Chapter 4 CASE STUDIES

In the past decade a series of relevant field studies were conducted that contribute to the understanding of West African fly ecology and spatial epidemiological patterns: the Togo national study, the study of the Sideradougou pastoral area in Burkina Faso and the Mouhoun river basin study in Burkina Faso.

Togo

The Togo study was conducted during the 1990s as part of a Belgium-funded FAO trypanosomiasis control project (GCPTOG-013-BEL). The holistic approach developed to collate field data on flies, disease and cattle at a 0.125 degree resolution across the country provides us with a unique data set for analysis (Figure 5). Published results include an analysis of fly distribution patterns (Napala et al., 1995; Hendrickx et al., 1999a), the study of the spatial epidemiology of trypanosomiasis in the wider sense (Hendrickx et al., 1999b), the mapping of trypanotolerant breeds (Hendrickx et al., 1996; Dao, 1998) and the use of remote sensing to contribute to the reduction of field surveys for tsetse (Hendrickx et al., 2001b) and trypanosomiasis (Hendrickx et al., 2000). Based on these results a georeferenced decision-support methodology for trypanosomiasis management was developed (Hendrickx et al., 1997, 1999b, 1999c; Hendrickx, 1999).

Spatial epidemiology

The traditional response to trypanosomiasis in West Africa has been trypanotolerance (Hoste et al., 1988; Hoste, 1992) and grazing management allowing cattle raising without the need for constant or regular veterinary care in tsetse-infested areas. Results discussed in the Togo spatial epidemiological study (Hendrickx et al., 1999b) suggested that trypanotolerant taurine cattle did indeed make a contribution towards reducing the impact of the disease in Togo, but this still did not enable farmers to exploit fully the production potential of their cattle. The presence of tsetse in the country remained a major obstacle to cattle raising and particularly constrained mixed farming development and intensification (Figure 6a and b). As a comparison, data from outside the tsetse belt (Mali, the Niger, Nigeria, the Sudan and Chad) show a very strong positive relationship between sedentary cattle densities and agriculture intensity levels (Wint and Bourn, 1994).

Also of particular relevance was the fact that riparian tsetse appeared to be less susceptible to crop encroachment in the humid south (1 300 mm annual rainfall), i.e. densities of G. p. palpalis remained high and unrelated to cropping intensity levels, while 600 km up-country in the dry northern part of Togo (1 000 mm annual rainfall) G. tachinoides densities dropped with increasing cropping intensity levels.

Livestock production systems

The distribution of animal husbandry systems was analysed (Figure 5). Obtained spatial patterns revealed two major systems (Hendrickx et al., 1999b): (i) a traditional extensive village system, prevailing in the northern half of the country, where cattle owned by a group or family are a patrimony for future generations. This system may or may not include the presence of draught oxen, and (ii) a semimodern, more market-oriented system where ownership is often individual (civil servants, traders) and savings are invested in cattle. This analysis did not include the seasonal presence of transhumance cattle herds.

Interestingly, the difference between the two systems also depends on the proportion of trypanotolerant taurine cattle out of total cattle. There was a higher proportion of taurine cattle in the traditional rural system while cross-bred were more numerous in the market-oriented system. This difference between the data sets was particularly true for bulls, highlighting distinct purpose-oriented livestock management strategies and different levels of risk taking, i.e. the risk of diluting trypanotolerance in a tsetse-infested area, once given access to veterinary inputs.

Decision-support systems

The results summarized above clearly stress the fact that trypanosomiasis management should not simply be regarded as a stand-alone animal health problem. Any control measure should be adapted to the animal husbandry system, and integrated into the wider perspective of the improvement of basic animal health and production, including the use of trypanotolerant breeds, and the integration of livestock and crop agriculture.

Hence the developed decision-support system integrates these various aspects (Hendrickx et al., 1997, 1999c; Hendrickx, 1999). A first model was developed to set priorities for trypanosomiasis management (Figure 7 top maps), i.e. veterinary follow-up of valuable cattle in rural systems (e.g. draught oxen) and of marketoriented (mainly cross-bred) herds. The resulting map (Output 1) highlighted areas where improved animal health follow-up should be of major concern. The second model looked at priorities to boost mixed farming practices (Figure 7 bottom maps) and is mainly relevant for rural systems. The resulting map (Output 2) highlighted areas where removal of existing disease constraints, in which trypanosomiasis prevails, might benefit most the development of mixed farming practices. Finally, particular care should be taken with regard to environmentally fragile areas, i.e. areas where the level of soil degradation is such that the further development of livestock activities, should trypanosomiasis control be a success, may result in irreversible environmental damage.

This problem was addressed during a pilot study (Van Camp, Biesemans and De Wulf, 1999; Van Camp et al., 2002) conducted in two areas in the northern part of Togo (Figure 8). Using a data set derived from ground validated satellite imagery (Landsat TM) and other physical land-resource databases, a mathematical model based on fuzzy relational computation was developed. As a result cantons, the smallest administrative unit, chosen as “application units”, could be ranked according to their environmental fragility (Figure 8), further refining the decision-making process. It is important to highlight here that the interpretation of results of this type may not always be as straightforward as it appears. “High environmental fragility” does not necessarily mean, “refrain from control of trypanosomiasis”. On the contrary, it may suggest, “improve the integration of livestock and crop agriculture towards environmentally more sustainable practices”. Sustainable practices include sound trypanosomiasis management.

These various results provided evidence of the importance of planning adapted T&T control activities at a national level (Hendrickx et al., 1997; Napala et al., 1999; Batawui, 1999; Kouagou et al., 2000; Hendrickx, Batawui and De Deken, 2002; Bastiaensen et al., 2003; Batawui et al., 2003).

• National trypanosomiasis extension campaign

  - Target population (nationwide trypanosomiasis management priorities, Figure 7, top maps): draught oxen, market-oriented herds and advanced village herds.

  - Method: series of posters depicting extension topics such as: (i) tsetse and trypanosomiasis, (ii) diagnostic-driven trypanocidal treatment, (iii) draught oxen and (iv) options for tsetse control (pour-on, insecticide-treated targets). These posters were used as a basis for household discussions led by trained extension workers and/or private veterinarians.

  - Implemented by: private rural veterinarians - self-installed but promoted by the project - as well as official extension workers and existing “barefoot veterinarians”, in areas not covered by private veterinarians.

  - Results obtained: important reduction in trypanosomiasis prevalence, less spectacular results on average herd health.

• Integrated T&T control

  - Target population: selected village clusters in mixed farming priority areas.

  - Method: community participatory approach combining regular veterinary follow-up and epicutaneous use of insecticide (pour-on technique) scheme.

  - Implemented by: community supervised by private and government veterinarians.

  - Results obtained: while the application of insecticide on cattle in small village clusters was successful in controlled experimental conditions (Batawui, 1999; Hendrickx, Batawui and De Deken, 2002; Batawui et al., 2003), in practice pouron schemes required permanent monitoring to compensate for organizational problems at the village level (Bastiaensen, personal communication).

Burkina Faso - Sideradougou agropastoral area

The Sideradougou agropastoral area (1 200 km², Figure 9) is situated south of Bobo-Dioulasso in western Burkina Faso, and is representative of the Sudanese ecoclimatic zone, with wooded savannah and bushy riparian gallery along the riverbanks. Because of a significant curve of the isohyets in the medium longitudes of West Africa, and despite comparative latitudes, this area is more humid than northern Togo.

Comparative analysis of remote sensing data showed significant land-use pattern changes over the past 20 years owing to increasing demographic pressure and immigration. This effect was mainly noticed in the eastern part, which is under considerable pressure from agricultural activities (Figure 9) such as cotton growing, a major cash crop. Savannah flies, G. morsitans submorsitans, the most sensitive species to crop encroachment, disappeared from the area and only riparian fly species (e.g. G. p. gambiensis and G. tachinoides) persist. The description of their distribution patterns was the initial objective of the “Sideradougou project” conducted between 1996 and 2000 in collaboration between the Centre de Coopération Internationale en Recherche Agronomique pour le Développement (CIRAD) and the Centre International de Recherche-Développement sur l’Élevage en Zone Subhumide (CIRDES) (de la Rocque et al., 2001a).

Change in riparian tsetse distributions with cultivation

The 1996 study was based on an ecological and entomological survey during the dry season. Along 120 km of the major river systems, biconical traps were placed at 100-metre intervals. All the results were integrated in a geographical information system. For each separate trap site, densities of riparian flies were correlated to ecological characteristics of the habitat (vegetation, river beds, waterflow patterns) and the most favourable habitats were identified for each species (de la Rocque, 1997).

Results obtained were compared with results from a similar survey conducted in 1982 (Cuisance et al., 1984) and tsetse distribution and abundance changes were related to environmental changes (de la Rocque et al., 2001b). Briefly, substantial and contrasting differences in tsetse fly distribution and density were observed along the river systems (Figure 9). On the western branch of the Koba river (Nakaka area), populations of both riparian species (G. tachinoides and G. p. gambiensis) were quite similar in terms of density and distribution between 1982 and 1996 whereas on the eastern branch (Ouara area), an increase of more than 2.5 times for G. tachinoides and 1.5 times for G. p. gambiensis occurred. Along the Tole river, G. p. gambiensis had almost disappeared while G. tachinoides was still found, although at reduced densities.

These contrasting tsetse dynamics between the Tole and eastern Koba river systems were surprising, considering both rivers run in areas of similar land-use intensity levels. Analysis revealed that tsetse density reductions were related to cropping distance from watercourses (Figure 10).

This strong link between localized agricultural development and impact on tsetse is not only due to direct human activities (wood cutting, hunting, drawing of water, use of insecticides), but also, and maybe mainly, to indirect impact induced by amplification of rainwater runoff. Land clearance and reduced plant cover exacerbate the effects of rainfall, which rapidly modifies the soil microstructure, renders it less permeable and finally exacerbates runoff (Riou, 1990). This favours riverbank erosion and the uprooting of riverside trees, essential components of riparian tsetse habitats.

This pattern of erosion was observed when the distance between the rivers and the cultivated plots was approximately ten metres. It was particularly marked along the Tole river. The resulting gaps in tree cover reduce the buffer effect played by the vegetation in relation to the drier conditions in the savannah and modify the abiotic conditions within the galleries. These changes in tsetse fly population dynamics as a whole can lead to the local disappearance of a fly species, as was observed with G. p. gambiensis. On the other hand, G. tachinoides seems better suited to forest environments transformed by human activity.

This situation contrasted with the conditions along the eastern branch of the Koba, where the cultivated plots were 300 metres or more from the river. The difference of valley landscape was mainly due to differences in soil characteristics. The Tole drainage system has good soils for cropping that extend close to the river. In contrast, on the Koba river alluvial soil formation, known as “yellow bank soil”, is a capping loamy soil that is difficult to till (Guillobez, 1979). As a consequence of the limited expansion of agriculture nearby the river, the gallery forests were not degraded. In such conditions, if habitat is preserved, increase of human and cattle densities may even favour tsetse presence by increasing feeding opportunities.

In conclusion, studies at various scales are required to forecast correctly tsetse distribution and abundance (Hendrickx et al., 2001a). Global analysis of land use is very useful, especially for the savannah species. For the riparian species, results from the Sideradougou study highlight the importance of local factors measured at high spatial resolution.

Contrasting epidemiological patterns at the herd level

Dissection of tsetse, and polymerase chain reaction (PCR) analysis of the infected organs, showed that infection patterns differ locally (Table 2) (Lefrançois et al., 1998). In areas where agricultural activity density is low (western Sideradougou), tsetse flies frequently feed on wild reptiles (monitor lizards or crocodiles) and rarely carry pathogenic trypanosomes. In areas with more dense agricultural activity (eastern Sideradougou), most flies feed upon cattle and pigs, and are highly infected by pathogenic parasites. Riparian tsetse flies appear to be opportunistic feeders, showing little host preferences and feeding on the most available host. This opportunism governs the interaction between vectors, domestic hosts and parasites. Consequently, entirely different epidemiological patterns may occur in areas only a few kilometres apart (de la Rocque, 1997).

TABLE 2
Epidemiological differences between eastern and western Sideradougou

¹ Parasites were identified using polymerase chain reaction (PCR). The PCR identification rate is the percentage of samples. recognized by the specific PCR primers for the parasites known in the area. Samples that have not been identified are considered as non-pathogenic parasites for cattle, e.g. reptile trypanosome species.

² Blood meal analysis was carried out using the enzyme-linked immunosorbent assay (ELISA) method.

A multidisciplinary approach to risk evaluation

Very different epidemiological situations thus occur on a fine scale. The risk of transmission primarily depends on the intensity of contact between hosts and vectors (de la Rocque et al., 1999). All the factors influencing this contact in time and space have to be considered (Laveissière, Couret and Hervouet, 1986). Some of these factors are related to the physical environment (riparian vegetation, structure of the rivers, geomorphologic profile, distribution and abundance of tsetse flies) and others to human activity (agricultural occupation, breeding systems, pastoral management). Epidemiological hot spots, i.e. major disease transmission foci areas, might thus only accurately be revealed through a multidisciplinary approach that includes all major components of the agro-ecological and socio-economic systems involved (de la Rocque et al., 2001a).

Towards highlighting epidemiological hot spots

To reveal these epidemiological spots, a model was proposed using GIS and spatial analysis tools.

• The most favourable habitats for each of the two species of riparian flies were identified using high-resolution remote sensing data (SPOT XS). However, at this scale forest galleries remain difficult to characterize because their width (from 20 to 40 metres) is often at the limit of the satellite resolution. Therefore a more general approach aiming at landscape recognition within valleys was developed and allowed us to segment the whole river network according to identified valley types. The most suitable landscapes for tsetse were recognized (see Figure 11).

• In parallel, an exhaustive and georeferenced census of all the stockbreeders, agropastoralists and farmer settlements was carried out in the field. Breeding systems were characterized and grazing areas were assigned to each of the settlements (Michel et al., 1999). These areas take into account the various breeding practices and in particular water management. A model was developed to establish a map of spatial distribution of cattle in the area (Figure 11).

• The overlay of tsetse and cattle layers revealed major host- vector interaction sites (i.e. host-vector interface) (Figure 11). It was concluded that, of the 120-km river system prospected in this study, 15 percent of it appears to harbour higher disease transmission risk (de la Rocque et al., 2001a).

The relevance of these results was validated by comparing identified epidemiological hot spots with the PCR dissection results and with parasitological and serological prevalence studies (Figure 11) (Michel et al., 2002). There are significantly more flies infected by the three species of pathogenic parasites in the red zones. This translates into a four-fold risk increase for cattle living or entering in the red zones.

Vector control in targeted areas

This study revealed that different epidemiological situations occur at the kilometre scale. This diversity is related to the narrow ecological requirements of the riparian flies, which are localized in circumscribed habitats. At this scale the actual rate of vector-host contact appears to be more determinant than fly density alone for risk evaluation.^[6] This finding may be of particular benefit with regard to fly control. To test this hypothesis a pilot trial was conducted during the last two years. The aim was to focus control operations only on identified epidemiological hot spots.

The developed control strategy was based on two principles:

• tsetse control aimed at suppression of tsetse populations below the threshold level where parasites cannot circulate thus leading to a significant decrease of infection prevalence in cattle;

• tsetse control was conducted only in identified epidemiological hot spots where rural populations do experience a trypanosomiasis problem, a prerequisite to obtain community involvement and a major factor for reducing costs of an eventual disease control campaign.

Two fly/disease control methods were integrated:

• impregnated targets along the drainage line at intervals of 100 metres, covering 2 or 3 km of the hot spot;

• epicutaneous treatment with pour-on formulations (live bait technique) of sedentary cattle visiting the hot spots daily.

The latter method was used only during the rainy season, when the tick burden is at its peak and may constitute an additional incentive to livestock holders.

The pilot trial was conducted in two distinct areas (Figure 12). The first site was located in a large forest gallery particularly suitable to tsetse, the second in a small and degraded gallery. A third site, located between the two, was used to evaluate the dispersion of the control effect.

Results were monitored by monthly entomological surveys and parasitological follow-up of sentinel herds. Preliminary results obtained for tsetse densities and trypanosomiasis prevalence (Figure 12) show some significant trends. An additional interesting result is the evolution of the antibody seroprevalence in the sentinel herds of site 1, decreasing from 85 percent to 60 percent in less than one year (data not shown).

In conclusion, such a targeted vector-control approach appears to be efficient, not for tsetse eradication, but for decreasing tsetse densities below the threshold of significant parasite transmission. Such a result, however, is fragile and supposes permanent effort by the beneficiaries. Socio-economic surveys are actually conducted for an appraisal of the livestock owners’ perception with regard to the control interventions and their sustainability. Their willingness to contribute either through labour or through financial participation should allow the creation of local associations charged with the continuation of control activities.

Burkina Faso - Mouhoun river basin

Following results obtained in both projects described above, activities were extended to drier areas of Burkina Faso, for example the Mouhoun river basin, formerly known as the Black Volta (Figure 13).

Epidemiological patterns at the river-basin level

A series of entomological and protozoological field surveys were conducted during the FAO regional project GCP-RAF-347-BEL, a follow-up of the Togo project GCP-TOG-013-BEL. The aim was to identify epidemiological patterns at the river-basin level. Results are summarized in Figure 13. Data collected by CIRDES teams in the Kenedougou province (joint CIRDES-ILRI-Bundesministerium für Wirtschaftliche Zusammenarbeit und Entwickling [BMZ] project, McDermott, personal communication) and at additional sites along the Mouhoun river (joint CIRAD-CIRDES project, Bouyer, personal communication) are also included.

The adapted methodology for identifying spatial epidemiological patterns is summarized in the legend of Figure 13 (for more details see Tamboura et al., 1999; Hendrickx and Tamboura, 2000; Tamboura, Béré and Hendrickx, 2000). The obtained map of epidemiological patterns shows a distinct trend. Trypanosomiasis mostly constitutes a problem along the main course of the Mouhoun stream from source to downstream. Different epidemiological classes appear not to be scattered randomly but as a rule the problem becomes progressively less “intense” when moving to smaller tributaries. The observed prevalence of trypanosomiasis in cattle drops (from red to pink areas) followed by the percentage of anaemic animals (from light blue to dark blue).

The combination of both parasitaemic data (positive for infection) and data on anaemia (the main occurring symptom in trypanosomeinfected animals) has the advantage of correcting for false negative samples, which may be due to the use of the relatively low sensitivity parasitological tests (wet smear - buffy coat - dark ground microscopy [Murray, Murray and McIntyre, 1977]). Interestingly, no areas were found with high parasitaemia and low prevalence of anaemic animals, stressing the link between trypanosomiasis occurrence and presence of anaemic animals in field situations. Nevertheless, other diseases may also cause anaemia and therefore the model may be considered not only as an estimation of the trypanosomiasis problem but also as a measurement of the general health status of the surveyed herds.

Tsetse ecology model

In a further study of the Mouhoun area, tsetse data from various available sources were used to model the spatial distribution in western Burkina Faso and southeastern Mali of the different fly ecology classes already discussed (see Figure 13). For more details on the rationale and approaches used see Hendrickx (2001). General information about discriminant analysis used to build the spatial prediction model is given in Hendrickx (1999).

The results obtained suggest a series of horizontal bands depicting the various fly-ecology classes. Interestingly some important matches can be made with the independently created epidemiological pattern map (Figure 13):

• a widespread trypanosomiasis problem at the “sources du Mouhoun” where riparian tsetse are ubiquitous;

• trypanosomiasis mainly restricted to linear fly distribution areas along the main stream.

Both results were important indicators suggesting pathways for future research towards further understanding epidemiological and fly-ecology patterns in this part of West Africa.

Towards understanding tsetse fragmentation

As may be suggested by the results described above, the “Boucle du Mouhoun” has a series of characteristics that make it particularly suitable as a study area representative of the dry northern part of the West African tsetse belt.

• It is entirely within the fly belt.

• It extends over 180 km from north to south within a rainfall range of 1 300-1 400 mm at the Mouhoun sources, 1 000 mm in the northern part of the river bend and 1 200 mm in the eastern leg.

• It includes a great variety of landscapes and valley profiles including large flood areas in the western ascending leg and highly cultivated areas with protected forest patches in the descending eastern leg.

Within given distribution limits, the distinction between suitable and non-suitable tsetse habitat may depend on apparently minor changes in ruling ecoclimatic settings. In areas where macroclimatic conditions are generally unfavourable, tsetse become highly susceptible to occurring habitat changes, for example, a breach in riparian vegetation continuity may induce rapid changes in tsetse populations. Physiological population fitness indicators such as body size or age group distribution patterns allow for a certain level of quantification of population stress (Rogers and Randolph, 1991). Preliminary results of population studies conducted on several sites by a team from CIRDES suggest that though actual population fragmentation is rarely observed along the Mouhoun main stream, significant age structure differences occur in areas subject to environmental stress (Bouyer, personal communication).

• In its northern part, though macroclimatic conditions appear least favourable (1 000 mm annual rainfall), large patches of riparian forests remain and this natural habitat is protected from human impact resulting in locally abundant healthy tsetse populations.

• Highest habitat stress is recorded in the eastern descending leg. Riparian forests are degraded because of soil erosion. Intensity depends on soil characteristics, slope, water flow intensity and agriculture. Tsetse populations are reduced and show a younger age structure.

• Along the western ascending leg, the main stream is surrounded by natural dams creating flood areas. This has a positive impact on erosion and vegetation is luxuriant. Tsetse populations are abundant with normal age structures.

In contrast, along the Mouhoun tributaries isolated riparian habitat patches separated by several kilometres of cultivated wetlands have been identified using satellite imagery and field surveys. As a next stage, population studies are planned in these areas. These will include genetic studies and modelling of fly dispersal patterns in time and space. Such studies will yield crucial data towards planning sustainable fly suppression and may be the first step towards a removal strategy.