The first outbreak of PPR in sheep and goats in Tanzania was confirmed in December 2008. Since then, sporadic outbreaks have been reported in many parts of the country. The three fundamental steps of exotic pathogen emergence are incursion, persistence and spread in a geographical context. In this article, incursion refers to the first confirmed arrival of PPR virus (PPRV) in the country, persistence refers to the obstinate survival of PPRV in sheep and goat populations in spite of vaccination and other biosecurity measures undertaken to eliminate the virus, and spread refers to the movement of PPRV between different geographical localities. We use data on the first occurrence and the subsequent events of PPR incursion and spread in Tanzania to illustrate the epidemiological factors that drive these sequential events amongst the pastoral sheep and goat population. The epidemiological data, clinical and pathological features of this outbreak and recovery and identification of PPRV are also reported.

However, reports of increased deaths in sheep and goats continued in the Ngorongoro and Mara districts. In June 2008, a second team was dispatched to the Ngorongoro and Mara districts, where, respectively, 404 and 84 serum samples were collected. Of the 404 samples collected in the Ngorongoro district, 129 (31.93%) tested positive for PPR during serological screening. None of the samples from the Mara district tested positive. Based on these rapid epidemiological assessment (REA) results, the director of veterinary services designed a cross-sectional study that was aimed at gauging the magnitude and extent of PPR spread in the country.

All sheep and goat flocks in the sampling area formed the population from which study flocks were randomly selected. Multistage sampling (Dohoo, Martin & Stryhn 2003), with four hierarchical stages, was used as sampling strategy. The first level of selection was the region; the Arusha, Kilimanjaro, Manyara and Tanga regions were purposely selected. Within each of the selected regions, specific districts (Karatu, Longido, Mbulu, Monduli, Ngorongoro, Siha, Simanjiro, Korogwe, Lushoto, Mkinga, Muheza and Tanga-municipal) were also purposely selected. Villages were randomly selected. Within each of the selected villages, 40 animals (either sheep or goats) were randomly selected. The study regions and districts were purposely selected based on perceived risk due to either their bordering Kenya or the presence of farming activities characterised by extensive livestock movement.

Owing to the characteristics of the pastoral and agro-pastoral livestock production systems, it is reasonable to consider all the sheep and goats in a village as one flock, with very little within-flock variation. A village was therefore considered to constitute a flock. The sample size was computed as suggested by Dohoo et al. (2003):

However, PPR is a highly contagious disease. To obtain a similar accuracy to that with simple random sampling, the sample size had to be recalculated. The aim was to sample at least 40 animals from each village, and from the REA study the average rate of PPR homogeneity (p) was estimated to be 0.156. The sample size was subsequently recalculated as:

The nucleic acid sequences obtained from PCR products amplified with the Nad1–Nad2 primer were aligned with sequences from PPRV strains that were maintained in our database. Phylogenetic analysis was performed on the 224 nucleotides located on the nucleoprotein gene of the virus. The known PPRV sequences were derived from strains isolated over a period of 20 years. They consisted of ‘historical’ strains, such as the Nigeria 75/1, from which the PPR vaccine was developed, and also the Dorcas strain isolated from a Dorcas gazelle in a zoological collection in the United Arab Emirates, (Furley, Taylor & Obi 1987). Strains isolated between 1988 and 2000 originated from Cameroon, India, Senegal, Turkey, Ghana, Guinea Bissau, Oman and the Ivory Coast. The alignment of nucleotide sequences was created by means of the Clustal W program (Vector NTI, Informax Inc., Rockville). Phylogenetic analysis was carried out as described by Kwiatek et al. (2007). Bootstrap confidence values were calculated on 1000 replicates. The RBOK vaccine strain of rinderpest virus was considered as an out-group.

In this study, R_t was estimated from the average age at seroconversion by the method of Dietz (1993). The deterministic formula used was:

The following assumptions were applied:

• From an epidemiological perspective, the small ruminant (sheep and goat) population is assumed to be a single, homogeneous, free-mixing host population (referred to as ‘the flock’) and contacts between individuals were random.

• The kidding and lambing rates were high enough to sustain an input of susceptible hosts into the population, thereby maintaining a constant chain of infection in the population.

• Screening of 250 sheep and 600 goat samples, randomly selected from the 2006–2008 sera that were collected from the study area during rinderpest elimination campaigns, yielded negative results for PPRV antibodies. It was therefore logical to assume that the study animals were seronegative during the 12 months before sampling, and therefore fully susceptible to the invading PPRV infection.

• The duration of the infectious period is not related to age.

• The deterministic model did not consider the probability of re-invasion of the recovered flock.

The probability of persistence or spread, p_pers, over the next generation was computed as p_pers = 1 – (1/R_t)ⁿ, where n is the number of infectious foci (the number of villages that reported PPR cases) and R_t equals R₀ when there is no immunity or previous infection within the population. The threshold level of vaccination (V_t) necessary to eradicate the disease was calculated according to V_t = 1 – R_t^–1, as described in Anderson and May (1991).

Competitive ELISA showed that antibodies against PPRV were present in 59% of the 217 serum samples and PPRV protein was detected in 34% of the screened ocular and nasal swabs. By March 2010 the disease had spread to the nearby districts of Tandahimba and Masasi, where, respectively, 84% of the 180 and 3% of the 125 samples screened by cELISA yielded positive results for antibodies against PPRV.

In April 2010 about 400 small ruminants were reported to have died because of an unknown disease in two divisions of Malinyi and Mtimbira in the Ulanga district of the Morogoro region. Epidemiological investigations suggested that the disease was PPR, which invaded the region through returning migratory pastoralists who had moved northwards from the Mtwara and Lindi regions. Serological investigations revealed the presence of antibodies against PPRV in 152 (76%) of the 200 serum samples tested by means of cELISA. In 2011, antibodies to PPRV were demonstrated in trade cattle sharing common grazing lands with 225 small ruminants in the Mukuranga district of the Coast region. Although not susceptible to the disease, cattle are able to seroconvert in case of a high prevalence of PPR in small ruminants (Abubakar et al. 2011).

In October 2010 the Director of Veterinary Services conducted risk-based serosurveillance in all districts that border Mozambique, Malawi and Zambia. Samples were collected from animals at randomly selected high-risk areas, such as border crossings, auctions, slaughter slabs and grazing and watering points, and other areas where animals from a number of villages congregate. A total of 720 serum samples were collected under the Vaccine for Control of Neglected Animal Diseases in Africa (VACNADA) programme (funded by the European Union) and screened for antibodies against PPRV at the PANVAC laboratories in Addis Ababa, Ethiopia. Of the 720 samples, 266 (36.9%) tested positive for antibodies against PPRV. Clinical disease was detected in the Mahuta ward of the Tandahimba district, Mtwara and according to RT-PCR, PPRV protein was detected in 33.0% of the tissue samples taken from 28 goats here.

The chronology and probable movement of PPR in Tanzania is depicted in Figure 1.

FIGURE 1: Chronology and spread of peste des petits ruminants in Tanzania.

Two necropsied goats showed inflammation of the apical lung lobes, with multiple abscesses evident in one of the goats. The mesenteric lymph nodes in both carcasses were swollen and the small intestines were hyperaemic. Microscopical lesions in the lungs were associated with bronchopneumonia and focal necrotic areas surrounded by inflammatory zones were also found.

As expected, the 2008 Tanzania strain of PPRV fell into lineage III, as is the case for East African strains, and it was closely related to the historical strains Ibri and Dorcas isolated since 1987 in the Sultanate of Oman (Furley et al. 1987). The lineage is distinct from the one circulating in Asia from south India to Turkey (Özkul et al. 2002).

Figure 2 shows the Bayesian inference of PPRV strains. The phylogenetic reconstruction was made using MRBAYES software as mentioned earlier (Huelsenbeck & Ronquist 2001). The sequence evolution model proposed by Jobb (2011) was HKY. Consequently, the number of substitution types was set at six and the Bayesian reconstruction was run for 6 000 000 iterations. Results were analysed when the standard deviation between the Monte Carlo Markov chains fell below 0.01. Trees were sampled every 1000 generations; thus the consensus tree was built up with a burn-in phase of 25% and posterior probabilities were displayed as node labels. The tree was rooted by two out-groups, namely measles and rinderpest morbilliviruses collected from Genbank (FJ416068 and X98291 respectively), and the four PPRV lineages are highlighted in black brackets. Tanzania_2009 analysed in this study is underlined in lineage III.

FIGURE 2: Phylogenetic tree based on the nucleoprotein gene of the peste des petits ruminants virus, positioning the Tanzania strain in lineage III.

As shown in Figure 3 the epidemic curve resembles a typical propagated outbreak, with peaks of primary, secondary and tertiary generations of cases spaced by an incubation period. The most probable period of exposure of the susceptible animals to PPRV was calculated as 3–5 days before the index case was reported.

It was of epidemiological interest to observe that clinical signs were not present in 68% of the 483 goats and 73% of the 286 sheep that tested seropositive for antibodies against PPRV. The presence of these ‘silent spreaders’, coupled with uncontrolled extensive livestock movement within the country and across the international borders, may explain the detection of antibodies against PPRV without clinical disease in the south-western parts of the country (Tanzania–Zambia and Tanzania–Malawi borders), as shown in Figure 1.

FIGURE 3: Epidemic curve of peste des petits ruminants outbreak in the Newala district of the Mtwara region in southern Tanzania.

Pastoral districts appeared generally to have suffered more from infection (Z = 15.38; p ≤ 0.0001) than the agro-pastoral districts (Table 2). Reasons for this may be related to extensive livestock movements and exchanges resulting in a high rate of contact between animals, which increases the potential for transmission of PPRV.

Pastoral animals will often travel long distances during the dry season in search of fodder and water. Consequently, a considerable proportion of the population becomes infected during this period, giving rise to the establishment of disease endemicity and continued, year-round circulation of the virus. Farmers are often not aware of the pitfalls of mixing sick and healthy animal for grazing. For example, the study team saw diseased small ruminants and in-contact animals being sold at the market despite the ongoing outbreak. The purchase of new stock from the market and the return of unsold animals will often precipitate in severe outbreaks. The movement of animals, therefore, plays an important role in the spread and maintenance of PPRV in flocks, as does the purchase of potentially infected animals and their subsequent introduction into naïve flocks.

Age appears to be a risk indicator (p ≤ 0.001) for seropositive PPR status, as suggested by the observed age profile according to which about 43% of the flock consists of young animals. Although goats are considered to be more susceptible to PPRV than sheep (Abubakar et al. 2011), there was no association (p = 1.000) between animal species and seropositivity in the current study. This could be due to the fact that sheep and goats are equally exposed to the PPR risk factors such as movement and nutritional stresses, or it could be that the circulating field strain has overcome the natural resistance of sheep; different types of breed might affect the outcome of PPRV infection. The significant relationship (p = ≤ 0.001) between sex and seropositive PPR status could be due to the uneven distribution of the two sexes in the study flocks.

The origin of PPRV in Tanzania is not well established, but it is believed that it might have been introduced to Tanzania through the movement of live, infected animals. This hypothesis is supported by the fact that the border districts such as Longido and Ngorongoro had the highest prevalence (Table 1) of PPRV antibodies; moreover, PPRV was isolated from tissue samples that originated from the Ngorongoro district. The results of molecular epidemiological techniques suggest cross-border transmission of actively circulating PPRV into Tanzania from neighbouring northern countries. The virus we isolated is PPRV lineage III, which includes viruses whose origins are in East Africa. An outbreak in Ethiopia in 1996 as well as one in the Middle East was of this type (Abubakar et al. 2011). Thus, one of the northern neighbouring countries in the East Africa region is most likely the source of infection.

The evolving East African situation regarding PPR for the period 2004–2009 can be traced in the World Organisation for Animal Health’s WAHID database ( http://www.oie.int/wahis/public.php?page=disease_timelines) (OIE 2009), where the different waves of incursion of PPR can be seen. The first is observed in Kenya in March 2006, a second in Uganda in 2007 and the last one in Tanzania in December 2008. The majority of north-eastern countries such as Ethiopia and Sudan have been reporting the disease for more than 10 years (Lefèvre et al. 1991). For PPRV transmission, infected and healthy animals need to be in contact (Lefèvre et al. 1991), which serves as an additional indication that the virus came into Tanzania through progressive introduction of live, infected animals from these countries. Trade, as well as the nomadic lifestyle of the people in this region, is to a great extent responsible for the phenomenon. Moreover, the general extension of PPR emergence in Africa is certainly linked to the expansion of trade in animals and subsequent movement of these animals into previously disease-free regions. This phenomenon is further advanced by the presence of a significant proportion of ‘silent spreaders’.

The epidemic curve revealed that during the outbreak cases occurred over multiple incubation periods. The shape and other features of an epidemic curve (Figure 3) can suggest hypotheses about the time and source of exposure, the mode of transmission, the causative agent, incubation period and the efficacy of control measures (Dicker & Gathany 1992). The outbreak had a successive series of peaks, reflecting increasing and decreasing numbers of cases in each generation. The epidemic waned after five generations, probably because either the number of susceptible animals fell below some critical level or the zoosanitary measures taken to control the disease became effective.

The estimates of R_t for PPRV infection (Table 3a) indicate that, on average, four secondary PPR infectious cases resulted from each primary infection in pastoral flocks. The variation in R_t estimates observed (Table 3a) is probably due to farming system and environmental variation, but it may also reflect the stochastic nature of PPRV. At this estimated R_t value (4), it might be expected that minor outbreaks of PPR infection would occur in which far fewer individuals seroconvert. This expectation is consistent with the number of reported sporadic PPR episodes. However, the generally poor reporting level suggests that an R_t value of 4 may be an underestimate of PPRV infection in the pastoral population. Furthermore, the morbilliviruses are known to be highly contagious, with a basic R₀ between 15 and 20 (Anderson & May 1991), which increases the chances of an infection spreading or persisting within the population. However, given that the serological survey was conducted when the epidemic was still evolving, the host–pathogen relationship may not have reached equilibrium. Furthermore, it is possible that the assumption that animals were seronegative during the previous 12 months might not be valid, as the epidemic had been evolving for some time before the animals were sampled. The value of R_t derived in this study should therefore be cautiously considered a minimum estimate of the true population values.

There are limitations to generalisation of this estimated R_t for PPRV infections in other sheep and goat populations. In the current study, the animals were assumed to constitute a free-mixing population, approximating a homogeneously mixed flock. Whilst representative of the small ruminant farming system found in Tanzania, this production environment is quite different from that found in highland areas, where grazing land is scarce. For example, the effective contact rate is likely to differ between tethered and free-ranging animals, which would affect estimates of R_t. Other factors, such as PPRV strain, genetics of the flock, pre-incursion flock immunity, climate, population density and frequency of contact between flocks, may also affect R_t. Thus, R_t for PPRV may vary from the present estimates in other sheep and goat populations and environments. Further refinement of the estimates, which should include variables such as seasonality, environmental contamination, and contribution of the wild life population, as well as estimation of R_t in other sheep and goat production settings will be required to determine the general applicability of these estimates.

An R_t value of 4 predicts that 63.8% – 84.6% of small ruminants in a population would have to be vaccinated with a 100.0% efficient vaccine to reduce R_t to less than 1 (Table 3a). If, as stated earlier, R_t is underestimated, the percentage vaccine coverage would also be underestimated and would need to be increased for effective control of PPRV. For an imperfect vaccine or other interventions that are unlikely to be completely efficient (e.g. quarantine, movement restriction and closure of primary and secondary livestock auctions), effective coverage would probably have to approach 100.0%. The situation is further complicated by the possibility of undetected disease in areas where stock populations are sparse, the possibility that the disease may establish in the wildlife population and the possibility of delayed detection and recognition, which would provide a potential source for re-infection of sheep and goats. Thus, whilst some interventions may reduce the number of infected sheep and goats, eliminating PPRV from pastoral flocks will require significant effort and development of highly effective intervention tools, as for rinderpest campaigns.

The control of transnational livestock movements across borders is difficult, especially in the northern districts where Maasai pastoralists are found on either side of the Tanzania–Kenya border. The resulting transborder movements of cattle, goats and sheep for breeding, trade and other socio-economic functions could be the main predictor for introduction, transmission and maintenance of PPRV in the country. Without proper control mechanisms, these animal movements could contribute to and accelerate the spread of the virus regionally. Indeed, the PPRV-positive samples from the Tanzania–Mozambique, Tanzania–Malawi and Tanzania–Zambia borders (Figure 1) indicate a southward movement of the disease. Therefore, the importance of PPR as a threat to small ruminants should be considered together with other economically important diseases and measures should be taken to prevent subsequent spread of such diseases into the region of the Southern African Development Community (SADC).

Anderson, R.M. & May, R.M., 1991, Infectious Diseases of Humans: Dynamics and Control, Oxford University Press, London.

Dicker, R. & Gathany, N.C., 1992, Principles of Epidemiology, 2nd edn., U.S. Department of Health and Human Services, Public Health Service, Centre for Disease Control and Prevention (CDC), Atlanta.

Dietz, K., 1993, ‘The estimation of the basic reproduction number for infectious diseases’, Statistical Methods in Medical Research 2, 23–41.

Dohoo, I., Martin, W. & Stryhn, H., 2003, Veterinary Epidemiologic Research, Atlantic Veterinary College, Charlottetown.

Food and Agriculture Organization/International Atomic Energy Agency (FAO/IAEA), 1993, ‘Peste des Petits Ruminants ELISA kit. Competitive enzyme immunoassay for detection of antibody to PPR virus. Bench protocol’, version PPR 1.0, January 1993. Joint FAO/IEAE Programme, Animal Production and Health. Pirbright, United Kingdom: World Reference Laboratory for Rinderpest.

Furley, C.W., Taylor, W.P. & Obi, T.U., 1987, ‘An outbreak of peste des petits ruminants in a zoological collection’, Veterinary Record 121, 443–447.

Heffernan, J.M. & Wahl, L.M., 2005, ‘Monte Carlo estimates of natural variation in HIV infection’, Journal of Theoretical Biology 236, 137–153.

Huelsenbeck, J.P. & Ronquist, F., 2001, ‘MRBAYES: Bayesian inference of phylogenetic trees ’, Bioinformatics 17, 754–755. ( http://mrbayes.sourceforge.net/links.php)

Jobb, G., 2011, TREEFINDER, computer software, viewed 17 August 2011 from http://www..treefinder.de

Kwiatek, O., Minet, C., Grillet, C., Hurard, C., Carlsson, E. & Karimov, B., 2007, ‘Peste des petits ruminants (PPR) outbreak in Tajikistan’, Journal of Comparative Pathology 136(2–3),111–119.

Lefèvre, P.C., Diallo, A., Schenkel, S., Hussein, S. & Staak, G., 1991, ‘Serological evidence of peste des petits ruminants in Jordan’, Veterinary Record 128, 110.

Libeau, G., Diallo, A., Calvez, D. & Lefèvre, P.C., 1995, ‘A competitive ELISA using anti-N monoclonal antibodies for specific detection of rinderpest in cattle and small ruminants’, Veterinary Microbiology 31, 147–160.

Office International des Epizooties (OIE), 2000, OIE manual of standards for diagnostic tests and vaccines. List A and B diseases of mammals, birds and bees, OIE, Paris.

Özkul, A., Akca, Y., Alkan, F., Barrett, T., Karaoglu., T., Dagalp, S.B. et al., 2002, ‘Prevalence, distribution, and host range of Peste des petits ruminants virus, Turkey’, Emerging Infectious Diseases 8, 708–712.

Scott, G.R., 1990, ‘Rinderpest and peste des petits ruminants’, in E.P.J. Gibbs (ed.), Virus Diseases of Food Animals, vol. 2, Disease Monographs, Academic Press, New York.

Snedecor, G.W. & Cochran, W.G., 1989, Statistical Methods, 8th edn., Iowa State University Press, Ames, Iowa.

Swai, E.S., Kapaga, A., Kivaria, F., Tinuga, D., Joshua, G. & Sanka, P., 2009, ‘Prevalence and distribution of Peste des petits ruminants virus antibodies in various districts of Tanzania’, Veterinary Research Communications 33, 927–936. http://dx/doi.org/10.1007/s11259-009-9311-7

Wambura, P., 2000, ‘Serological evidence of the absence of peste des petits ruminants in Tanzania’, Veterinary Record 146, 473–474.