The prevalence of serum antibodies to Ehrlichia ruminantium infection in ranch cattle in Tanzania : a cross-sectional study

INTRODUCTION Heartwater (cowdriosis), caused by the rickettsial organism Ehrlichia ruminantium (formely Cowdria ruminantium), is an infectious disease of ruminants. The organism is transmitted by 3 host ticks of the genus Amblyomma. Heartwater is one of the most important tick-borne diseases (TBD) affecting both wild and domesticated grazing ruminants in Tanzania, and it has been estimated that the direct economic loss due to the disease is about US$20.5–24.5 million. The mortality rate due to heartwater is around 15 % with a morbidity rate of 5 %, although the severity of the disease varies widely between cattle types, agro-ecological zones, socio-economic conditions and cattle production systems. In tropical and subtropical areas, the disease is endemic and results in considerable economic losses due to loss of production, treatment costs and reduced initiatives for the upgrading of local breeds of livestock with more susceptible exotic breeds . Heartwater can be unequivocally diagnosed by the identification of E. ruminantium by demonstration of rickettsial inclusion bodies in the endothelial cells of brain crush smears. This method is, however, only applicable to clinical cases. Serological tests specifically targeting bovines have been applied but are of limited use under field conditions owing to cross-reactions with other Ehrlichia species.. The MAP1-B ELISA based on the truncated major antigenic protein 1 of E. ruminantium has been shown to be of value under field and experimental conditions, especially in the regions where the distribution of cowdriosis is unknown. Although the disease is known to be endemic in various parts of Tanzania, there is very little information on the epidemiology of the disease between and across farming systems in the country. This investigation was designed to provide information on the risk of heartwater and to assist in the development of appropriate control strategies. The prevalence of E. ruminantium was assessed, and factors associated with infection in extensively raised indigenous cattle on Manyara Ranch, Tanzania, were explored.


INTRODUCTION
Heartwater (cowdriosis), caused by the rickettsial organism Ehrlichia ruminantium (formely Cowdria ruminantium), is an infectious disease of ruminants.The organism is transmitted by 3 host ticks of the genus Amblyomma 33 .Heartwater is one of the most important tick-borne diseases (TBD) affecting both wild and domesticated grazing ruminants in Tanzania 18 , and it has been estimated that the direct economic loss due to the disease is about US$20.5-24.5 million 10 .The mortality rate due to heartwater is around 15 % with a morbidity rate of 5 %, although the severity of the disease varies widely between cattle types, agro-ecological zones, socio-economic conditions and cattle production systems 10 .In tropical and subtropical areas, the disease is endemic and results in considerable economic losses due to loss of production, treatment costs and reduced initiatives for the upgrading of local breeds of livestock with more susceptible exotic breeds 16 .
Heartwater can be unequivocally diagnosed by the identification of E. ruminantium by demonstration of rickettsial inclusion bodies in the endothelial cells of brain crush smears 24 .This method is, however, only applicable to clinical cases.Serological tests specifically targeting bovines have been applied but are of limited use under field conditions owing to cross-reactions with other Ehrlichia species. 6,9.The MAP1-B ELISA based on the truncated major antigenic protein 1 of E. ruminantium has been shown to be of value under field and experimental conditions, especially in the regions where the distribution of cowdriosis is unknown 14,15,17,33 .
Although the disease is known to be endemic in various parts of Tanzania, there is very little information on the epidemiology of the disease between and across farming systems in the country.This investigation was designed to provide information on the risk of heartwater and to assist in the development of appropriate control strategies.The prevalence of E. ruminantium was assessed, and factors associated with infection in extensively raised indigenous cattle on Manyara Ranch, Tanzania, were explored.

Study site
Manyara Ranch (03°34'04''S, 36°05' 01"E) occupies 17 871 hectares (~45 000 acres) of land and is located 92 km southwest of Arusha city in the Maasai Steppe Heartland 26 .The ranch is located between 2 major national parks: Tarangire to the south and Lake Manyara to the north.Wildlife and livestock interact freely when grazing.The ranch is semi-arid with annual average rainfall of 600-700 mm and is at an elevation of 1200 m above sea level.The vegetation is mainly Acacia-Commiphora bushland, wooded savanna, with short grass.The rains are usually concentrated in 2 seasons: end of March to May, and end of October to December.The mean temperature ranges from 15 to 30 °C.The study was carried out in October and November 2006.

Cattle husbandry system
The cattle are maintained in a pastoral open grazing system from approximately 07:30 to 16:30 daily.No supplementary feeding is provided.The animals are housed at night in bomas or kraals constructed from thorny tree branches to protect them from theft and predators.
The breeding system used on the ranch is natural mating, with service bulls running freely with females year-round.Calves are weaned at 8 months of age.Young calves are isolated and penned until the return of their dams from grazing, and then allowed to suckle.Decatix ® (2.5 % m/v, Deltamethrin, Coopers, Zimbabwe).The frequency of dipping is coded as follows: none: no dipping, low: once every 4 weeks; moderate: twice every 4 weeks.Anthelmintic treatment using Albendazole (10 %, Kela N.V., Phenix, Belgium) is generally limited to calves.

Study animals and sampling
A systematic sampling procedure was adopted and every 4th animal was selected as the cattle passed through a crush 2 .A total of 360 cattle (all ages and sexes), out of the total ranch herd size of 1481, were used in the study.The cattle types reared on the ranch are Boran (75 %) and Tanzania short horn zebu (TSHZ) (20 %), and their crosses (5 %).Each selected animal was subjected to a thorough, whole-body inspection for the presence of ticks.Other information recorded included: sex, age (estimated by ear notch, examination of teeth and written records).The age of each animal was transformed age centred to normalise the data.Tick infestation, with both mature and immature stages, was coded as low: ≤20; moderate: >20 and ≤40; high: >40.Tick infestation was moderate, therefore whole body counting was adopted 8 .Ticks on the study animals were identified to genera and species and counted using standard procedures 8 .The body condition score of each sampled animal was determined using the 9-point system developed at the International Livestock Centre for Africa 20 .Scores were categorised as ≥1-3: thin; ≥4-6 medium; and ≥7-9: fat.The responses to many of these questions were investigated as explanatory variables in the analyses of seroconversion to E. ruminantium.

Collection of sera and laboratory analysis
Blood samples for sera were collected aseptically by jugular venipuncture into a plain vacutainer tube (Becto-Dickson, UK) from selected animals.The blood samples were labelled and transported in refrigerated cool boxes to the Veterinary Investigation Centre (VIC) laboratory where they remained overnight.Serum was then separated by centrifugation and stored at -20 °C until used.The sera were analysed for antibodies against E. ruminantium using indirect MAP1-B ELISA 15,34 .The results were expressed as per cent positive (PP) calculated as a percentage of the optical density (OD) value of the reference positive control.The cut-off point was obtained by plotting a graph of the OD readings of heartwater-negative sera.Seroconversion status was considered positive if the PP cut-off was ≥20 % 33

Statistical analysis
Descriptive statistics for the animal and ranch-level explanatory variables (inclusive of tick infestation) examined in the study were developed using Epi-Info version 6.04d 7 .Cross-correlations 7 were performed on all explanatory variables to investigate the potential for confounding amongst them (defined as those with a coefficient of 0.7 or greater).The relationship between explanatory variables and outcome response (seroconversion to E. ruminantium) were investigated in 2 steps by logistic regression (using Egret for Windows 4 version 2.0).Explanatory variables investigated were 'presence of a tick of each species' investigated as a binary variable (i.e.yes or no); body score (coded as fat, medium, thin); dipping (yes or no), acaricide application frequency (coded as none, low, moderate); sex (female or male) and age.
In the 1st step, the relationships between each explanatory and outcome variable were individually investigated.In the 2nd step, any variables that were significantly associated at P <0.25 were included in multivariate logistic regression 4 models, producing, by forwards and backwards substitution and elimination, the most parsimonious models in which all explanatory variables remained significant at the P < 0.05 level.The criteria for inclusion and exclusion were a change in deviance significant at the 5 % level according to the maximum likelihood ratio test-chi square distribution.
Forces of infection were estimated from age seroprevalence profiles using maximum likelihood methods (MLM) in Excel (Microsoft, USA) with the Solver add-in 29 .Assuming a stable population size and age structure and a constant force of infection across all age groups, the log likelihood (LL) was derived using the following equation: where Ri = number of seropositive in group i, Ni = number tested in age group i and λ = the force of infection.

Descriptive analysis of antibody response to E. ruminantium
Of the animals sampled (n = 360), 249 (69.2 %) were females and 111 (30.8 %) were males.The mean serum antibody prevalences of the 360 animals studied by variable categories are shown in Table 1.The overall mean antibody prevalence was 50.3 % (95 % CI, 44.9 -55.5).

Tick infestation
A total of 2600 ticks (mature and immature) were counted during sampling.Of the 360 cattle examined and sampled, 320 were found to carry ticks, giving an infestation rate of 88.9 % and the overall mean (mean± SE) tick density of 8.12 ± 0.44 tick/cattle.Tick infestation rate was significantly (χ 2 = 31.7,df = 4, P = 0.001) higher in low, intensive and non-dipped and least in moderately dipped cattle.The mean tick-specific densities per animal are detailed in Table 2.

Age-antibody prevalence profiles
An increasing seropositivity was recorded with respect to increasing age.The estimated force of infection was at 0.18 animals per cattle year-risk.The graphical age seropositivity relationship and forces of infection are shown in Fig. 1.

Factor influencing antibody response to E. ruminantium
The factors that were significantly associated with E. ruminantium antibody prevalence in the multivariate regression model are given in Table 3. Cattle treated with acaricide at an interval of once per month and not dipped were associated with significantly lower antibody prevalence than those that were treated at an interval of twice per month (OR = 0.31 for once per month, OR = 0.25 for not dipped, P < 0.001).Animals that carried ticks (of any species) at the time of survey were associated with significantly higher seropositivity than those that did not carry ticks (OR = 3.30, P < 0.001).Multivariate analysis of risk factors showed that age was the risk factor for the occurrence of seropositivity to heartwater in cattle (β = 0.19 per year of age, P < 0.001).

DISCUSSION
Based on a serological survey, the present investigation revealed that bovine cowdriosis due to E. ruminantium infection exists on Manyara Ranch, supporting the presence of heartwater (cowdriosis) (A Chang'a, pers.obs., 2006) and in other parts of Tanzania 10,13 .The detected prevalence of infection in the cattle was intermediate in comparison to that observed in other studies in Africa; from 31 % 11 in the Ivory Coast, 41 % in Malawi 25 , and up to 61 % in Ghana 12 , but generally higher than observed in studies in Caribbean, e.g. from 7.3 % reported in indigenous cattle on 19 islands 19 to 30 % in cattle in Guadeloupe 3 .The estimated seroprevalence was considerably less than 70 %, suggesting that this pathogen exists in a state of 'endemic instability' as defined 5,21 .However, these levels are sufficiently high to ensure that clinical disease would be a risk.
The most commonly detected tick species was A. variegatum, which was consistent with serological responses to E. ruminantium (transmitted by A. variegatum), having the highest prevalence amongst the cattle 13,25 .R. evertsi evertsi constituted 13.9 % and was the 2nd-most abundant tick.However, this result is not consistent with findings in Ethiopia, where it formed only 6.4 % of the total ticks counted in a survey 28 .The ability of this tick to survive in open grassland, comparable to Manyara Ranch, and its perennial breeding habit, have both been advanced as the reasons for high incidences.Tick burdens were low (mean 8.2, range 4-64), but with no adult ticks of any species being observed on cattle in 11 % of observations.Such low tick burdens could reflect the poor off-host survival of ticks in cattle ranchland.
The present study indicated that animals that were not dipped, and those dipped at an interval of once every 4 weeks, were all associated with lowered seropositivity.It is not clear, however, why the detected rate of infection was low, particularly for undipped cattle, considering the number and level of tick infestation recorded during the survey.One possible explanation could be that the current acaricide (at least at ranch level) regimen (once every 4 weeks, and twice every 4 weeks) is more frequent than necessary, although it might not necessarily be true because of other tick species that are potential vectors of other tick-borne diseases.It is not clear, therefore, to what extent these cattle contribute to the maintenance of ticks and the E. ruminantium pathogen on Manyara Ranch, and this warrants further investigation.
Presence of vectors (ticks) was another factor associated with cattle seropositive  to E. ruminantium.Animals that carry ticks (of any species) had higher seroprevalence since they were at a higher risk of exposure than those that had no ticks.Other factors investigated such as body score were not associated with E.ruminantium infection.
Age-related differences in prevalence and force of infection were significant.A strong positive correlation was observed with increasing age and reactor rates, reflecting more innate resistance to primary infection and a high level of exposure to infective ticks.This is in agreement with the findings of other authors 32 .High antibody response to the E. ruminantium parasite in animals above 6 months of age (the youngest animals sampled) demonstrated a high level of and early exposure to infective ticks consistent with a high degree of infection.Studies have shown that maternally derived antibodies decline to zero within 2-4 months 1 .In the light of our findings, the detected antibody titres could be due to field exposure to infective ticks rather than maternally derived antibodies.As far as is known, immunisation against E. ruminantium has never been carried out in Tanzania, and any seropositive cattle are most likely to have been naturally infected.Nonetheless, this is the first structured study to have identified and quantified the determinants of heartwater in an extensively grazed local herd in rural Tanzania.
Although MAP-1B ELISA is rated as more specific and sensitive to ovine and caprine than bovine sera compared to polyclonal competitive ELISA (PC-ELISA) 27 , our seroconversion estimates are likely to be lower or higher than the true prevalence in the cattle.A wider use of MAP-1B-ELISA in cattle under field conditions is hampered by the down-regulation of antibody responses to E. ruminantium following 1st exposure to the organism, and also by cross-reactions with other Ehrlichia secies such as E. canis and E. phagocytophilum 22,23 .However, consistent with the results of Zimbabwean study 14 , the findings of this study did provide evidence of previous exposure to the pathogen and suggest that there is a need for strict tick-control measures.The serological evidence of cattle exposure to E. ruminantium infection highlights the need for taking cowdriosis into consideration in the envisaged future crossbreeding scheme on Manyara Ranch, where over 70 % of the ranch herd is of the Boran breed.This breed has been shown to have a higher growth rate and reproductive performance, and better adaptability to stressful nutritional, parasitic and disease conditions, than the exotic (Bos taurus) breed 30 .These potentials have made Boran cattle a standard breed in crossbreeding for both beef and milk production in low-input pasture-based systems 31

CONCLUSIONS
The use of serum antibody profiles in the detection of exposure of cattle to the E. ruminantium parasite has an important diagnostic role.Despite the fact that animals seem well adapted to ranch environments, the use of antibody profiles alone is not adequate to explain disease status, as a number of animals developed antibodies without obvious clinical disease.Consistent with serology, other prospective morbidity and mortalitity studies should be carried out to arrive at a better-informed indication of the disease status in this complex livestock/wildlife interaction system.0038-2809 Tydskr.S.Afr.vet.Ver.( 2008) 79(2): 71-75

Table 1 :
List of variables, proportion and seroprevalence of each category investigated (n = 360) (CI = confidence interval).

Table 3 : Significant factors associated with seropositivity to Ehrlichia ruminantium of cattle on Manyara Ranch using multivariable logistic models
(β = coefficient of regression (or parameter estimate), SE = standard error of coefficient, LRS = likelihood ratio statistic, LRP = likelihood ratio P-value)