCP 6.0 (Low Climate Mitigation)—stabilization Received: 30 May 2018; Accepted: 13 September 2019;
scenario in which total radiative forcing is stabilized shortly after 2100, without
Published online: 15 October 2019
overshoot, by the application of a range of technologies and strategies for reducing
greenhouse gas emissions (iii) MESSAGE-RCP8.5 (Business as Usual Emissions)—
worsening scenarios with ongoing increasing, unchecked, greenhouse gas emissions
over time, leading to high greenhouse gas concentration levels. Although we only
used a single overall climate model (HADGem2-AO) due to computational con-
straints, this model offers good agreement with other alternative models65 and a References
‘middle of the road’ option in terms of realised changes to future host distributions 1. Whitmee, S. et al. Safeguarding human health in the Anthropocene epoch:
compared to 32 other models (Supplementary Fig. 7; Supplementary Table 5). To report of The Rockefeller Foundation–Lancet Commission on planetary
estimate host presence probability in the future we needed to predict fine-scale health. The Lancet, https://doi.org/10.1016/S0140-6736(15)60901-1 (2015).
future habitat data under the RCP scenarios. As only coarse land-use categorisa- 2. Civitello, D. J. et al. Biodiversity inhibits parasites: Broad evidence for the
tions are currently available66 with, for instance, a ‘primary’ land-use having a wide dilution effect. Proc. Natl Acad. Sci. USA 112, 8667–8671 (2015).
variety of possible natural habitats from arctic tundra to tropical rainforest, we 3. Costello, A. et al. Managing the health effects of climate change. Lancet 373,
therefore separately empirically estimated future land use-land cover (LULC) 1693–1733 (2009).
change using the spatiotemporal patterns contained within the MODIS land-cover 4. Lafferty, K. D. Calling for an ecological approach to studying climate change
time series35. For each grid cell, at the same grid cell resolution as set out above, we and infectious diseases. Ecology 90, 932–933 (2009).
calculated the frequency of each LULC change seen in the 2001–2012 MODIS 5. Keesing, F. et al. Impacts of biodiversity on the emergence and transmission of
dataset for the surrounding square of 25 grid cells around each grid cell. We infectious diseases. Nature 468, 647–652 (2010).
converted this frequency to a probability by dividing by the total possible number 6. Taylor, L. H., Latham, S. M. & Woolhouse, M. E. Risk factors for human
of changes, and based on these probabilities, we simulated yearly LULC change disease emergence. Philos. Trans. R. Soc. Lond., Ser. B: Biol. Sci. 29, 983–989
across the region of interest for each grid cell from 2012 until 2070, and ran this (2001).
simulation 100 times to create a bank of future possible landscapes. These were 7. Grace, D. et al. Mapping of poverty and likely zoonoses hotspots (International
then summarized into three consensus landscapes representing low (with Livestock Research Institute, Nairobi, Kenya, 2012).
anthropogenic changes rejected where possible), medium (by choosing the majority 8. Hotez, P. J., Fenwick, A., Savioli, L. & Molyneux, D. H. Rescuing the bottom
consensus across all 100 runs) and high anthropogenic change, (more anthro- billion through control of neglected tropical diseases. Lancet 373, 1570–1575
pogenic changes (e.g. urban, cropland, mosaic cropland) were preferentially chosen (2009).
across the landscape) and we aligned these three scenarios to SSP1, SSP2 and SPP3, 9. Hotez, P. J. et al. Control of neglected tropical diseases. New Engl. J. Med. 357,
respectively (for scenario details see below)38,39. 1018–1027 (2007).
(b) Future force of infection λ. Using predicted human demographic variables 10. Alexander, K. A. et al. What factors might have led to the emergence of Ebola
and poverty levels for 2070, we recalculated values for the force of infection λ, by in West Africa? PLoS Neglected Tropical Dis. 9, e0003652 (2015).
estimating the number of individuals per grid cell, n and effective reproduction

NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6

11. Centre for Disease Control. 2014 Ebola Outbreak in West Africa, http://www. 42. Redding, D. W., Lucas, T. C., Blackburn, T. M. & Jones, K. E. Evaluating
cdc.gov/vhf/ebola/outbreaks/2014-west-africa/index.html (2015). Bayesian spatial methods for modelling species distributions with clumped
12. Pigott, D. M. et al. Mapping the zoonotic niche of Ebola virus disease in and restricted occurrence data. PloS ONE 12, e0187602 (2017).
Africa. eLife 3, e04395 (2014). 43. Dalrymple, U. et al. Quantifying the contribution of Plasmodium falciparum
13. Legrand, J., Grais, R. F., Boelle, P.-Y., Valleron, A.-J. & Flahault, A. malaria to febrile illness amongst African children. Elife 6, e29198 (2017).
Understanding the dynamics of Ebola epidemics. Epidemiol. Infect. 135, 44. Wells, C. R. et al. Ebola vaccination in the Democratic Republic of the Congo.
610–621 (2007). Proc. Natl Acad. Sci. USA 116, 10178–10183 (2019).
14. Leroy, E. M. et al. Fruit bats as reservoirs of Ebola virus. Nature 438, 575–576 45. Peterson, A. T., Bauer, J. T. & Mills, J. N. Ecologic and geographic distribution
(2005). of filovirus disease. Emerg. Infect. Dis. 10, 40 (2004).
15. Leendertz, S. A. J. Testing new hypotheses regarding ebolavirus reservoirs. 46. Shuaib, F. et al. Ebola virus disease outbreak—Nigeria, July–September 2014.
Viruses 8, 30 (2016). Mmwr. Morbidity Mortal. Wkly. Rep. 63, 867 (2014).
16. Pigott, D. M. et al. Updates to the zoonotic niche map of Ebola virus disease in 47. Mickleburgh, S., Waylen, K. & Racey, P. Bats as bushmeat: a global review.
Africa. Elife 5, e16412 (2016). Oryx 43, 217–234 (2009).
17. Mickleburgh, S. P., Hutson, A. M. & Racey, P. A. Old World fruit bats. An 48. Deen, G. F. et al. Ebola RNA persistence in semen of Ebola virus disease
action plan for their conservation. Gland, Switzerland: IUCN (1992). survivors—final report. New Engl. J. Med. 377, 1428–1437 (2015).
18. Chowell, G., Hengartner, N. W., Castillo-Chavez, C., Fenimore, P. W. & 49. Mackay, I. M. & Arden, K. E. Ebola virus in the semen of convalescent men.
Hyman, J. M. The basic reproductive number of Ebola and the effects of Lancet Infect. Dis. 15, 149–150 (2015).
public health measures: The cases of Congo and Uganda. J. Theor. Biol. 229, 50. Tompkins, E. L. & Neil Adger, W. Defining response capacity to enhance
119–126 (2004). climate change policy. Environ. Sci. Policy 8, 562–571 (2005).
19. Kiskowski, M. A. A three-scale network model for the early growth dynamics 51. Rivers, C. M., Lofgren, E. T., Marathe, M., Eubank, S. & Lewis, B. L.
of 2014 west Africa ebola epidemic. PLoS Curr 6, http://currents.plos.org/ Modeling the Impact of Interventions on an Epidemic of Ebola in Sierra
outbreaks/index.html%3Fp=44141.html (2014). Leone and Liberia. PLoS Curr. 6, https://doi.org/10.1371/currents.
20. Weitz, J. S. & Dushoff, J. Post-death transmission of Ebola: challenges for outbreaks.1374d1341fe1375d1376c1305e1379df1330ddce1333c1366d1084c
inference and opportunities for control. Sci. Rep. 5, 8751 (2015). (2014).
21. Althaus, C. L., Low, N., Musa, E. O., Shuaib, F. & Gsteiger, S. Ebola virus 52. R: A language and environment for statistical computing (R Foundation for
disease outbreak in Nigeria: transmission dynamics and rapid control. Statistical Computing, Vienna, Austria, 2009).
Epidemics 11, 80–84 (2015). 53. Leendertz, S. A. J., Gogarten, J. F., Düx, A., Calvignac-Spencer, S. & Leendertz,
22. Nishiura, H. & Chowell, G. Early transmission dynamics of Ebola virus disease F. H. Assessing the evidence supporting fruit bats as the primary reservoirs for
(EVD), West Africa, March to August 2014. Eurosurveillance 19, 20894 Ebola viruses. EcoHealth 13, 18–25 (2016).
(2014). 54. Phillips, S. J., Anderson, R. P. & Schapire, R. Maximum entropy modelling of
23. Valdez, L. D., Aragão Rêgo, H. H., Stanley, H. E. & Braunstein, L. A. species geographic distribution. Ecol. Model. 190, 231–259 (2006).
Predicting the extinction of Ebola spreading in Liberia due to mitigation 55. Global Biodiversity Information Facility. http://www.gbif.org/ (2017).
strategies. Sci. Rep. 5, 12172 (2015). 56. Elith, J. et al. Novel methods improve prediction of species’ distributions from
24. Li, S.-L. et al. Essential information: uncertainty and optimal control of Ebola occurrence data. Ecography 29, 129–151 (2006).
outbreaks. Proc. Natl Acad. Sci. 114, 5659–5664 (2017). 57. Althaus, C. L. Estimating the Reproduction Number of Ebola Virus (EBOV)
25. Pourrut, X. et al. Large serological survey showing cocirculation of Ebola and During the 2014 Outbreak in West Africa. PLoS Curr. 6, https://doi.org/
Marburg viruses in Gabonese bat populations, and a high seroprevalence of 10.1371/currents.outbreaks.1391afb1375e1370f1279e1377f1329e70560952
both viruses in Rousettus aegyptiacus. BMC Infect. Dis. 9, 159 (2009). 55b7056095288. (2014).
26. Messina, J. P. et al. Mapping global environmental suitability for Zika virus. 58. Gridded Population of the World Version 3 (GPWv3) Population Grids
eLife 5, e15272 (2016). CIESIN. (SEDAC, Columbia University, New York, 2005).
27. Pigott, D. M. et al. Local, national, and regional viral haemorrhagic fever 59. Ghosh, T. et al. Shedding light on the global distribution of economic activity.
pandemic potential in Africa: a multistage analysis. The Lancet https://doi.org/ Open Geogr. J. 3, 147–160 (2010).
10.1016/S0140-6736(17)32092-5 (2017). 60. Global Roads Open Access Data Set Version 1 (gROADSv1) CIESIN. (NASA
28. Bogoch, I. I. et al. Potential for Zika virus introduction and transmission in Socioeconomic Data and Applications Center (SEDAC), Palisades, NY, 2013).
resource limited countries in Africa and Asia-Pacific: A modeling study. 61. Merler, S. & Ajelli, M. The role of population heterogeneity and human
Lancet Infect. Dis. 16, 1237–1245 (2016). mobility in the spread of pandemic influenza. Proc. R. Soc. B: Biol. Sci. 277,
29. Redding, D. W., Moses, L. M., Cunningham, A. A., Wood, J. & Jones, K. E. 557–565, https://doi.org/10.1098/rspb.2009.1605 (2010).
Environmental-mechanistic modelling of the impact of global change on 62. OpenFlights. https://openflights.org/data.html (2012).
human zoonotic disease emergence: a case study of Lassa fever. Methods Ecol. 63. Van Vuuren, D. P. et al. The representative concentration pathways: an
Evolution 7, 646–655 (2016). overview. Clim. Change 109, 5–31 (2011).
30. Dudas, G. et al. Virus genomes reveal factors that spread and sustained the 64. Kriegler, E. et al. A new scenario framework for climate change research: the
Ebola epidemic. Nature 544, 309–315 (2017). concept of shared climate policy assumptions. Clim. Change 122, 401–414
31. Mayhew, M. World Bank Group Ebola Response Fact Sheet. (2017). (2014).
32. Kramer, A. M. et al. Spatial spread of the West Africa Ebola epidemic. Roy. 65. Zubler, E. M., Fischer, A. M., Fröb, F. & Liniger, M. A. Climate change signals
Soc. Open Sci. 3, 160294 (2016). of CMIP5 general circulation models over the Alps – impact of model
33. Garchitorena, A. et al. Disease ecology, health and the environment: a selection. Int. J. Climatol. 36, 3088–3104 (2016).
framework to account for ecological and socio-economic drivers in the control 66. Hurtt, G. C. et al. Harmonization of land-use scenarios for the period
of neglected tropical diseases. Philos. Trans. R. Soc. B: Biol. Sci. 372, 20160128 1500–2100: 600 years of global gridded annual land-use transitions, wood
(2017). harvest, and resulting secondary lands. Climatic Change 109, 117 (2011).
34. Henao-Restrepo, A. M. et al. Efficacy and effectiveness of an rVSV-vectored 67. Riahi, K. et al. The shared socioeconomic pathways and their energy, land use,
vaccine expressing Ebola surface glycoprotein: interim results from the Guinea and greenhouse gas emissions implications: an overview. Glob. Environ.
ring vaccination cluster-randomised trial. Lancet 386, 857–866 (2015). Change 42, 153–168 (2017).
35. Friedl, M. A. et al. MODIS Collection 5 global land cover: algorithm
refinements and characterisation of new datasets. Remote Sens. Environ. 114,
168–182 (2010). Acknowledgements
36. World Bank. World Development Indicators (World Bank, Washington DC, We thank P. Sivasubramaniam for technical assistance, and A.P. Jones, M. Wilson,
USA, 2014). G. Mace, M. Leach, B. Sheldon, and C. Watts for comments on previous versions of the
37. Bellouin, N. et al. The HadGEM2 family of Met Office Unified Model climate manuscript.
configurations. Geosci. Model Dev. 4, 723–757 (2011).
38. O’Neill, B. C. et al. A new scenario framework for climate change research: the
concept of shared socioeconomic pathways. Climatic Change 122, 387–400 Author contributions
(2014). D.W.R. and K.E.J. designed the study, D.W.R. devised and ran analyses, D.W.R., P.M.A.,
39. Van Vuuren, D. P. et al. A new scenario framework for climate change A.A.C., G.L.I., L.M.M., J.L.N.W. and K.E.J. all contributed to the writing and editing
research: scenario matrix architecture. Clim. Change 122, 373–386 (2014). the paper.
40. Olivero, J. et al. Mammalian biogeography and the Ebola virus in Africa.
Mammal. Rev. 47, 24–37 (2017).
41. Murray, K. A. et al. Global biogeography of human infectious diseases. Proc. Competing interests
Natl Acad. Sci. USA 112, 12746–12751 (2015). The authors declare no competing interests.

10 NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-12499-6 ARTICLE

Additional information Open Access This article is licensed under a Creative Commons
Supplementary information is available for this paper at https://doi.org/10.1038/s41467- Attribution 4.0 International License, which permits use, sharing,
019-12499-6. adaptation, distribution and reproduction in any medium or format, as long as you give
appropriate credit to the original author(s) and the source, provide a link to the Creative
Correspondence and requests for materials should be addressed to D.W.R. or K.E.J. Commons license, and indicate if changes were made. The images or other third party
material in this article are included in the article’s Creative Commons license, unless
Peer review information Nature Communications thanks Cyril Caminade, Kris Murray indicated otherwise in a credit line to the material. If material is not included in the
and the other, anonymous, reviewer for their contribution to the peer review of this article’s Creative Commons license and your intended use is not permitted by statutory
work. Peer reviewer reports are available. regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this license, visit http://creativecommons.org/
Reprints and permission information is available at http://www.nature.com/reprints licenses/by/4.0/.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
published maps and institutional affiliations. © The Author(s) 2019

NATURE COMMUNICATIONS | (2019)10:4531 | https://doi.org/10.1038/s41467-019-12499-6 | www.nature.com/naturecommunications 11

 © 2019. This work is published under
http://creativecommons.org/licenses/by/4.0/(the “License”). Notwithstanding
the ProQuest Terms and Conditions, you may use this content in accordance
with the terms of the License.