- Email: [email protected]
Spatial modelling and ecology of Echinococcus multilocularis transmission in China
Parasitology International 55 (2006) S227 – S231 www.elsevier.com/locate/parint
Spatial modelling and ecology of Echinococcus multilocularis transmission in China F. Mark Danson a,*, Patrick Giraudoux b, Philip S. Craig a a
Centre for Environmental Systems Research and Biosciences Research Institute, School of Environment and Life Sciences, University of Salford, Manchester M5 4WT, UK b Department of Environmental Biology EA3184 aff. INRA, Universite´ de Franche-Comte´, 25030, Besanc¸on Cedex, France Available online 20 December 2005
Abstract Recent research in central China has suggested that the most likely transmission mechanism for Echinococcus multilocularis to humans is via domestic dogs which are allowed to roam freely and hunt (infected) small mammals within areas close to villages or in areas of tented pasture. This assertion has led to the hypothesis that there is a landscape control on transmission risk since the proximity of suitable habitat for susceptible small mammals appears to be the key. We have tested this hypothesis in a number of endemic areas in China, notably south Gansu Province and the Tibetan region of western Sichuan Province. The fundamental landscape control is its effect at a regional scale on small mammal species assemblages (susceptible species are not ubiquitous) and, at a local scale, the spatial distributions of small mammal populations. To date the research has examined relationships between landscape composition and patterns of human infection, landscape and small mammal distributions and recently the relationships between landscape and dog infection rates. The key tool to characterize landscape is satellite remote sensing and these data are used as inputs to drive spatial models of transmission risk. This paper reviews the progress that has been made so far in spatial modeling of the ecology of E. multilocularis with particular reference to China, outlines current research issues, and describes a framework for building a spatial – temporal model of transmission ecology. D 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Satellite remote sensing; Landscape ecology; Transmission
1. Introduction The fox tapeworm Echinococcus multilocularis is transmitted in a wildlife cycle that may involve multiple definitive and intermediate hosts. In the Chinese/central Asian endemic areas, definitive hosts include the red fox (Vulpes vulpes), Tibetan fox (Vulpes ferrilata) and corsac fox (Vulpes corsac) and intermediate hosts a wide range of small mammal species including for example Pika (Ochotona sp.) and vole species (Arvicolid sp.) [1]. Although intermediate host infection, from E. multilocularis eggs does not require direct contact with definitive hosts, it does require spatial (and temporal) interaction between fox home range and a landscape patch with susceptible small mammals. Foxes may visit many different patches however, and, given that one fox may harbour up to 10,000 worms [2] producing eggs for up to 40 days, and that eggs may remain
* Corresponding author. Tel.: +44 161 295 4038; fax: +44 161 295 5015. E-mail address: [email protected] (F..M. Danson). 1383-5769/$ - see front matter D 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.parint.2005.11.047
viable in favourable microclimatic conditions for several months [3], environmental contamination is likely to be widespread but spatially aggregated and thus patchy at various scales [4,5]. Human infection with E. multilocularis causes the rare but highly pathogenic disease human alveolar echinococcosis (HAE) and results from the ingestion of eggs either by direct contact with infected foxes or other canid, or through contact with an egg-infected environment (Fig. 1). Viable eggs must exist within the small mammal intermediate host habitat for sustainable transmission of the parasite, but may also be distributed within the fox home range and in the human environment where transmission to domestic dogs is active [6]. Investigations of transmission to humans have used epidemiological surveys to determine risk factors for HAE [7 –9]. Although the results are not uniform, with risk factor varying from one study area to another, a common finding in China and central Asia is that dog ownership is frequently identified as a key risk factor. In areas with high rates of HAE, dog populations are large; dogs are allowed to roam freely and
S228
F.M. Danson et al. / Parasitology International 55 (2006) S227 – S231
Fig. 1. Conceptual model of transmission cycle of Echinococcus multilocularis (modified from Giraudoux et al. [4]).
hunt small mammals; and there is often close contact between dogs and humans [10]. Following the conceptual model (Fig. 1), this implies that dogs must hunt in landscape patches with susceptible small mammals, and that these patches are close to the human environment. These observations have driven the hypothesis that there is an important spatial dimension to the ecology of E. multilocularis transmission related to landscape composition, and that the risk of HAE is also affected by spatial relationships with landscape [11]. Research over the last 10 years has tested this hypothesis in a range of environments and has introduced the new technologies of satellite remote sensing and geographical information systems. It has also applied geostatistical modeling approaches to account for spatial correlation in HAE and landscape data. This paper first outlines research undertaken by the authors in an endemic area in central China. Second, it reviews the related research that is contributing to increased understanding of the spatial characteristics of E. multilocularis transmission, and third, it highlights a range of issues that is shaping the development of spatial transmission models for the parasite. 2. Spatial modeling of E. multilocularis transmission to humans in China The fundamental landscape control on E. multilocularis transmission is its effect at regional scale on the species composition of small mammal assemblages and the population dynamics of species and, at a local scale, on the spatial distributions of small mammal populations. At a regional scale, the distribution of counties with HAE across China appears to be related to the distribution of grasslands which occur in
central and western China [12]. Land cover data were obtained from the International Geosphere Biosphere Programme land cover map and HAE data from Zhou and others [13]. The qualitative correlation confirmed that the transmission of E. multilocularis to humans requires grassland landscapes supporting suitable intermediate hosts and work in Kazakhstan provided further support [14]. This scale of investigation is problematic, however, because of the incomplete records of HAE and the generalization of land cover characteristics which cannot explain the particular small mammal species assemblage found in a given grassland area. Studies at local scale have been more productive since it is possible to produce detailed landscape maps with satellite remote sensing, and data on parasite host distributions and human cases is more readily collected. In China, Craig and others [7] first suggested a link between HAE prevalence and landscape in an upland agricultural area in southern Gansu Province. Prevalence rates in 33 villages (determined by mass screening) varied from 0% to 16% across an area of approximately 40 40 km, and high rates were associated with shrubland and grassland habitat close to villages. In follow-up studies [15,16], Landsat satellite images from the 1970s, 1980s and 1990s were used to produce landscape maps of the study area and the landscape composition around the target villages was determined from classified images using spatial analysis tools. Stepwise regression analysis showed a strong correlation between the landscape composition in a buffer zone of 2000 m radius around the villages and the prevalence of HAE, and a four-class landscape model explained approximately 70% of the spatial variation in HAE [15]. However, analysis of the same data showed that there was spatial correlation in the village prevalence rates, which caused
F.M. Danson et al. / Parasitology International 55 (2006) S227 – S231 Table 1 Logistic regression of prevalence and HAE in south Gansu (modified from Danson et al. [12]) Variable
Model 1
Model 2
Parameter Significance Parameter Significance estimate ( P) estimate ( P) Intercept Age Gender Occupation Scavenging dog Tree/shrub, 750 m All rodent habitat, 1750 m
5.1 0.019 0.44 1.3 1.13
<0.0001 0.027 0.074 0.031 0.001
9.0 0.26
<0.0001 0.0008
0.0095 <0.0001 0.00031 <0.0001
an inflation of the estimate of regression model fit [17]. A Getis spatial filter was applied to account for the spatial dependence in both HAE prevalence and the landscape data, and the new model explained 59% of the spatial variation in HAE prevalence confirming that there was spatial correlation in the data, but showing that landscape was still highly significant in explaining prevalence patterns. A further development in the Gansu study area was to examine the aspatial risk factors, such as age and occupation, which had been obtained in the earlier work, and to combine these in a logistic regression model with the landscape data discussed above. Logistic regression on the aspatial data showed that age, gender, occupation and ownership of a scavenging dog were significantly related to HAE infection (Table 1, Model 1), but when landscape variables were introduced, of the aspatial variables, only age remained statistically significant (Table 1, Model 2) [12]. This result enabled the first age-specific risk map of HAE infection to be produced for the study area. Analysis of the Gansu landscape was also undertaken using a landscape metrics approach where the size, shape and spatial arrangement of land cover patches was determined from classified satellite imagery [16]. A large number of metrics was calculated and the Fmean shape index_, a metric of shape complexity, of all landscape patches in a buffer zone around villages, were found to be most strongly correlated with village HAE prevalence [16]. Landscape metrics are notoriously difficult to interpret however, and the ecological interpretation of such indices is often speculative. 3. Landscape and local scale transmission of E. multilocularis The incidence of HAE in a population is a clear indication of active transmission of E. multilocularis at some time in the preceding 10 – 15 years, since there may be a time lag between infection, symptoms and diagnosis, but it is not necessarily related to the current ecology of parasite transmission. In the Gansu study area, the absence of foxes and dogs after about 1993, following indirect poisoning due to rodent control, means that the parasite may now be locally extinct. To study active transmission ecology, it is necessary to examine the spatial and temporal distribution of the parasite in the definitive or intermediate hosts, or in the environment. Study of the parasite in intermediate hosts is highly problematic since
S229
parasite prevalence rates in intermediate hosts in endemic areas are normally only a few percent, and patterns of infected individuals are likely to be spatially heterogeneous and temporally highly variable. Localised high levels of infection are most likely to be caused by interaction with a highly infected individual fox, or hotspots of viable parasite eggs in environmentally favourable micro-foci [4,18]. Analysis of eggs in the environment is also difficult given their patchy distribution, variable lifetime, pathogenic nature and problematic identification. Significant progress, however, has been made in identifying E. multilocularis in definitive hosts which may have parasite prevalence rates of up to 60% in endemic areas [8]. In Europe, Japan and North America, the focus has been on foxes and a number of studies have examined the spatial distribution of E. multilocularis in foxes in relation to landscape. Spatial variation in fox infection in the Jura region of France was related to landscape variations associated with topography and altitude [2] and, using the same data set, spatial patterns of fox infection were also shown to be related to variation in grassland area when included in a spatial model based on kriging [19]. Patterns of E. multilocularis in shot foxes in Germany were found to be correlated with higher grassland pasture and lower forest land cover and, based on data from satellite imagery, related to moist areas close to rivers [20]. Local scale variation in E. multilocularis infection in foxes has also been related to landscape in peri-urban areas [21]. In China and central Asia, the infection of domestic or semi-domestic dogs is likely to be important and current research by the authors is examining the relations between dog infection, dog movement and landscape in a high altitude area on the eastern Tibetan Plateau. The relationships between landscape and small mammal population distributions have been extensively studied, but few have examined such relationships in the context of E. multilocularis transmission. Infected rodents are an indicator of parasite presence but they do not provide useful information about the infection pressure to the definitive host. Nevertheless, it has been hypothesized that landscapes supporting cyclically large populations of susceptible small mammal hosts are likely to support sustained transmission of E. multilocularis and increase the risk of human transmission [11]. Research is now underway to relate the spatial and temporal distribution of small mammal populations to landscape characteristics in endemic areas in China [5]. 4. Contribution of satellite remote sensing and spatial modeling An underpinning technology in much of the research reviewed here is the application of satellite remote sensing for landscape mapping, and the implementation of spatial modelling techniques to analyze landscape and epidemiological data. The application of satellite remote sensing in epidemiology is now well established and in the context of E. multilocularis transmission is reviewed elsewhere [12]. Key features of such data are the near-complete coverage of the Earth’s surface, the historical archive of data available which
S230
F.M. Danson et al. / Parasitology International 55 (2006) S227 – S231
enables landscape changes over the last 30 years to be mapped, the range of spatial resolution available and the potential range of quantitative information that may be extracted. All of these attributes have been exploited in research on E. multilocularis transmission. Further developments are likely as very high spatial resolution data (60 cm or less) become routinely available to facilitate sub-patch scale mapping of vegetation cover. Topographic data at 30 m resolution from the Shuttle Radar Topography Mission (SRTM) will soon be available globally and provide information on topographically controlled variables like soil moisture and solar radiation which may influence parasite egg survival. Free satellite and topographic data may now be downloaded from the World Wide Web for any place on the Earth’s surface and an increasing number of research groups are using these data as baseline maps for planning and executing field-based investigations. Parallel advances in the application of global positioning system data (GPS) are also important. Many researchers now use hand-held GPS for spatial mapping of epidemiological data and many of the studies reviewed above used GPS to map foxes, dogs, small mammals and a range of relevant landscape features. The application of GPS tracking of foxes or dogs, using collars with receivers, offers an exciting new way to examine definitive host movements in relation to landscape composition over a number of days, months or even years. Data collected in studies of E. multilocularis, or other parasites, transmission must be mapped and analyzed and there is now an increased emphasis on the application of geostatistical techniques to optimize sampling strategies and to predict spatial variation in the variables of interest. The multiple spatial and temporal scales of operation of the variables involved in the transmission of E. multilocularis presents a particular challenge. For example, regional-scale variations in climate influence the vegetation properties in an area and the small mammal species present; at the local scale, the spatial arrangement of the landscape may affect the population dynamics of susceptible small mammals; and at the patch scale, microclimatic variation may influence egg survival. It is critical that sampling of these and other relevant variables is carried out optimally so as to capture the spatial variation present whilst not oversampling the variable space. Geostatistics provides the suite of tool to address this problem [22]. 5. Conclusion The ecology of transmission of E. multilocularis is a complex process that involves time-dependent interactions of multiple hosts, at multiple spatial scales. A goal of current research is to develop a spatially explicit process-based of E. multilocularis transmission that can be used to predict transmission risk and to carry out simulations to test control scenarios. Such a model must be detailed enough to accurately capture the complexity of the transmission system, but at the same time simple enough to allow testing on real world data. The development of such a model requires landscape to be characterized in terms of the key characteristics that affect
transmission [23]. In the work reviewed in this paper, and described elsewhere in this volume, there is a rich resource of field measurements of hosts, parasites, landscapes and human disease. The challenge now is to set up the modeling framework, populate and test the model with field data and apply it to predict spatial variation in E. multilocularis transmission in China and other endemic areas across the Northern Hemisphere. Acknowledgements The authors acknowledge the U.S. National Institutes of Health and National Science Foundation for support through research grant TWO1565-02, and all co-workers who contributed to this research. References [1] Eckert J, Raush RL, Gemmell MA, Giraudoux P, Kamiya M, Liu FJ, et al. Epidemiology of Echinococcus multilocularis, Echinococcus vogeli and Echinococcus oligarthrus. In: Eckert J, Gemmell MA, Meslin F, Pawlowski ZS, editors. WHO/OIE manual on echinococcosis in humans and animals: a public health problem of global concern. Paris’ World Organisation for Animal Health; 2001. p. 164 – 82. [2] Raoul F, Deplazes P, Nonaka N, Piarroux R, Vuitton DA, Giraudoux P. Assessment of the epidemiological status of Echinococcus multilocularis in foxes in France using ELISA coprotests on fox faeces collected in the field. Int J Parasitol 2001;31:1579 – 88. [3] Veit P, Bilger B, Schad V, Schafer FW, Lucius R. Influence of environmental factors on the infectivity of Echinococcus multilocularis eggs. Parasitology 1995;110:79 – 86. [4] Giraudoux P, Delattre P, Takahashi K, Raoul F, Que´re´ JP, Craig P, et al. Transmission ecology of Echinococcus multilocularis in wildlife: what can be learned from comparative studies and multi-scale approaches? In: Craig P, Pawlowski Z, editors. Cestode zoonoses: echinococcosis and cysticercosis. An emergent and global problem. Amsterdam’ IOS Press; 2002. p. 267 – 85. [5] Giraudoux P, Pleydell DRJ, Raoul F, Que´re´ JP, Qian W, Yang Y, et al. Transmission ecology of Echinococcus multilocularis: what are the ranges of parasite stability among various host communities in China? Parasitol Int 2006;55:S237 – 46 [in this issue]. [6] Graham AJ, Danson FM, Craig PS. Ecological epidemiology: the role of landscape structure in the transmission risk of the fox tapeworm Echinococcus multilocularis (Leukart 1863) (Cestoda: Cyclophyllidea: Taeniidae). Prog Phys Geogr 2005;9:77 – 91. [7] Craig PS, Giraudoux P, Shi D, Bartholomot B, Barnish G, Delattre P, et al. An epidemiological and ecological study of human alveolar echinococcosis transmission in south Gansu, China. Acta Trop 2000;77:167 – 77. [8] Kern P, Ammom A, Kron M, Sinn G, Sander S, Petersen LR, et al. Risk factors for alveolar echinococcosis in humans. Emerg Infect Dis 2004; 10:2088 – 93. [9] Yamamoto N, Kishi R, Katakura Y, Miyake H. Risk factors for human alveolar echinococcosis: a case-control study in Hokkaido, Japan. Ann Trop Med Parasitol 2001;95:689 – 96. [10] Budke CM, Campos-Ponce M, Qian W, Torgerson PR. A canine purgation study and risk factor analysis for echinococcosis in a high endemic region of the Tibetan plateau. Vet Parasitol 2005;127:43 – 9. [11] Giraudoux P, Craig PS, Delattre P, Bao G, Bartholomot B, Harraga S, et al. Interactions between landscape changes and host communities can regulate Echinococcus multilocularis transmission. Parasitology 2003; 127:121 – 31. [12] Danson FM, Graham AJ, Pleydell DRJ, Campos-Ponce M, Giraudoux P, Craig PS. Multi-scale spatial analysis of human alveolar echinococcosis risk in China. Parasitology 2003;127:S133 – 41.
F.M. Danson et al. / Parasitology International 55 (2006) S227 – S231 [13] Zhou H. Epidemiology and transmission ecology of Echinococcus multilocularis in Xinjiang, the Peoples Republic of China: a preliminary study. 2001 PhD Thesis, Universite´ de Franche-Comte´, Besanc¸on, France. [14] Shaikenov BS, Torgerson PR. Distribution of Echinococcus multilocularis in Kazakhstan. In: Craig PS, Pawlowski ZS, editors. Cestode zoonoses: echinococcosis and cysticercosis. An emergent global problem. NATO science series, p. 299 – 307. [15] Danson FM, Craig PS, Man W, Shi D, Giraudoux P. Landscape dynamics and risk modelling of human alveolar echinococcosis. Photogramm Eng Remote Sensing 2004;70:359 – 66. [16] Graham AJ, Danson FM, Giraudoux P, Craig PS. Ecological epidemiology: landscape metrics and human alveolar echinococossis. Acta Trop 2004;91:267 – 78. [17] Pleydell DRJ, Danson FM, Craig PS, Shi, D. Spatial modeling of disease prevalence using Landsat TM data. Proceedings of the 2001 Annual Conference of the Remote Sensing and Photogrammetry Society, Remote Sensing and Photogrammetry Soc Nottingham: 396 – 406.
S231
[18] Morgan EC, Milner-Gulland EJ, Torgerson P, Medley GF. Ruminating on complexity: macroparasites of wildlife and livestock. Trends Ecol Evol 2004;19:181 – 8. [19] Pleydell DRJ, Raoul F, Danson FM, Graham A, Craig PS, Giraudoux P. Modeling the spatial distribution of Echinococcus multilocularis infection in foxes. Acta Trop 2004;91:23 – 34. [20] Staubach C, Thulke HH, Tackmann K, Hugh-Jones M, Conraths FJ. Geographic information system-aided analysis of factors associated with the spatial distribution of Echinococcus multilocularis infections of foxes. Am J Trop Med Hyg 2001;65:943 – 8. [21] Stieger C, Hegglin D, Schwarzenbach G, Mathis A, Deplazes P. Spatial and temporal aspects of urban transmission of Echinococcus multilocularis. Parasitology 2002;124:631 – 40. [22] Graham A, Danson FM, Atkinson PM. Spatial analysis for epidemiology. Acta Trop 2004;91:219 – 25. [23] Hansen F, Jeltsch F, Tackmann K, Staubach C, Thulke HH. Processes leading to spatial aggregation of Echinococcus multilocularis in its natural intermediate host Microtus arvalis. Int J Parasitol 2004;34:37 – 44.
Spatial modelling and ecology of Echinococcus multilocularis transmission in China F. Mark Danson a,*, Patrick Giraudoux b, Philip S. Craig a a
Centre for Environmental Systems Research and Biosciences Research Institute, School of Environment and Life Sciences, University of Salford, Manchester M5 4WT, UK b Department of Environmental Biology EA3184 aff. INRA, Universite´ de Franche-Comte´, 25030, Besanc¸on Cedex, France Available online 20 December 2005
Abstract Recent research in central China has suggested that the most likely transmission mechanism for Echinococcus multilocularis to humans is via domestic dogs which are allowed to roam freely and hunt (infected) small mammals within areas close to villages or in areas of tented pasture. This assertion has led to the hypothesis that there is a landscape control on transmission risk since the proximity of suitable habitat for susceptible small mammals appears to be the key. We have tested this hypothesis in a number of endemic areas in China, notably south Gansu Province and the Tibetan region of western Sichuan Province. The fundamental landscape control is its effect at a regional scale on small mammal species assemblages (susceptible species are not ubiquitous) and, at a local scale, the spatial distributions of small mammal populations. To date the research has examined relationships between landscape composition and patterns of human infection, landscape and small mammal distributions and recently the relationships between landscape and dog infection rates. The key tool to characterize landscape is satellite remote sensing and these data are used as inputs to drive spatial models of transmission risk. This paper reviews the progress that has been made so far in spatial modeling of the ecology of E. multilocularis with particular reference to China, outlines current research issues, and describes a framework for building a spatial – temporal model of transmission ecology. D 2005 Elsevier Ireland Ltd. All rights reserved. Keywords: Satellite remote sensing; Landscape ecology; Transmission
1. Introduction The fox tapeworm Echinococcus multilocularis is transmitted in a wildlife cycle that may involve multiple definitive and intermediate hosts. In the Chinese/central Asian endemic areas, definitive hosts include the red fox (Vulpes vulpes), Tibetan fox (Vulpes ferrilata) and corsac fox (Vulpes corsac) and intermediate hosts a wide range of small mammal species including for example Pika (Ochotona sp.) and vole species (Arvicolid sp.) [1]. Although intermediate host infection, from E. multilocularis eggs does not require direct contact with definitive hosts, it does require spatial (and temporal) interaction between fox home range and a landscape patch with susceptible small mammals. Foxes may visit many different patches however, and, given that one fox may harbour up to 10,000 worms [2] producing eggs for up to 40 days, and that eggs may remain
* Corresponding author. Tel.: +44 161 295 4038; fax: +44 161 295 5015. E-mail address: [email protected] (F..M. Danson). 1383-5769/$ - see front matter D 2005 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.parint.2005.11.047
viable in favourable microclimatic conditions for several months [3], environmental contamination is likely to be widespread but spatially aggregated and thus patchy at various scales [4,5]. Human infection with E. multilocularis causes the rare but highly pathogenic disease human alveolar echinococcosis (HAE) and results from the ingestion of eggs either by direct contact with infected foxes or other canid, or through contact with an egg-infected environment (Fig. 1). Viable eggs must exist within the small mammal intermediate host habitat for sustainable transmission of the parasite, but may also be distributed within the fox home range and in the human environment where transmission to domestic dogs is active [6]. Investigations of transmission to humans have used epidemiological surveys to determine risk factors for HAE [7 –9]. Although the results are not uniform, with risk factor varying from one study area to another, a common finding in China and central Asia is that dog ownership is frequently identified as a key risk factor. In areas with high rates of HAE, dog populations are large; dogs are allowed to roam freely and
S228
F.M. Danson et al. / Parasitology International 55 (2006) S227 – S231
Fig. 1. Conceptual model of transmission cycle of Echinococcus multilocularis (modified from Giraudoux et al. [4]).
hunt small mammals; and there is often close contact between dogs and humans [10]. Following the conceptual model (Fig. 1), this implies that dogs must hunt in landscape patches with susceptible small mammals, and that these patches are close to the human environment. These observations have driven the hypothesis that there is an important spatial dimension to the ecology of E. multilocularis transmission related to landscape composition, and that the risk of HAE is also affected by spatial relationships with landscape [11]. Research over the last 10 years has tested this hypothesis in a range of environments and has introduced the new technologies of satellite remote sensing and geographical information systems. It has also applied geostatistical modeling approaches to account for spatial correlation in HAE and landscape data. This paper first outlines research undertaken by the authors in an endemic area in central China. Second, it reviews the related research that is contributing to increased understanding of the spatial characteristics of E. multilocularis transmission, and third, it highlights a range of issues that is shaping the development of spatial transmission models for the parasite. 2. Spatial modeling of E. multilocularis transmission to humans in China The fundamental landscape control on E. multilocularis transmission is its effect at regional scale on the species composition of small mammal assemblages and the population dynamics of species and, at a local scale, on the spatial distributions of small mammal populations. At a regional scale, the distribution of counties with HAE across China appears to be related to the distribution of grasslands which occur in
central and western China [12]. Land cover data were obtained from the International Geosphere Biosphere Programme land cover map and HAE data from Zhou and others [13]. The qualitative correlation confirmed that the transmission of E. multilocularis to humans requires grassland landscapes supporting suitable intermediate hosts and work in Kazakhstan provided further support [14]. This scale of investigation is problematic, however, because of the incomplete records of HAE and the generalization of land cover characteristics which cannot explain the particular small mammal species assemblage found in a given grassland area. Studies at local scale have been more productive since it is possible to produce detailed landscape maps with satellite remote sensing, and data on parasite host distributions and human cases is more readily collected. In China, Craig and others [7] first suggested a link between HAE prevalence and landscape in an upland agricultural area in southern Gansu Province. Prevalence rates in 33 villages (determined by mass screening) varied from 0% to 16% across an area of approximately 40 40 km, and high rates were associated with shrubland and grassland habitat close to villages. In follow-up studies [15,16], Landsat satellite images from the 1970s, 1980s and 1990s were used to produce landscape maps of the study area and the landscape composition around the target villages was determined from classified images using spatial analysis tools. Stepwise regression analysis showed a strong correlation between the landscape composition in a buffer zone of 2000 m radius around the villages and the prevalence of HAE, and a four-class landscape model explained approximately 70% of the spatial variation in HAE [15]. However, analysis of the same data showed that there was spatial correlation in the village prevalence rates, which caused
F.M. Danson et al. / Parasitology International 55 (2006) S227 – S231 Table 1 Logistic regression of prevalence and HAE in south Gansu (modified from Danson et al. [12]) Variable
Model 1
Model 2
Parameter Significance Parameter Significance estimate ( P) estimate ( P) Intercept Age Gender Occupation Scavenging dog Tree/shrub, 750 m All rodent habitat, 1750 m
5.1 0.019 0.44 1.3 1.13
<0.0001 0.027 0.074 0.031 0.001
9.0 0.26
<0.0001 0.0008
0.0095 <0.0001 0.00031 <0.0001
an inflation of the estimate of regression model fit [17]. A Getis spatial filter was applied to account for the spatial dependence in both HAE prevalence and the landscape data, and the new model explained 59% of the spatial variation in HAE prevalence confirming that there was spatial correlation in the data, but showing that landscape was still highly significant in explaining prevalence patterns. A further development in the Gansu study area was to examine the aspatial risk factors, such as age and occupation, which had been obtained in the earlier work, and to combine these in a logistic regression model with the landscape data discussed above. Logistic regression on the aspatial data showed that age, gender, occupation and ownership of a scavenging dog were significantly related to HAE infection (Table 1, Model 1), but when landscape variables were introduced, of the aspatial variables, only age remained statistically significant (Table 1, Model 2) [12]. This result enabled the first age-specific risk map of HAE infection to be produced for the study area. Analysis of the Gansu landscape was also undertaken using a landscape metrics approach where the size, shape and spatial arrangement of land cover patches was determined from classified satellite imagery [16]. A large number of metrics was calculated and the Fmean shape index_, a metric of shape complexity, of all landscape patches in a buffer zone around villages, were found to be most strongly correlated with village HAE prevalence [16]. Landscape metrics are notoriously difficult to interpret however, and the ecological interpretation of such indices is often speculative. 3. Landscape and local scale transmission of E. multilocularis The incidence of HAE in a population is a clear indication of active transmission of E. multilocularis at some time in the preceding 10 – 15 years, since there may be a time lag between infection, symptoms and diagnosis, but it is not necessarily related to the current ecology of parasite transmission. In the Gansu study area, the absence of foxes and dogs after about 1993, following indirect poisoning due to rodent control, means that the parasite may now be locally extinct. To study active transmission ecology, it is necessary to examine the spatial and temporal distribution of the parasite in the definitive or intermediate hosts, or in the environment. Study of the parasite in intermediate hosts is highly problematic since
S229
parasite prevalence rates in intermediate hosts in endemic areas are normally only a few percent, and patterns of infected individuals are likely to be spatially heterogeneous and temporally highly variable. Localised high levels of infection are most likely to be caused by interaction with a highly infected individual fox, or hotspots of viable parasite eggs in environmentally favourable micro-foci [4,18]. Analysis of eggs in the environment is also difficult given their patchy distribution, variable lifetime, pathogenic nature and problematic identification. Significant progress, however, has been made in identifying E. multilocularis in definitive hosts which may have parasite prevalence rates of up to 60% in endemic areas [8]. In Europe, Japan and North America, the focus has been on foxes and a number of studies have examined the spatial distribution of E. multilocularis in foxes in relation to landscape. Spatial variation in fox infection in the Jura region of France was related to landscape variations associated with topography and altitude [2] and, using the same data set, spatial patterns of fox infection were also shown to be related to variation in grassland area when included in a spatial model based on kriging [19]. Patterns of E. multilocularis in shot foxes in Germany were found to be correlated with higher grassland pasture and lower forest land cover and, based on data from satellite imagery, related to moist areas close to rivers [20]. Local scale variation in E. multilocularis infection in foxes has also been related to landscape in peri-urban areas [21]. In China and central Asia, the infection of domestic or semi-domestic dogs is likely to be important and current research by the authors is examining the relations between dog infection, dog movement and landscape in a high altitude area on the eastern Tibetan Plateau. The relationships between landscape and small mammal population distributions have been extensively studied, but few have examined such relationships in the context of E. multilocularis transmission. Infected rodents are an indicator of parasite presence but they do not provide useful information about the infection pressure to the definitive host. Nevertheless, it has been hypothesized that landscapes supporting cyclically large populations of susceptible small mammal hosts are likely to support sustained transmission of E. multilocularis and increase the risk of human transmission [11]. Research is now underway to relate the spatial and temporal distribution of small mammal populations to landscape characteristics in endemic areas in China [5]. 4. Contribution of satellite remote sensing and spatial modeling An underpinning technology in much of the research reviewed here is the application of satellite remote sensing for landscape mapping, and the implementation of spatial modelling techniques to analyze landscape and epidemiological data. The application of satellite remote sensing in epidemiology is now well established and in the context of E. multilocularis transmission is reviewed elsewhere [12]. Key features of such data are the near-complete coverage of the Earth’s surface, the historical archive of data available which
S230
F.M. Danson et al. / Parasitology International 55 (2006) S227 – S231
enables landscape changes over the last 30 years to be mapped, the range of spatial resolution available and the potential range of quantitative information that may be extracted. All of these attributes have been exploited in research on E. multilocularis transmission. Further developments are likely as very high spatial resolution data (60 cm or less) become routinely available to facilitate sub-patch scale mapping of vegetation cover. Topographic data at 30 m resolution from the Shuttle Radar Topography Mission (SRTM) will soon be available globally and provide information on topographically controlled variables like soil moisture and solar radiation which may influence parasite egg survival. Free satellite and topographic data may now be downloaded from the World Wide Web for any place on the Earth’s surface and an increasing number of research groups are using these data as baseline maps for planning and executing field-based investigations. Parallel advances in the application of global positioning system data (GPS) are also important. Many researchers now use hand-held GPS for spatial mapping of epidemiological data and many of the studies reviewed above used GPS to map foxes, dogs, small mammals and a range of relevant landscape features. The application of GPS tracking of foxes or dogs, using collars with receivers, offers an exciting new way to examine definitive host movements in relation to landscape composition over a number of days, months or even years. Data collected in studies of E. multilocularis, or other parasites, transmission must be mapped and analyzed and there is now an increased emphasis on the application of geostatistical techniques to optimize sampling strategies and to predict spatial variation in the variables of interest. The multiple spatial and temporal scales of operation of the variables involved in the transmission of E. multilocularis presents a particular challenge. For example, regional-scale variations in climate influence the vegetation properties in an area and the small mammal species present; at the local scale, the spatial arrangement of the landscape may affect the population dynamics of susceptible small mammals; and at the patch scale, microclimatic variation may influence egg survival. It is critical that sampling of these and other relevant variables is carried out optimally so as to capture the spatial variation present whilst not oversampling the variable space. Geostatistics provides the suite of tool to address this problem [22]. 5. Conclusion The ecology of transmission of E. multilocularis is a complex process that involves time-dependent interactions of multiple hosts, at multiple spatial scales. A goal of current research is to develop a spatially explicit process-based of E. multilocularis transmission that can be used to predict transmission risk and to carry out simulations to test control scenarios. Such a model must be detailed enough to accurately capture the complexity of the transmission system, but at the same time simple enough to allow testing on real world data. The development of such a model requires landscape to be characterized in terms of the key characteristics that affect
transmission [23]. In the work reviewed in this paper, and described elsewhere in this volume, there is a rich resource of field measurements of hosts, parasites, landscapes and human disease. The challenge now is to set up the modeling framework, populate and test the model with field data and apply it to predict spatial variation in E. multilocularis transmission in China and other endemic areas across the Northern Hemisphere. Acknowledgements The authors acknowledge the U.S. National Institutes of Health and National Science Foundation for support through research grant TWO1565-02, and all co-workers who contributed to this research. References [1] Eckert J, Raush RL, Gemmell MA, Giraudoux P, Kamiya M, Liu FJ, et al. Epidemiology of Echinococcus multilocularis, Echinococcus vogeli and Echinococcus oligarthrus. In: Eckert J, Gemmell MA, Meslin F, Pawlowski ZS, editors. WHO/OIE manual on echinococcosis in humans and animals: a public health problem of global concern. Paris’ World Organisation for Animal Health; 2001. p. 164 – 82. [2] Raoul F, Deplazes P, Nonaka N, Piarroux R, Vuitton DA, Giraudoux P. Assessment of the epidemiological status of Echinococcus multilocularis in foxes in France using ELISA coprotests on fox faeces collected in the field. Int J Parasitol 2001;31:1579 – 88. [3] Veit P, Bilger B, Schad V, Schafer FW, Lucius R. Influence of environmental factors on the infectivity of Echinococcus multilocularis eggs. Parasitology 1995;110:79 – 86. [4] Giraudoux P, Delattre P, Takahashi K, Raoul F, Que´re´ JP, Craig P, et al. Transmission ecology of Echinococcus multilocularis in wildlife: what can be learned from comparative studies and multi-scale approaches? In: Craig P, Pawlowski Z, editors. Cestode zoonoses: echinococcosis and cysticercosis. An emergent and global problem. Amsterdam’ IOS Press; 2002. p. 267 – 85. [5] Giraudoux P, Pleydell DRJ, Raoul F, Que´re´ JP, Qian W, Yang Y, et al. Transmission ecology of Echinococcus multilocularis: what are the ranges of parasite stability among various host communities in China? Parasitol Int 2006;55:S237 – 46 [in this issue]. [6] Graham AJ, Danson FM, Craig PS. Ecological epidemiology: the role of landscape structure in the transmission risk of the fox tapeworm Echinococcus multilocularis (Leukart 1863) (Cestoda: Cyclophyllidea: Taeniidae). Prog Phys Geogr 2005;9:77 – 91. [7] Craig PS, Giraudoux P, Shi D, Bartholomot B, Barnish G, Delattre P, et al. An epidemiological and ecological study of human alveolar echinococcosis transmission in south Gansu, China. Acta Trop 2000;77:167 – 77. [8] Kern P, Ammom A, Kron M, Sinn G, Sander S, Petersen LR, et al. Risk factors for alveolar echinococcosis in humans. Emerg Infect Dis 2004; 10:2088 – 93. [9] Yamamoto N, Kishi R, Katakura Y, Miyake H. Risk factors for human alveolar echinococcosis: a case-control study in Hokkaido, Japan. Ann Trop Med Parasitol 2001;95:689 – 96. [10] Budke CM, Campos-Ponce M, Qian W, Torgerson PR. A canine purgation study and risk factor analysis for echinococcosis in a high endemic region of the Tibetan plateau. Vet Parasitol 2005;127:43 – 9. [11] Giraudoux P, Craig PS, Delattre P, Bao G, Bartholomot B, Harraga S, et al. Interactions between landscape changes and host communities can regulate Echinococcus multilocularis transmission. Parasitology 2003; 127:121 – 31. [12] Danson FM, Graham AJ, Pleydell DRJ, Campos-Ponce M, Giraudoux P, Craig PS. Multi-scale spatial analysis of human alveolar echinococcosis risk in China. Parasitology 2003;127:S133 – 41.
F.M. Danson et al. / Parasitology International 55 (2006) S227 – S231 [13] Zhou H. Epidemiology and transmission ecology of Echinococcus multilocularis in Xinjiang, the Peoples Republic of China: a preliminary study. 2001 PhD Thesis, Universite´ de Franche-Comte´, Besanc¸on, France. [14] Shaikenov BS, Torgerson PR. Distribution of Echinococcus multilocularis in Kazakhstan. In: Craig PS, Pawlowski ZS, editors. Cestode zoonoses: echinococcosis and cysticercosis. An emergent global problem. NATO science series, p. 299 – 307. [15] Danson FM, Craig PS, Man W, Shi D, Giraudoux P. Landscape dynamics and risk modelling of human alveolar echinococcosis. Photogramm Eng Remote Sensing 2004;70:359 – 66. [16] Graham AJ, Danson FM, Giraudoux P, Craig PS. Ecological epidemiology: landscape metrics and human alveolar echinococossis. Acta Trop 2004;91:267 – 78. [17] Pleydell DRJ, Danson FM, Craig PS, Shi, D. Spatial modeling of disease prevalence using Landsat TM data. Proceedings of the 2001 Annual Conference of the Remote Sensing and Photogrammetry Society, Remote Sensing and Photogrammetry Soc Nottingham: 396 – 406.
S231
[18] Morgan EC, Milner-Gulland EJ, Torgerson P, Medley GF. Ruminating on complexity: macroparasites of wildlife and livestock. Trends Ecol Evol 2004;19:181 – 8. [19] Pleydell DRJ, Raoul F, Danson FM, Graham A, Craig PS, Giraudoux P. Modeling the spatial distribution of Echinococcus multilocularis infection in foxes. Acta Trop 2004;91:23 – 34. [20] Staubach C, Thulke HH, Tackmann K, Hugh-Jones M, Conraths FJ. Geographic information system-aided analysis of factors associated with the spatial distribution of Echinococcus multilocularis infections of foxes. Am J Trop Med Hyg 2001;65:943 – 8. [21] Stieger C, Hegglin D, Schwarzenbach G, Mathis A, Deplazes P. Spatial and temporal aspects of urban transmission of Echinococcus multilocularis. Parasitology 2002;124:631 – 40. [22] Graham A, Danson FM, Atkinson PM. Spatial analysis for epidemiology. Acta Trop 2004;91:219 – 25. [23] Hansen F, Jeltsch F, Tackmann K, Staubach C, Thulke HH. Processes leading to spatial aggregation of Echinococcus multilocularis in its natural intermediate host Microtus arvalis. Int J Parasitol 2004;34:37 – 44.