Mapping the potential distribution of Phlebotomus martini and P. orientalis (Diptera: Psychodidae), vectors of kala-azar in East Africa by use of geographic information systems

Mapping the potential distribution of Phlebotomus martini and P. orientalis (Diptera: Psychodidae), vectors of kala-azar in East Africa by use of geographic information systems
of 14
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Similar Documents
  ELSEv2ER Available on]ine at Acta Tropica 90 (2004) 73-86 ACTA TROPICA, Mapping the potential distribution of Pklebotomus martini and P. orientalis (Diptera: Psychodidae), vectors of kala-azar in East Africa by use of geographic information systems T. Gebre-Michael a b J.B. Malone b M. Balkew a, A. All a, N. Berhe a, A. Hailu a, A.A. Herzi c Institute qflPathobiology, Addia Ababa Universil3; P.O. Box 1176, Addis Ababa, Ethiopia Pathobiologicai Sciences, Schoo of Veterinary Medicine Louisiana State UniversiO,, Baton Rouge, LA 70803, USA Department q'Morphology and Patholog3; Facu O, of Medicine and Surger?; Somali Nationai Universi.,, Mogadishu, SomaliaReceived 17 September 2002 received in revised brm 5 September 2003 accepted 29 September 2003 Abstract The distribution of two principal vectors of kala-azar in East Africa, Phlebotomus martini andPk ebotomus orientalis were analysed ttsing geographic information system (GIS) based on (1) earth observing satellite sensor data: Normalized Differ- once Vegetation Index (NDVI) and midday Land Surface Temperature (LST) derived fromadvanced very high resolution ra- diometer (AVHRR) of the global land km project of United States GeologicalSurvey (USGS), (2) agroclimatic data from the FAO Crop Production System Zone (CPSZ) of the Intergovernmettal Authority on Development(1GAD) sub-region, and (3) the FAO  998 soils digital map for the IGAD sub-region. Theaim was to produce a predictive risk model forthe two vectors. Data used for the analysis were based on presence and absence of the two species from previous surveycol- lections in the region (mainiy Ethiopia, Kenya and Somalia). Anmmt, wet season and dry season models were constructed. Aithough ail models resulted in more than 85% positive predictive values for both species, the best fit for the distribu- tion of P. martini was the dry season composite (NDVI 0.07-0.38 and LST 22-33 C) with a predictive value of 93.8%, and thebest fit for P. orientalis was the wet season composite (NDVI -0.01 to 0.34 and LST 23 34C) with a predictive value of 96.3%. The two seasonal composites models derived from satellite data were largely similar withbest fit models developed based on the CPSZ climate data: average altitnde (l 2-1900 m), average annual mean temperature (15-30 C), an- nual rainfall (274-1212 turn), average annual potential evapotranspiration (1264-1938 ram) and readily available soil mois- ture (62-1 i3 ram) for P. mart#'& and average altitude (200-2200 m), annual rainfall (I80 1050 ram), annual mean temper- amre (16 36C) and readily available soil moisture (67-108mm) for P. orieta/is. Logistic regression analysis indicated LST dry season composite of the satellite data, average altitude, mean annual temperature and readily avaiiable soil mois- ture of the CPSZ data as thebest ecological determinants for P. martini while LST annual composite was the only important ecological determinant for P. orientalis. Spearman's rank correlation revealed several factors to be important determinants Corresponding athor. Tel.: +25i-i-763091; fax: ?251-1-755296. E-mail addreas: (T. Gebre-Michaei). 0001-706X/$ see front matter (0 2003 Elsevier B.V. All rights reserved. doi: 10. 016/j.actatropica.2003.09.021  74 Z Gebre-Michae eta . /Acta Tropica 90 (2004) 73-86 for thedistribution of the two vectors. None of the soil types analysed appeared to be important determinant for the two species in East Africa, anlike in Sudan where P. orientalis is mainly associated with eutricverfisol (black cotton clay soil). (c) 2003 Elsevier B.V. AI] rights reserved. Keywords: Kala-azm-; Visceral ]eishmaniasis; Geographic information systems; Ph ebotomus martini; Phlebotomus orienta is; AVHRR; Remote sensing; East i. ]Introduction Kala-azar (visceral leishmaniasis) caused by Leish- mania donovani s.1. is a-major endemic health problem in countries of East Africa.Several epidemics have occarred in the past and continue to occur presently in some countries, such as in Sudan. The main endemic areas in Ethiopia occur primar- ily in the south, southwest and northwest lowland regions with isolated foci in other low-lying areas of the country. Some of these areas have experi- enced repeated epidemics of thedisease associatedwith military activities (Cole et al.,  942; Anderson, 1943) or large-scaleagricultural development projects (Mengesha and Abahoy, t978; Mare, 1979). The most recent epidemic in Humera (northwest Ethiopia) started in 1995 with 56 patients, and is currently pro- ducing 600-800 new cases annually following recent extensive agricultural and settlement projects (IPB, unpublished data). In Kenya, kala-azar occurs in semi-arid lowlands with thorn-bush vegetation. Themain endemic areas lie in Kitui and Machakos Districts east of the Rift altey, where epidemics have also occurred; and in Baringo and ',Vest Pokot Districtsin the Rift valley re- gion (Ashford and Bettini, 1987; Lawyer etal., 1989). A small outbreak has recently been reported in refugee camps in northeastern Kenya (Boussei 3' etal., 200t). One of the worst affectedcountries by kala-azar in the region is Sudan, where the disease consti- tutes one of the most important vector-borne diseases (Elnaiem et al., 1998). Themain endemic region ro,aghly occurs between Medani-Gedm'ef in the north to Ma]akal-Benim [Western Upper Nile Province (WUP)] areas in the south extending eastwards to the Ethiopian border (Hoogstraal and Heyneman, 1969; Seaman et ai., 1996). The crrent major epidemic area in sotthern S)adan (WUN) has resulted in about 100,000 deaths (Zijistra and E1-Hassan,2002). Availableinformation also indicatesthat kala-azar is highly endec in southern Somalia along the Shebelle and Jaba river basins (Cahilt et ai., 1967; Shiddo etal., 1995a,b). Two species, Phlebotomus martini and Phleboto- mus orientalis are the principal vectors of kala-azar in East Africa (Killick-Kendrick, 1999). From a number of studiesin East Africa, kala-azar is known to occur in two markedly different ecolog- ical situations. The first is where kala-azar is associ- ated with Acacia-Balanites forest and deeply cracldng 'black cotton clay' soils. Hoogstraal and Heyneman (1969) and Ftfiler et al. (1979) have stressedthe im- portance ofsuch forest and soils on the distribution of kala-azar and its vector, P. orientalis. Itis thought that such soil is hygroscopic and swells during the rainy season, then loses water and shrinks quickly during the dry season resulting in very deep cracks, creating essential micro-habitats for the vector, q orientalis. Fociof thedisease with this type of ecological sit- uation are found in endemic areas of somhwest and northwest Ethiopia, and much of the endemic areas of Stdan except perhaps in the extreme southeast. Re- cently, a kala-azar risk map has been developed based on a geographical information system forthe Sudan delineating where the vector (P. orientcdis) might be found inthat country and the adjacent endemic area of Humera in NW Ethiopia (Thomson et al., 1999). The second type of ecological feature is whereMacrotermes termite mounds play an important role in the epidemiology of kala-azar in East Africa. The ventilation shafts of these termite mounds are be-lieved to provide ideal breeding and resting habitats for the main vector, P. martini and its close rela- tives, P. vansomerenae and P. celiae (Minter, 1964; Gebre-Michael and Lane, 1996). Endemic foci of this type of ecology exist in Kenya (Minter, 1963,1964), southern Ethiopia (Gebre-Michael and Lane, 1996),  Z Gebre-Michael et al. / Acta Tropica 90 (2004) 73-86 75 southern Somalia (Herzi, 1987) and in the extreme sotthern Stdan areas close to the Kenya-Ethiopia border area (Hoogstraai and Heyneman, 1969). Inthe'termite mound habitat', P. orienmlis is either absent or rare. Conversely, P. mar ini is usually absent or rare in the R orientaiis habitat. However, the distri- butionof thetermite hills are far more widespread in the region than that of the vectors and thedisease and it is not clearly understood what other factors govern the distribution of the vector (P. martini and its close relatives). The present knowledge of thedistribtion of the disease and the vectors in the region as a whole is mostly based on crossectionalsurveys and cot]ection of sandflies in accessible areas of the region: none, however, predicts where disease transmission may oc- cur or where it might not occur. In recent years,the availability of climatic, geologi- cal and phytographic data in digital formathas greatly increased. Image ana]ysis of such data and GIS soft- ware packages suitable for use on desktop computers have led to increasing application of GIS in studies of the population dynamics of variotts arthropod vectors inrelation to a range of ecological factors and disease prediction (Hay et al., 1996; Beck et al., 2000). The use ofenvironmental variables obtained from satellites instudies of disease vectors is particularly important in countries and remote areas where ground-based mete- orological and other environrnental data are frequently unavailable. Recently, a G[ S model has been described for Phle- botomus papatasi, a vector of Leishmania major and sundry fever in southwestAsia (Cross eta]., 1996) and for the kala-azar vector (P. orientalis) in Sudan (Thomson et al., 1999). Examples of simi]ar success- ful applications of G1S in other vector-borne disease include malaria (Roberts et al., i996; Beck et al., 1997), African trypansomosis (Rogers et al., 1996; Rogers, 2000), schistosomiasis (Malone et al., 2001; Kristensen et al., 2001), onchocerciasis (Richards, 1993) and fasciolosis (Yilrna and Malone, 1998). The objective of this study was to develop a GIS predictive risk map of P. martini and P. orienmIis dis- trib,ation and probable areas of kala-azar infectionfor Ethiopia and theEast African region by matching the known biological distribution of +/-e two vectors to digital databases on potential environmental determi- nants. 2. Materials andmethods All GIS databases were developed using ArcView 3.2 GIS (ESRI, Redlands, CA) and ERDAS imagine 8.3.1 (ERDAS, Atlanta, GA) sofkware. Selected en- vironmental features from the databasesof the Crop Production System Zones (CPSZ) of the Intergovern- mental Autrhority on Development ( GAD) sub-region, Eastern Africa (Velthuizen and Verelst, 1995) were imported intothe ArcView 3.2 (ESRI, Redlands Ca) as shapefiles for extraction, analysis and compilation of the data. 2. I. Global  and 1 km data Globai ]and km data derived from the Advanced Very High Resolution Radiometer (AVHRR) sensor of National Oceanic and Atrnosphere Administration (NOAA) environmental satellites were downloaded from theinternet ( imported as image files and calibrated to prodce the Normalized Difference Veg- etation Index (NDVi) and midday land (earth) surface emperature (LST) va ues as described by Malone et al. (2001). o re&ice the effect of variations of climate years, anntial composite maps for 72 dekadal maps of NDVI and LST between January 1992 to December t992and January 1995 to December 1995 were combined to create average annal composite NDVI and LST maps (1992 + 1995)/2 using ERDAS Imagine software. NDVI and LST composite maps for the dry season (November-March, i992 and 1995) and wet season (April-October, 1992 and 1995) fortheEast African region were similarly prepared. The global landkrn NDVI product is scaled to positive values (0-2i0). After mask removal, the data were converted to real NDVI vaiues which range from -1 to +1 [(NDVI i00)/100]. The LST data calibration gave midday landsurface temperature vataes. 2.2. Sand[Iy data Data on the distribttion of the two sundry vec- tors, P. martini and P. orienta is were compiled from punished and unp-ablished records by past and present investigators between t936 and 2002 in East Africa (mainly Ethiopia, Kenya and Somalia) to as- sess the ecological determinantsof the distribution  76 Z Gebre-Michae eta . /Acta Tropica 90 (2004) 73-86 ofboth vectors. Chief among these were Minter (1963,1964) and Kfiiick-Kendrick etal. (i994) for Kenya; Herzi (1987) for Somalia; Ashford (1974), Gernetchu etal. (1975,i976), Gemetcha and Fuller (1976), Gebre-Michael and Lane (1996), Balkew et al. (1999, 2002) and other unpablished recordsof the institute of Pathobiology for Ethiopia. The data were based on presence and absence, and not on density of the two species, becausesuch infor- mation was lacking in most publications. Data from Sudan by several investigators were not used for the constrtctionof the model except forvalidation. The latitude-longitude coordinatesof ail survey sites were determined and coaverted to decima degree vales by reference to the website (iqttp://www.gnpswww. and the Microsoft Encarta Reference Library 2002 programme. 2.3. Environmental extractions and model development The NDVI and LST values for the 1992 and 1995 annuai composite, dry and wet season composites cre- ated for the IGAD sub-region were extracted from a circular 5 km bv, ffer area around a survey site (64 pos- itive and 47 negative sites f\'r P. martini and 54 pos- itive and 57 negative sites for P. (rientalis). Analysis of a scatter diagram made by plotting the extracted mean vaiues against thestrvey sites ailowed definition of the range of mean NDVt and LST valtes. These valne ranges were then used in combination to querydatabases and display map overlays of areas suggested to be most sfitfblefor P. mart i and R orientalis. The query procedure was done repeatedly by G[S analy- sis, tsing different combinationsof NDVI and LST va|ues to derive a 'best fit mode1 for thedistribution of the two vectors. 2.4. FAO Crop Production System Zone CPSZ, database The FAO CPSZ agroclimatic database for 1GAD stb-region which is based on 10years observation (i981-1991) was srcina y developed for predicting theretative ecological suitability of an area for variouscrops. Selected weather and environmental features (average aititade, mean annual temperature, annual rainfaii, average aititude, raln-potentialevapot;anspi- ration, average NDVI, readily available soil moisture, soil water logging potential, terrain slope) were ex- tracted from CPSZ database and imported into Ar- cView as shapefites br comparison to survey sites.2.5. FAO soils database As outlined above P. oriemalis has been associated with a particular soft type ('black cotton clay soil ) and P. martini with termite hills. The FAO 1998 soils digital map for the 1GAD sub-region were incorpo- ratedinto ArcView as a shapefile. Dominant and theassociated soil features (e.g., slope class) were 1inked to the srvey points within the map unit to define anyassociation of certain soil types as important determi- nants for thedistribution of the vectors. 2.6. Statistical analysis Logistic regression and Spearman's rank correlation analysis were done on the association of the presence or absence of each vector species (P. martini and P. oriemaiis) in the survey areas as the dependent vari- able, and satellite derived variables (annual and sea- sonal NDVI and LST composites) and the CPSZ agro- ciimatic variables.Results were considered to be sig- nificant if P < 0.05. 3. Resnlts 3.1. Evaluation qf the predictive value of the global land  kmAVHRR annual attd seasonal composite models Analysis of q,aeries resuiting from 1992 and 1995 annual and seasonal composites maps of NDV1 and LST for area overlays that best fit the distribution of the -vectors, P martini and P. orientalis are summarised in Table 1. For P. mariini, qery results that best fit its distribution in EastAfrica for theannual com- posite were NDVI (0.t2-0.31) and LST (21-30':'C), for the wet season composite were NDVI (0.08-0.35) and LST (18-28C) and  'or the dry season com posite were NDVI (0.07-0.38) and LST (22-33 These are mean valueranges of the  992 and 1995 NDV and LST composites which revealed area over-  ays that best fit he disvbution of P. martini for the respective seasons. The annm composite model had  T. Gebre-Michael eta . / Acre Tropica 90 (2004) 73-86 77 TablePredictive valteof the 1992 and 1995 NDVI and LST (C) anniai and seasonal composke models of Phlebotomus martini and P. orientalis in East Africa Composite NDVI and LST (C) No. of positive Positive No. of negativeNegative sites within predictive sites oatsidethe predictive the model va,aes (%) model values () P. martini Annual 0.12-0.31 and 2/-30 53/64 82.8 40/47 85.1 Wet season 0.08-0.35 and 18-28 56/64 87.539/47 83.0 Dry season 0.07-0.38 and 22-33 60/64 93.8 29/47 61.7 P. orientalis Annuai 0.05 0.28 and23 36 49/5490.7 28/57 49.1 Wet -0.0i 0.34 and 23 34 52/54 96.3 32/57 56.1 Dry Season 0.06-0.28 and 2640 49/54 90.7 26/57 45.6 a positive predictive vake of 82.8% while the wet (Fig. a) and dry season composites had positive pre- dictive vakes of 87.5 and 93.8%, respectively. Sim- i]arly, for P. orientalis, queries that best fit its dstri- bttion in East Africa were for the anmal composite NDVI (0.05-0.28) and LST (23-36 C) (Fig. lb), for the wet season composite were NDVI (-0.01 to 0.34) and LST (23-34C), and for the dry season com- posites were NDVI (0.06-0.28) and LST (26/10 C). The annLm] composite model had a positive predictive value of 90.7% while the wet and dry season com- posites had 96.3 and 90.7% predictive vaktes, respec- tively. As a dry season species mainly, the apparent high positivepredictive vaiueof the wet season com posite was also unexpected. 3.2. Evaluc'tion of tle CPSZ models Severa] queries based on scatter diagrams of the CPSZ variab es revealed that a model based on aver- age altitude (12-I900m), annual mean temperature (15-30C), annual rainfall fall (274-1212mm), an- nual potential evapotranspiration (1264-1938mm) and readi y available soit moisture (62-t 13 ram) pro- duced thebest fit to thedistribution of P. martini with the exception of a few sites missed in north- ern Somalia (Fig. 2a). Some areas sch as northern Ethiopia and Eritrea, parts of the Ogaden in eastern Ethiopian lowlandshave been predicted as endemic in the model. Similarly, average a]fitude (200-2200 m), annaal rainfall (180-1050mm), annuai mean tem- perature ( 6-36C), readily availabie soil moistnre (67-108 ram) produced thebest fitting model tkrthedistribution of P. oriented;is i East Africa (Fig. 2b). The CPSZ models were largely in agreement wih the resorts of G obai land 1kin models describedabove. 3.3. Soils Analysis of dominant soils from the FAO 1998 dig- ital map for theEast African region showed various soil types in association with the various sites of col- lection, bm provided no evidenceof a sing e soil type being associated with presence or absence of a species(Figs. 3 and 4). Twenty-five and 18 soil types were as- sociated with areas of P. martini and P. orienmis, re- spectively. Of these, euic vertisols, eurtic cambisols and humic nitoso]s were most frequently associated with P. martini (Table 2), while eutric vertisols, vertic cambisols, lithic Ieptosois and hapiic solonchaks were associated with areas of ,e orientalis (Table 3). The most common of these, edtric vertisols, occun'ed in noAhwest Ethiopia and Eritrea adjacent to the Sndan where this soil type is dominant and widespread. On theother hand, areas of P. orienm is in the Awash and the iower Omo valleys, vertic cambisols and eric visols were the dominant soii types, respectively. Else- where in Ethiopia, Kenya and Somalia other soil types were associated with thedistribttion ofboth species. Some ofthese soils were also common to both species in East Africa, the most important of these being lithic leptosols and eutric vertiois.3.4. Statistical analysis The logisfic regressionanaiysis on the global km variables, CPSZ agroclimatic variabies anddominant
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!