Mapping high probability area for the Bacillus anthracis occurrence in wildlife protected area, South Omo, Ethiopia

Abstract

Anthrax is a zoonotic disease caused by a spore-forming gram-positive bacterium, Bacillus anthracis (BA), in which soil is the primary reservoir. The geographic distribution of the disease appears to be restricted by a combination of climatic and environmental conditions. Among the top five zoonotic diseases, Anthrax is the second priority zoonosis in Ethiopia. Increased anthropogenic factors inside wildlife protected areas may worsen the spillover of the disease from domestic animals to wildlife. Consequently, the prediction of the environmental suitability of BA spores to locate a high-risk area is urgent. Here we identified a potentially suitable habitat for BA spores survival and a high-risk area for appropriate control measures. Our result revealed that a relatively largest segment of Omo National Park located on the western side and more than half of the total area of Mago National Park bordering Hamer, Bena Tsamay, and south Ari were categorized under a high-risk area for the anthrax occurrence in the current situation. Therefore, the findings of this study provide the priority area to focus and allocating resources for effective surveillance, prevention, and control of anthrax before it cause devastating effect on wildlife.

Full text

Page 1/17
Mapping high probability area for the Bacillus
anthracis occurrence in wildlife protected area,
South Omo, Ethiopia
Fekede Regassa Joka (  wjoragge@gmail.com )
Ethiopian Wildlife Conservation Authority
Article
Keywords:
Posted Date: June 9th, 2023
DOI: https://doi.org/10.21203/rs.3.rs-3009574/v1
License:   This work is licensed under a Creative Commons Attribution 4.0 International License. 
Read Full License

Page 2/17
Abstract
Anthrax is a zoonotic disease caused by a spore-forming gram-positive bacterium,
Bacillus anthracis
(BA),
in which soil is the primary reservoir. The geographic distribution of the disease appears to be
restricted by a combination of climatic and environmental conditions. Among the top ve zoonotic
diseases, Anthrax is the second priority zoonosis in Ethiopia. Increased anthropogenic factors inside
wildlife protected areas may worsen the spillover of the disease from domestic animals to wildlife.
Consequently, the prediction of the environmental suitability of
BA
spores to locate a high-risk area is
urgent. Here we identied a potentially suitable habitat for
BA
spores survival and a high-risk area for
appropriate control measures. Our result revealed that a relatively largest segment of Omo National Park
located on the western side and more than half of the total area of Mago National Park bordering Hamer,
Bena Tsamay, and south Ari were categorized under a high-risk area for the anthrax occurrence in the
current situation. Therefore, the ndings of this study provide the priority area to focus and allocating
resources for effective surveillance, prevention, and control of anthrax before it cause devastating effect
on wildlife.
Background
Anthrax, which is typically associated with bioterrorism1, is a zoonotic disease caused by a spore-
forming gram-positive bacterium,
Bacillus anthracis
(
BA
). Anthrax is acute to peracute, highly contagious
infectious disease of domestic and wild animals, including birds2–4 . Humans, suids, and carnivores are
considered incidental hosts3.
BA
is widely distributed in all continents and several islands in which soil
naturally rich in organic matter and calcium promotes the survival of resilient
BA
spores and is the
primary reservoir1,5. Based on the biblical concept that mentioned anthrax as the fth and sixth plague
of Egypt (Exodus, Chapters 7 to 9), which occurred in about 1491 BC, available literature stipulated that
the probable origin of anthrax could be the former Mesopotamia and northern Africa3. However, other
evidence pointed out that anthrax originated from sub-Saharan Africa, linked with its diverse and dense
fauna6. However, the geographic distribution of the
BA
appears to be restricted by a combination of
climatic and environmental conditions7,8.
Because of the practical diculties encountered in vaccinating free-living wild animals, anthrax retains a
place in the ecology of free-ranging wildlife in several regions of the world3. In wildlife, anthrax has been
reported in many African countries like Cameroon and Coˆtedvar9,10, Ethiopia4, Tanzania11,12,
Zambia13, South Africa14, Namibia15 and Zimbabwe16 affecting different wildlife population in and
around the wildlife protected areas17. Both wild herbivores and livestock get infected by
BA
spores while
grazing and usually contaminate the soil when they die and decompose18. These serve as the potential
source of infection and important biotic agents that create localized nutrient pulses while at the same
time aggregating spores in environments , which can persist for several years18–21. Most of the time,
domestic and wildlife often share grazing grounds in which wildlife epizootics can lead to infections in
livestock and humans and Vis versa. In some regions, anthrax is hyperendemic and occurs following

Page 3/17
regular seasonal trends, while in others, the disease re-emerges sporadically after years or decades
without a single incident3. Anthrax is the second priority zoonosis in Ethiopia, which occur seasonally in
different parts of the country25,26, often associated with the dry season with the onset of the initial
rains27. In Ethiopia from 2009 to 2013 about 26,737 cases and 8523 deaths of animal anthrax reported
all over the country27.
Ethiopia has the largest population of livestock than any African country that sustains the livelihood and
the main assets for the pastoral communities, especially in the South Omo zone22,23. Due to lack of
comprehensive land use plan and policy24 most livestock share similar habitat for grazing and watering
with wildlife in and around Omo, Mago National park and Tama wildlife reserve from here on collectively
called wildlife protected areas.
Annual Vaccination and active surveillance program to identify early outbreaks in the epidemic course is
a cost-effective control measure for anthrax in livestock, while it is costly and labor intensive in wildlife.
However, advanced epidemiological tools have been found useful in facilitating the prevention and
control of anthrax.
Ecological niche modelling or Species distribution models (SDMs) methods provide good opportunities
that predict environmental suitability for pathogen survival, which assists the control strategy
implementation before the disease-causing devastating effect28–30. Thus, assessment of ecologically
suitable areas for
BA
spores, and hence potential disease riskmapping, is critical for the surveillance and
management of the disease in wildlife, as wide-scale immunization in wildlife remains untenable14.
SDMsarecommonly used to forecast environmental suitabilityfor the species as a function of a set of
selected environmental variables31,32. This technique increasingly applied to model the geographic
distribution of different diseases30,33–37 ,animal38,39 and vectors40.SDMs provided researchers with an
innovative tool to answer different questions in biogeography, biology, ecology, evolution, and
conservation41,42. It isincreasingly used todevelop riskmaps that summarize landscape suitability for
species that have the potential to spread and affect native ecosystems43.
Generally, the predictive model involves genetic pattern matching algorithms (genetic algorithm for
ruleset prediction), statistical modelling (logistic regression, discriminant function), or probability-based
algorithms (MaxEnt)44. These models, however, are subject to substantial assumptions and
limitations.To minimize this, we employed a machine learning method known as ensemble models to
combine multiple predictive models to overcome the technical challenges of building a single model that
causes high variance, low accuracy, features noise, and bias. By integrating different independent but
complementary methods, ensemblemodelscreated robust environmental suitability for
BA spores
.
BIOMOD2 allows enormous simulations by randomly re-sampling species distribution data, tting
various models for each sample. It helps to bring together the output of different algorithms in which the
ensemble results outperforms individual models37,45. So far ensemble modelling approach have been

Page 4/17
extensively implemented to develop suability model to map
BA
spores either alone or in combination for
several landscapes 2,7,8,14,28,29,37,46–51
Though Anthrax is endemic to Ethiopia and outbreaks occur sporadically in the southern part, specically
Mogo and Omo National Park, up to now, there is no risk map developed that identies a high priority
area for intervention before an outbreak in wildlife. Thus, the main objective of this study is to identify
and map a high probabilityareas for
BA
spores occurrence in wildlife-protected areas to implement
appropriate control strategies before the disease causes devastating effects on wildlife.
Results
After the assessment of multicollinearity, nine predictor variables (four climatic and ve soil) were
retained to develop the nal model (Table1). The value of VIF analysis was below 2.5 with a high
tolerance of 0.8, indicating no severe correlation between the nal selected predictor variables. The
evaluation score of variable importance showed that precipitation of December (Prec Dec.) contributed
the most to the individual and the nal ensemble models except for MARS and ANN. Soil Ca+, Mean T
Jan, and soil PH7 contribute the least for individual models (Figs. 2 and 3).
Table 1.Predictor Variable selected to develop the nal model.
Variable code Predictor variables unit Sources
Prec Dec Precipitation of December mm www.worldclim.org
Prec W quart Precipitation of wettest quarter mm www.worldclim.org
Mean T Jan Mean Temperature of January °C www.worldclim.org
Min T July Minimum Temperature of July °C www.worldclim.org
Soil PH 7 Soil PH 7 standard depths - www.isric.org
Soil Silt 2 Soil Silt content 6 standard depths g/100g (w%) www.isric.org
Soil Ca+ Extractable Calcium 30 cm depth mg/kg (ppm) www.isric.org
Soil K+ Extractable Potassium 30 cm depth mg/kg (ppm) www.isric.org
Soil P+ Extractable Phosphorus 30 cm depth mg/kg (ppm) www.isric.org
The Ensemble models outperformed the individual model with the highest evaluation score for ROC
(0.99), KAPPA (0.98), and TSS (0.96), combined with the highest sensitivity and specicity, indicating
accurate discrimination of habitat suitability for
BA
spores survival (Fig. 4). The largest suitability areas
for the
BA
spore occurrences in ONP, about 40% located on the western side bordering Surma following
Baysh Mt. The smallest areas are located in the northern part delimited by Maji near Sayi Mt. More than
50% of MNP bordering Hamer, Bena Tsamay, and south Ari was classied under a highly suitable area. A
large proportion of the central part of TWR is categorized as a highly suitable area (Fig. 5). The southern

Page 5/17
part of the three protected areas and a few patches of the northern and eastern part of TWR was low
suitable for BA spore survival in the current situation (Fig. 5).
Discussions
Anthrax is hyper-endemic in Ethiopia that affects livestock while it occurs sporadically and causes a
catastrophic effect on different wildlife species in MNP 4,52. In MNP, anthrax continues to be a problem
for the animal, human, and wildlife populations. While ecologically suitable modelling was done in the
northern part of Ethiopia, focusing on domestic animals 37 and the global anthrax distribution modelling
that categorized Ethiopia as a less suitable area 2, our current model presents the rst ecological
suitability assessment for
BA
spore persistence, targeting wildlife protected areas. A localized modelling
approach in the endemic area with inadequate information resulted in accurate predictions 37.
Our ltering process of environmental predictor variables to minimize multicollinearity maintains nine
predictor variables. Prec Dec contributes the most to the individual and the nal ensemble model
indicating the importance of this predictor variable for
BA
spore survival (Fig.2). The highest contribution
of perspiration in this study corroborates the emergence of precipitation from a set of environmental
variables as the best predictor in a spatial model 37. In MNP outbreaks of anthrax periodically occurred
succeeding a cyclical pattern, most often associated with a dry climatic resulting in above-average
rainfall 4. Anthrax outbreak requires favorable seasonal changes such as warm weather and precipitation
that favor the possibility of spore survival and sporulation 53. These environmental factors are the main
determinants for the onset of anthrax outbreaks 5. Anthrax revealed an anity for low precipitation
values during the driest month, followed by a signicant decrease as precipitation increased 8.
While soil Ca + and soil PH7 contribute less to the nal ensemble model as compared to other predictor
variables in this study (Fig.3), several studies asserted that soil parameters such as PH, calcium, and
organic matter content were the most inuential variables for the long-term persistence of
BA
spores in
the environment 29,51,54,55.
BA
spores remain viable for years in the deeper layer of clayey soils rich in
calcium and decaying organic matter with slightly alkaline pH and humus 29,37,51,54,56. However, it
corroborates that soil specimens with anthrax positive contain lower calcium than negative samples 53.
Weakly acidic soil provide good condition for
BA
spore 53,57,58.
Compared to other soil properties, soil K + and Silt 2 contributed highly to individual and nal ensemble
models. This nding corroborates that the hot and dry climatic conditions with slightly alkaline soil rich in
potassium favor
BA
spore survival 8,49. The presence of various nutrients in the soil creates a favorable
environment for spore survival and growth 53–55, 59,60. Further, amoebas, common in moist soils and
pools of standing water, serve as ampliers of
BA
spores by enabling germination and intracellular
multiplication 1.

Page 6/17
In this study, more than half of MNP bordered by Hamer, Bena Tsamay, and south Ari ranked under a
highly suitable area for
BA
(Fig.5). This coincides and overlaps with the reported anthrax case in wildlife
by 4 indicating the accuracy and transferability of the model. In South Omo, pastoral and agro-pastoral
cattle production system, anthrax was ranked as the second priority disease, which affected livelihood by
hindering cattle productivity 22.
During model building, other environmental predictor variables revealed a high correlation with the most
signicant predictor variables maintained to generate the nal model. Accordingly, we are convinced that
apart from the signicantly contributing predictor variable, which favors the persistence of
BA
spores in
our high and very high-risk areas, other environmental factors may contribute to the occurrence of
anthrax in the study area. However, we still trust that the identied risk area should be given more
attention for prevention and control of the impact of anthrax on wildlife conservation.
In this study, an ensemble modelling approach, which pulls the result of different ecological niche
modelling, was used to predict the potentially suitable habitat of
BA
spores in wildlife protected areas. We
mapped for the rst time the potential risk area of
BA
spores using the environmental predictor variable.
Although anthrax is a zoonotic disease that affects the public, livestock, and wildlife in the southern Omo
zone, risk mapping has not been done. Consequently, there is no evidence-based allocation of scarce
resources to prevent and control the disease. Therefore, the ndings of this study provide important
insights for spatially allocating and prioritizing resources for anthrax surveillance, prevention, and control
based on the predicted risk level within the three protected areas.
Materials and Methods
Research area
Our research area focused on wildlife-protected areas situated in the southern part of Ethiopia, inspired
and established by the story of uninhabited wilderness and unspoiled natural beauty 61. It is bordered
from west and southwest by south Sudan, in which the Tama wildlife reserve is found between Omo and
Mago national parks and serves as the connecting corridor (Fig 1). Omo National Park (ONP) is located
within the upper Lake Turkana rift, is characterized by a complex topography of hills, Rivers, and plains
62,63 with climate comparatively dry, having mean annual precipitation of 784 mm  64 and 810 mm, with
an average annual temperature range from 20°C to 40°C 63. While sporadic rainfall may occur every
month of the year 64, March and April are characterized by a long rainy season, whereas October and
November short rains. The vegetation of ONP comprises extensive grassland, dense bushland, and
Riverine forest 62,64. Currently, some portion of the grasslands converted into built-up and industrial
farmland 63.
Mago National Park (MNP) is located west of the Rift Valley in the lower Omo River valley in the transition
zone between the highlands and the arid lowlands, which continue south into Kenya 65,66. The central

Page 7/17
part of the park is a relatively at plain, while the periphery is formed by mountains and chains of hills 66.
The park area is crossed by the Mago River and two of its tributaries on the southwest side, bordered by
the Omo River 67.
The lower elevated area climate of MNP is distinguished by hot, dry, and semi-arid, with average annual
temperatures ranging from 24oC to 38oC. Though there is variation in the rainfall distribution between low
and highlands, the long rainy season is the same as ONP. About 50% of vegetation is mainly bush, while
the remaining comprises forest, savanna bushland, savanna grassland, and open grassland 67. The
fauna of the three wildlife-protected areas includes diverse species of mammals, birds, sh, and an
unknown number of reptiles, amphibians, and invertebrates 63,67–69.
The economy of communities around these wildlife protected areas largely depends on agriculture and
natural resources utilization, which trigger intense disputes that lead to a dramatic decline in wildlife,
especially the large savanna game animals 69. Both the three protected areas are surrounded by settled
and agropastoralist nomadic communities 67 that frequently enter with their livestock 4. The ecosystem
of those wildlife-protected areas is tremendous and dynamic, with the communities' social and economic
structure linked closely to natural resource utilization. Thus, it is under the signicant inuence of
anthropogenic factors like settlement, subsistence, and commercial agricultural practices such as sugar
cane farms/plantations 63. This increased encroachment into wildlife-protected areas and increased
utilization of wildlife resources may aggravate the occurrence of diseases like anthrax. 
Data collection and processing
Conrmed cases of
BA
occurrence point (n=30) in wildlife were extracted from FAO (https://empres-
i.apps.fao.org), incorporating eight African countries (Botswana, Kenya, Malawi, Mozambique, Namibia,
Uganda, Tanzania, and Zambia) with recorded report of anthrax in wildlife. The collected data contains
geographic coordinates, observation date, wildlife species infected, died, and susceptible population for
each outbreak as available from 8 June 2010 to 23 July 2015. Predictor variable comprising monthly
precipitation (n = 12), monthly mean, minimum and maximum temperature (n = 36), derived bioclimatic
variables (n =19) and elevation were extracted from worldclim, with data from 1970 to 2000 at 30 arc‐
second resolutions (www.worldclim.org); combination of physical (sand , clay, slit), chemical (soil pH, Ca,
K, and Na) content of soil 70 at 250 m resolution extracted from ISRIC-World Soil Information
(https://www.isric.org). To adjacent the cell size, extant and projection system  of predictor variable
collected from different source all spatial data were resampled at the spatial resolution (30 arc-seconds)
and the cell value range of the elevation layer by the Extract/ Mask tool in the ArcGIS Spatial Analyst
(ESRI, Redlands, CA, USA) following 36. A variance ination factor (VIF) analysis was done in IBM®
SPSS® statistics version 25 to avoid multicollinearity of predictors variables 71. To minimize spatial
autocorrelation of the occurrence point we raried the points following 30.

Page 8/17
Spatial modelling and evaluation process
An ensemble modelling approach was employed using BIOMOD2 package (Thuiller et al., 2016) in R and
Maxent, v. 3.4.4 73,74. Ten algorithms were run: Articial Neural Networks (ANN) 75, Surface Range
Envelope (SRE), Flexible Discriminant Analysis (FDA), General Linear Models (GLM), General Boosted
Models (GBM) 76, General Additive Models (GAM) 77,78, Classication Tree Analysis (CTA) 79, Multiple
Adaptive Regression Splines (MARS) 80, Random Forests (RF), and Maximum Entropy (Maxent) 81 82. The
model accuracy and stability were assessed using the area under the curve of the receiver operating
characteristic (AUC) 32,83, KAPPA and the true skill statistic (TSS) and sensitivity with a portion of original
data set aside for testing (Thuiller et al., 2009). Model with greater accuracy on the test data were
projected on the current environmental predictor variables to develop an ensemble model. Modelling
options in BIOMOD2 set to default and the algorithm runs 3-fold with 30 outputs for the ten models with
the ‘build. clamping. Mask’ option set to ’TRUE’ to determine the uncertainty of the model. Data were split
into training and evaluation sets in which 80 % of the data were used to develop the model, while 20 %
was used to evaluate the model’s performance 37.
Declarations
Data Availability Statement
The data that support the ndings of this study are freely downloadable from the following websites:
Food and Agricultural Organization (FAO) (https://empres-i.apps.fao.org), worldclim (www.worldclim.org),
ISRIC-World Soil Information (https://www.isric.org).
Additional information
Competing Interests
The author declare no competing interests.
References
1. Dey, R., Hoffman, P. S. & Glomski, I. J. Germination and amplication of anthrax spores by soil-
dwelling amoebas.
Appl. Environ. Microbiol.
78, 8075–8081 (2012).
2. Carlson, C. J.
et al.
The global distribution of Bacillus anthracis and associated anthrax risk to
humans, livestock and wildlife.
Nat. Microbiol.
4, 1337–1343 (2019).
3. Hugh-Jones, M. E. & Vos, V. de. Anthrax and wildlife.
Rev. sci. tech. Off. int. Epiz.
21, 359–383 (2002).
4. Shiferaw, F., Abditcho, S., Gopilo, A. & Laurenson, M. K. Anthrax outbreak in Mago National Park,
southern Ethiopia.
Vet. Rec.
150, 318–320 (2002).
5. FAO. Anthrax outbreaks: a warning for improved prevention, control and heightened awareness.
Empres Watch
37, 8 (2016).

Page 9/17
. Smith, K. L.
et al.
Bacillus anthracis diversity in Kruger National Park.
J. Clin. Microbiol.
38, 3780–
3784 (2000).
7. Blackburn, J. K., McNyset, K. M., Curtis, A. & Hugh-Jones, M. E. Modeling the geographic distribution
of Bacillus anthracis, the causative agent of anthrax disease, for the contiguous United States using
predictive ecologic niche modeling.
Am. J. Trop. Med. Hyg.
77, 1103–1110 (2007).
. Kracalik, I. T.
et al.
Modeling the environmental suitability of anthrax in Ghana and estimating
populations at risk: Implications for vaccination and control.
PLoS Negl. Trop. Dis.
11, 1–17 (2017).
9. Klee, S. R.
et al.
Characterization of Bacillus anthracis-like bacteria isolated from wild great apes
from Cote d’Ivoire and Cameroon.
J. Bacteriol.
188, 5333–5344 (2006).
10. Leendertz, F. H.
et al.
A new Bacillus anthracis found in wild chimpanzees and a gorilla from west
and central Africa.
PLoS Pathog.
2, 0001–0004 (2006).
11. Hampson, K.
et al.
Predictability of anthrax infection in the Serengeti, Tanzania.
J. Appl. Ecol.
48,
1333–1344 (2011).
12. Gainer, R. S. Epizootiology of anthrax and Nyasa wildebeest in the Selous Game Reserve, Tanzania.
J. Wildl. Dis.
23, 175–178 (1987).
13. Hang’ombe, M. B.
et al.
Human – animal anthrax outbreak in the Luangwa valley of Zambia in 2011.
Trop. Doct.
42, 136–139 (2012).
14. Steenkamp, P. J., Heerden, H. van & Schalkwyk, O. L. van. Ecological Suitability Modelling for Anthrax
in the Kruger National Park, South Africa.
PLoS One
110 (2013).
15. Turnbull, P. C. B., Doganay, M., Aygen, B., Lindeque, P. M. & Mclaughlin, J. Serology and anthrax in
humans, livestock and Etosha National Park wildlife.
Epidemiol. Infect.
108, 299–313 (1992).
1. Clegg, S. B., Turnbull, P. C. B., Foggin, C. M. & Lindeque, P. M. Massive outbreak of anthrax in wildlife
in the Malilangwe Wildlife Reserve, Zimbabwe.
Vet. Rec.
160, 113–118 (2007).
17. Nsoh, A. E.
et al.
Mapping as a tool for predicting the risk of anthrax outbreaks in northern region of
Ghana.
Pan Afr. Med. J.
25, 1–5 (2016).
1. Turner, W. C.
et al.
Fatal attraction: Vegetation responses to nutrient inputs attract herbivores to
infectious anthrax carcass sites.
Proc. R. Soc. B Biol. Sci.
281, (2014).
19. Towne, E. G. Prairie vegetation and soil nutrient responses to ungulate carcasses.
Oecologia
122,
232–239 (2000).
20. Bump, J. K., Peterson, R. O. & Vucetich, J. A. Wolves modulate soil nutrient heterogeneity and foliar
nitrogen by conguring the distribution of ungulate carcasses.
Ecology
90, 3159–3167 (2009).
21. Melis, C.
et al.
Soil and vegetation nutrient response to bison carcasses in Białowieża Primeval
Forest, Poland.
Ecol. Res.
22, 807–813 (2007).
22. Adane, Z., Yemane, N. & Hidosa, D. Cattle Production System and Breeding Practices in Bena-Tsemay
District of South Omo, South-Western Ethiopia.
Int. J. Agric. Biol. Sci.
1–19 (2021).
23. Bereda, A. Livestock and Livestock Products and By-Product Trade in Ethiopia : A Review.
Dev. Ctry.
Stud.
6, 1–9 (2016).

Page 10/17
24. Damtie, M. Legal and institutional issues for environment in Ethiopia in 2008. in
Ethiopian
Environment Review
(ed. Edwards, S.) 1–232 (2010).
25. Seid, K., Shiferaw, A. M., Yesuf, N. N., Derso, T. & Sisay, M. Livestock owners’ anthrax prevention
practices and its associated factors in Sekota Zuria district, Northeast Ethiopia.
BMC Vet. Res.
16, 1–
8 (2020).
2. Olani, A., Dawo, F. & Lakew, M. Laboratory diagnostic methods and reported outbreaks of anthrax in
Ethiopia.
Eur. J. Biol. Res.
10, 81–95 (2020).
27. Romha, G. & Girmay, W. Knowledge, attitude and practice towards anthrax in northern Ethiopia: a
mixed approach study.
BMC Infect. Dis.
20, 1–17 (2020).
2. Abdrakhmanov, S. K.
et al.
Maximum entropy modeling risk of anthrax in the Republic of
Kazakhstan.
Prev. Vet. Med.
144, 149–157 (2017).
29. Chikerema, S. M., Murwira, A., Matope, G. & Pfukenyi, D. M. Spatial modelling of Bacillus anthracis
ecological niche in Zimbabwe.
Prev. Vet. Med.
111, 25–30 (2013).
30. Fekede, R. J., van Gils, H., Huang, L. Y. & Wang, X. L. High probability areas for ASF infection in China
along the Russian and Korean borders.
Transbound. Emerg. Dis.
66, 852–864 (2019).
31. Pearson, R. G., Raxworthy, C. J., Nakamura, M. & Peterson, A. T. Predicting species distributions from
small numbers of occurrence records: A test case using cryptic geckos in Madagascar.
Biogeography
34, 102–117 (2007).
32. Forester, B. R., Dechaine, E. G. & Bunn, A. G. Integrating ensemble species distribution modelling and
statistical phylogeography to inform projections of climate change impacts on species distributions.
Divers. Distrib.
1480–1495 (2013) doi:10.1111/ddi.12098.
33. Mischler, P.
et al.
Environmental and socio-economic risk modelling for Chagas disease in Bolivia.
Geospat. Health
6, 59–66 (2012).
34. Wang, S. Analysis of the spatial distribution risk of wild bird avian inuenza on the migration route in
eastern China and construction network of nature reserve. (Northeast Forestry University, 2018).
35. Wang, H. The Spatio-temporal dynamic of canine distemper and its risk assessment in eastern part
of Heilongjiang province, China. (Northeast Forestry University, 2019).
3. Fekede, R. J., HaoNing, W., Hein, V. G. & XiaoLong, W. Could wild boar be the Trans-Siberian
transmitter of African swine fever?
Transbound. Emerg. Dis.
00, 1–11 (2020).
37. Assefa, A., Bihon, A. & Tibebu, A. Anthrax in the Amhara regional state of Ethiopia; spatiotemporal
analysis and environmental suitability modeling with an ensemble approach.
Prev. Vet. Med.
184,
105155 (2020).
3. van Gils, H., Westinga, E., Carafa, M., Antonucci, A. & Ciaschetti, G. Where the bears roam in Majella
national park, Italy.
Nat. Conserv.
22, 23–34 (2014).
39. Bosch, J., Mardones, F., Pérez, A., Torre, A. D. & Muñoz, M. . A maximum entropy model for predicting
wild boar distribution in Spain.
Spanish J. Agric. Res.
12, 984–999 (2014).

Page 11/17
40. Levine, R. S., Peterson, A. T. & Benedict, M. Q. Geographic and ecologic distributions of the Anopheles
gambiae complex predicted using a genetic algorithm.
Am. J. Trop. Med. Hyg.
70, 105–109 (2004).
41. Elith, J.
et al.
Novel methods improve prediction of species ’ distributions from occurrence data.
Echography
29, 129–151 (2006).
42. Elith, J.
et al.
A statistical explanation of MaxEnt for ecologists.
Divers. Distrib.
17, 43–57 (2011).
43. Jimenez-Valverde, A.
et al.
Use of niche models in invasive species risk assessments.
Biol Invasions
13, 2785–2797 (2011).
44. Alexander, K. A., Lewis, B. L., Marathe, M., Eubank, S. & Blackburn, J. K. Modeling of Wildlife-
Associated Zoonoses: Applications and Caveats.
Vector-borne zoonotic Dis.
12, 1005–1018 (2012).
45. Thuiller, W., Lafourcade, B., Engler, R. & Araújo, M. B. BIOMOD - A platform for ensemble forecasting
of species distributions.
Ecography (Cop.).
32, 369–373 (2009).
4. Blackburn, J. K. Integrating geographic information systems and ecological niche modeling into
disease ecology: A case study of bacillus anthracis in the United States and Mexico.
NATO Sci.
Peace Secur. Ser. A Chem. Biol.
59–88 (2010) doi:10.1007/978-90-481-9637-1_7.
47. Joyner, T. A.
et al.
Modeling the potential distribution of Bacillus anthracis under multiple climate
change scenarios for Kazakhstan.
PLoS One
5, (2010).
4. Mullins, J. C.
et al.
Ecological niche modeling of bacillus anthracis on three continents: Evidence for
genetic-ecological divergence?
PLoS One
8, 1–8 (2013).
49. Driciru, M.
et al.
Environmental determinants inuencing anthrax distribution in Queen Elizabeth
Protected Area, Western Uganda.
PLoS One
15, 1–21 (2020).
50. Barro, A. S.
et al.
Redening the Australian Anthrax Belt: Modeling the Ecological Niche and
Predicting the Geographic Distribution of Bacillus anthracis.
PLoS Negl. Trop. Dis.
1–16 (2016)
doi:10.1371/journal.pntd.0004689.
51. Mwakapeje, E. R.
et al.
Ecological niche modeling as a tool for prediction of the potential geographic
distribution of Bacillus anthracis spores in Tanzania.
Int. J. Infect. Dis.
79, 142–151 (2019).
52. Cherie, E. Overview on the status of the African elephant (Loxodonta africana) in the Omo and Mago
National Parks. in
The Wildlife Conservation Area of Ethiopia: Current Status and Future Prosprcts
(eds. Mengistou, S. & Getahun, A.) 1–122 (2002).
53. Ahsan, M. M.
et al.
Investigation into Bacillus anthracis spore in soil and analysis of environmental
parameters related to repeated anthrax outbreak in Sirajganj, Bangladesh.
Thai J. Vet. Med.
43, 449–
454 (2013).
54. Hugh-Jones, M. & Blackburn, J. The ecology of Bacillus anthracis.
Mol. Aspects Med.
30, 356–367
(2009).
55. Dragon, D. C. & Rennie, R. P. The ecology of anthrax spores: tough but not invincible.
Can. Vet. J.
36,
295–301 (1995).
5. Salgado, J. R. S.
et al.
Detection of bacillus anthracis and bacillus anthracis-like spores in soil from
state of Rio de Janeiro, Brazil.
Mem. Inst. Oswaldo Cruz
115, 1–10 (2020).

Page 12/17
57. Artenstein, A. W.
et al.
Chloroquine enhances survival in Bacillus anthracis intoxication.
J. Infect. Dis.
190, 1655–1660 (2004).
5. Jula, M., Jabbari, G. R. & Malek, B. Isolation of anthrax spores from soil in endemic regions of
Isfahan, Iran.
Arch. Razi Inst.
58, 29–38 (2004).
59. Pabian, S. E. & Brittingham, M. C. Soil Calcium and Forest Birds: Indirect Links Between Nutrient
Availability and Community Composition.
Ecosystems
15, 748–760 (2012).
0. Jula, G. M., Jabbari, A. & Darmian, F. V. Determination of anthrax foci through isolation of Bacillus
anthracis form soil samples of different regions of Iran.
Arch. Razi Inst.
62, 23–30 (2007).
1. Turton, D. Wilderness, wasteland or home? three ways of imagining the lower Omo valley.
J. East.
African Stud.
5, 158–176 (2011).
2. Ikeda, H., Ono, Y., Baba, M., Doi, T. & Iwamoto, T. Ranging and activity patterns of three nocturnal
viverrids in Orno National Park, Ethiopia.
Afr. J. Ecol.
20, 179–186 (1982).
3. Armaw, A. & Molla, M. B. Assessing the trend and magnitude of land cover dynamics and its major
driving forces in Omo National Park, Southern Ethiopia.
Cogent Soc. Sci.
8, (2022).
4. Enawgaw, C. Distribution , Abundance And Age Structure Of Elephants In Omo National Park ,
Ethiopia.
walia
17, 17–26 (1994).
5. Demeke, Y. The Birds of Mago National Park.
Walia
38–48 (1995).
. Demeke, Y. & Bekele, A. Study on the Elephants of Mago National Park, Ethiopia.
Pachyderm
32–41
(2000).
7. Demeke, Y. Elephants in the Mago National Park: An assessment of number, distribution and
movements.
Walia
15, 23–32 (1994).
. Debebe, D. F. Conservation status of bird fauna of South West of Omo National Park, Ethiopia.
Int. J.
Biodivers. Conserv.
10, 276–291 (2018).
9. Nishizaki, N. Differing Local Attitudes Toward Conservation Policy: a Case Study of Mago National
Park, Ethiopia.
Afr. Study Monogr.
29, 31–40 (2005).
70. Nath, S. & Dere, A. Soil geochemical parameters inuencing the spatial distribution of anthrax in
Northwest Minnesota, USA.
Appl. Geochemistry
74, 144–156 (2016).
71. Claudia, P., Skidmore, A. K., Mccall, M. K., Gils, H. A. M. J. Van & Mccall, M. K. Smallholder Farms as
Stepping Stone Corridors for Crop-Raiding Elephant in Northern Tanzania : Integration of Bayesian ...
Smallholder Farms as Stepping Stone Corridors for Crop-Raiding Elephant in Northern Tanzania :
Integration of Bayesian Expert System. (2015) doi:10.1007/s13280-013-0437-z.
72. Thuiller, A. W., Georges, D., Engler, R., Georges, M. D. & Thuiller, C. W. The biomod2 package: the
updated object-oriented version of BIOMOD package. 1–104 (2016) doi:Artn 20141776 Doi
10.1098/Rspb.2014.1776.
73. Phillips, S. J., Anderson, R. P., Dudík, M., Schapire, R. E. & Blair, M. E. Opening the black box: an open-
source release of Maxent.
Ecography (Cop.).
40, 887–893 (2017).

Page 13/17
74. Phillips, S. J. & Dudík, M. Modeling of species distribution with Maxent: new extensions and a
comprehensive evalutation.
Ecograpy
31, 161–175 (2008).
75. Ripley, B. D.
Pattern recognition and neural networks
. (1996).
7. Dunn, P. K. & Smyth, G. K.
Generalized Linear Models With Examples in R Springer Texts in Statistics
.
(2018).
77. Edition, S.
Generalized Additive Models An Introduction with R
. (2017).
7. Hastie, T. J. & Tibshirani, R. Generalized additive models.
Stat. Sci.
1, 297–310 (1986).
79. Breiman, L., Friedman, J. H., Olshen, R. A. & Stone, C. J.
Classication and Regression Trees
.
(Chapman & Hall, 1984).
0. Naser, A. H., Badr, A. H., Henedy, S. N., Ostrowski, K. A. & Imran, H. Application of Multivariate
Adaptive Regression Splines (MARS) approach in prediction of compressive strength of eco-friendly
concrete.
Case Stud. Constr. Mater.
17, (2022).
1. Phillips, S. J., Dudík, M. & Schapire, R. E. A maximum entropy approach to species distribution
modeling.
Ecography (Cop.).
1–8 (2004) doi:10.1145/1015330.1015412.
2. Forester, B. R. Integrating ensemble species distribution modelling and statistical phylogeography to
inform projections of climate change impacts on species distributions.
Kathleen L. Kitto, Dean of the
Graduate School
(2012). doi:10.1111/ddi.12098.
3. Fielding, A. H. & Bell, J. F. A review of methods for the assessment of prediction errors in
conservation presence / absence models.
Environ. Conserv.
24, 38–49 (1997).
Figures

Page 14/17
Figure 1
Location Map of the three Wildlife protected areas (Omo NP, Tama WR and Mago NP) in southern part of
Ethiopia.

Page 15/17
Figure 2
Predictor Variable selected to develop the nal model and their contribution to the individual model.
Figure 3
Predictor Variable selected to develop the nal ensemble model and their contribution

Page 16/17
Figure 4
Individual and ensemble model evaluation score.
Figure 5

Page 17/17
Probability of environmental suitability for occurrence of anthrax. The red color indicate a high risk area
for Anthrax occurrence in Omo NP, Tama WR and Mago NP