Ecological Niche Modeling and Spatial Risk Analysis

for potential spread of Rift Valley Fever Vectors in Kenya

Presented at 3rd MVVR Conference at Hilton Hotel Nairobi, 16th October 2014

Nanyingi M,Bayoh N, Ogola E, Thumbi M,Mosmtai G,Gachie T, Muchemi G,

Kiama G , Munyua P, Sang R, Njenga K and Bett B

History, Etiology and Epidemiology

Montgomery , 1912, Daubney 1931, Davies 1975, Jost et al., 2010

RVF viral zoonosis of cyclic

occurrence(5-10yrs), described In

Kenya in 1912 isolated in 1931 in

sheep with hepatic necrosis and

fatal abortions.

RVFV is an OIE transboundary

high impact pathogen and CDC

category A select agent.

Etiology: Phlebovirus in

Bunyaviridae (Family).

Genome: tripartite RNA segments

designated large (L), medium (M),

and small (S) contained in a

spherical (80–120 nm in diameter)

lipid bilayer.

Risk factors:

Precipitation: > 600mm, flooding

Altitude: <1100masl

Vector +: Aedes, culicines spp?

NDVI: 0.1 units > 3 months

Soil : Solonetz, Solanchaks,

planosols

Historical Outbreaks

Epidemics in Africa and recently

Arabian Peninsula; in Egypt (1977),

Kenya (1997–1998, 2006-2007),

Saudi Arabia (2000–2001) and

Yemen (2000–2001), Sudan (2007)

and Mauritania (2010)

3

RVF Vector Emergence (Ecological and Climatic)

Precipitation: ENSO/Elnino above

average rainfall leading hydrographical

modifications/flooding

(“dambos”,dams,irrigation channels).

Vector Presence: 35/38 spp.

(interepidemic transovarial maintenance by

aedes 1º and culicine 2º (vectorial capacity/

competency)

Dense vegetation cover =Persistent

NDVI.(0.1 units > 3 months)

Soil types: Solonetz, Solanchaks,

planosols (drainage/moisture)

Elevation : altitude <1,100m asl

Linthicum et al., 1999; Anyamba et al., 2009;Sang et al ., 2010; Hightower et al., 2012

r

h

Culex

eggs

Aedes

eggs

t0Jan Dec

t20

h

Aedes

eggsr

Culex

eggs

t0

Jan Dec

Adult

Den

sity

A

dult

Den

sity

(Aedes- Culex complex) responsible for maintenance and amplification of the virus

Objectives and Approaches

To evaluate the correlation between mosquito distribution

and environmental-climatic attributes favoring emergence of

RVF.

(Statistical modeling the climatic, ecological and

environmental drivers of RVF outbreaks).

To develop a risk map for spatial prediction of RVF

outbreaks in Kenya based on potential vector distribution

(Spatial and temporal analysis and risk modelling by GIS

Analysis)

Study Design and Research Approach

Cross-sectional and purposive design

1. Randomization of 15 high and 15 low risk (Case & Control)

districts based on RVF occurrence data (2006-2007).

2. Seasonality based on precipitation : Wet and dry

3. Monthly multisite sampling: 40 points in 4 quadrants.

4. Population based: Livestock and household distribution.

5. Socioeconomic survey (SES) and health care access.

6. Spatiotemporal analysis and ENM for RVF risk prediction

(Maxent, GARP,BRT,RF) using R- Statistics,ArcGIS,QGIS

Geographical Distribution of Arthropod Vectors and Exploration of

Pathogens they Transmit in Kenya (Approved KEMRI ERC, SSC 1849)

Maximum Entropy (Maxent) Model

Culex species was highly influence by the number of dry months variable (dm),

mean annual rainfall (bio12), Aedes was influenced by rainfall derived variables

Boosted Regression Trees(BRT)

Number of dry months (dm), longest dry seasons (llds) and

rainfall of wettest month (bio 13), had the highest influence on

culex species distribution.

Comparative Random Forest(RF) Output

Aedes is highly influence by moisture index of moist quarter (mimq)

rainfall of driest quarter (bio 17), rainfall of wettest month (bio13).

What Next?? Regional Models = Model Validation

Multisite country level surveillance coupled with RVF

seroepidemiology profiles for hotspots is promising for

validation and genomic pathogen discovery.

Maxent

Geographically linked phylogenetic models?

11

Limitations of the study

Lack of data from “hotspots” may complicate conclusive

associations between the vector presence, epidemiological data and

ecological predictors.

Temporal and spatial distribution was not explicitly examined due to

insufficient vector presence data.

Lack of reliable climatic and ecological parameters from local

databases hence leading to risk generalization projected from the

regional- global databases.

Despite excellent model agreement in prediction of habitat suitability

for vectors, species taxonomic identification is underway for specific

niche modelling.

Overfitting due to clustered sampling can lead to misinterpretation of

geographical spread of vector( corrected by stratification and cross-

validation)

12

Conclusions and Recommendations

This is an empirical attempt to predict large-scale country

level spatial patterns of RVF occurrence using vector data

and ecological predictor variables.

The vector predictive risk maps will be useful to animal

and human health decision-makers for planning

surveillance and control in RVF known high-risk areas.

The forecasting and early detection of RVF outbreaks

using VSS contributes to comprehensive risk assessment

of pathogen diffusion to naive areas, hence essential in

disease control preparedness.

GIS tools and ENM can contribute to existing model

frameworks for mapping the areas at high risk of RVFV

and other vector borne diseases.

ACKNOWLEDGEMENTS

Data sources

AFRICLIM database

World Clim - Global Climate data, available at http://www.worldclim.org/

Collaborating Institutions

DVS, DDSR,DVBD,MOPH, ZDU,USAMRU

Individuals

IHAHP team, study participants, CHW, Local administrators

Contact : mnanyingi@kemricdc.org, mnanyingi@gmail.com