The Gambia is a small, sub-Saharan country located on the West Coast of Africa. A short, tropical rainy season begins in July and ends in September and a long dry season extends from October to June. Animal traction plays a significant role in the intensification of crop production and, in addition, equidae are used for transportation of goods, harvests and people over long distances. The 1974 National Agricultural Sample Survey estimated a population of 10,500 donkeys and 5000 horses. There has been a > 400% increase in the number of equidae over the 28 year period to 2002, when the census recorded 22,236 horses and 42,268 donkeys (Dr Kebba Daffeh, pers comm).

The pathogenic species of trypanosome found in The Gambia include Trypanosoma congolense, Trypanosoma vivax and Trypanosoma brucei, which are transmitted cyclically by tsetse flies (Glossina spp), although T. vivax can also be transmitted mechanically by biting flies [1]. The prevalence of trypanosomosis reported in Gambian equidae has primarily been determined using classical parasitological methods [2–4]. In general, horses are considered to be highly susceptible to infection, while donkeys are considered to be more resistant, although the reasons for lower prevalence in donkeys could also be due to feeding preferences of the tsetse vector. Snow et al. [3] reported that given the susceptibility of equidae and reported foaling rates, the numbers of animals in The Gambia must be maintained by import from areas of lower challenge, while other authors suggest that the level of tsetse challenge has reduced as a result of changes in the habitat. The data on prevalence do not seem to support these views as the surveys undertaken in 1989–91 show a prevalence of 6–10% in horses and 9.2% in donkeys [2, 3], which does not seem to be consistent with a very high challenge, whereas studies undertaken in the mid to late 1990s show an increased prevalence with values ranging from 43–63% in horses and 6.2–43% in donkeys [4, 5], suggesting that the challenge has increased. As all these studies used the buffy coat method of diagnosis [6] the differences cannot be methodological. However, one of the most recent studies [4] was undertaken on animals attending a veterinary clinic, which could, in part, explain the higher prevalence. In each of the studies reported to date, the species of infecting trypanosome were defined; with a general conclusion that T. congolense is the most prevalent followed by T. vivax with T. brucei being rare, although the prevalence of T. vivax was shown to be higher in the dry season. There has been a significant amount of work carried out developing more sensitive PCR based methods for detecting the trypanosome species infecting livestock. Whereas the buffy coat technique has a threshold of detection of around 1 × 10² parasites/ml [6], PCRs have been developed that target multi-copy targets, such as the satellite repeats on the minichromosomes [7] or the Internal Transcribed Spacer (ITS) of ribosomal DNA [8], the sensitivity of which are simply limited by the presence of trypanosome DNA in the sample i.e. a single trypanosome. For a comprehensive review of these techniques, see [9]. Furthermore, in a recent study from The Gambia, it was shown that the number of detected infections was seven times higher using PCR [5].

The aims of this study were to estimate the prevalence of trypanosomes in both clinically presented and asymptomatic horses and donkeys in both the wet and dry season, as well as determining which species of trypanosomes might be responsible. The development of highly sensitive, species specific PCR based methods of diagnosis [9] provides the possibility of more comprehensive data on prevalence. This allowed the testing of previous conclusions based on less sensitive methodology, as well as the investigation of interactions between species in the development of clinical disease. The results provided a series of important findings that challenge some of the conclusions of previous studies, and provide justification for more extensive future investigation into the epidemiology in order to inform control and treatment strategies.

Although the prevalence of all Trypanosoma species was slightly higher in the wet season (92%) compared to the dry season (90%) this difference was not significant (P = 0.7) for any of the species or for mixed infections (Table 2). The prevalence in the 73 samples from asymptomatic animals sampled in surrounding villages (87%) was not significantly different to that in animals presented to GHDT clinics (92%). There was a higher overall prevalence in horses (93%) compared to donkeys (83%) (Table 2) and horses had a significantly higher prevalence of T. vivax infection. However, significantly more donkeys had T. brucei sp infections and donkeys had a greater proportion of mixed infections compared to horses.

The presenting clinical signs of infected animals were variable ranging from lethargy and anorexia reported by the owner, with no abnormalities detected on clinical examination, to marked pyrexia, tachycardia and tachypnoea. Dependant oedema was an occasional finding in 27 positive and 3 negative animals (diagnosed by PCR). The majority of animals had a body condition score (BCS) of 2 or lower. However there was no significant difference in BCS, total protein or other recorded clinical parameters between animals with and without trypanosome infection diagnosed by PCR. However, 91% of animals were positive for trypanosome infection and a study with larger numbers of uninfected animals is required to evaluate clinical signs associated with infection.

Model diagnostics showed that the assumptions of normality were met and removal of observations with a large influence showed the model to be stable.

This study found the prevalence of trypanosomosis in horses and donkeys in the Central River Division of The Gambia by microscopy was 18.25%, a rate of detection similar to that seen in other studies by microscopy in this region (7.5 – 34.2%[5]). Although speciation was not carried out under microscopy, accurate speciation was undertaken using species specific primers [7]. This allowed the detection of mixed infections much more accurately, as it is not dependent on a patent parasitaemia. The prevalence as determined by species specific PCR after Whole Genome Amplification (WGA) on the samples was 93% and 83% in horses and donkeys, respectively. These values are considerably higher than any of the previously published figures, which range from 7% [3] to 61% [4], depending on the region of the country sampled, the population sampled and the time of year. However, the published data are largely based on microscopy, which from this study and that of Faye et al. [5] exhibits a much lower sensitivity. The sensitivity of PCR based methods has been reviewed [9] and these methods can detect as few as 1–20 trypanosomes/ml, although the exact sensitivity depends in part on whether prior concentration of trypanosomes is undertaken before PCR amplification. We have previously estimated the sensitivity of whole genome amplification followed by PCR, the technique used in this study and, using the satellite PCR, can detect 100 trypanosomes/ml from blood spotted on FTA filters [11]. Similar to the work by Dhollander et al. [4], the majority of animals sampled in this study were those presented for clinical problems to the mobile or home clinic, and therefore, the prevalence estimates are unlikely to be representative of the general population. However, the prevalence in this study for the samples from apparently healthy animals sampled in surrounding villages was also very high (87%). The use of target repetitive sequences in PCR to detect T. congolense and T. vivax has to be treated with an element of caution, since it has been suggested that PCR using these highly repetitive elements can amplify from DNA that remains in circulation for some time after parasite death [5], and does not necessarily indicate active infection. This explanation is unlikely to be the case for the single copy microsatellite used for T. brucei sp detection, as the sensitivity of the PCR is much lower [11]. The minimum conclusion from a positive result using the satellite repeat PCRs is that the animal either is, or has been infected with T. congolense and T. vivax at some point. In every case the possibility of cross-contamination was kept to a minimum, with positive and negative controls included to ensure no contamination, and by processing each sample individually. The main difference between the present study and that of Faye et al. [5], where PCR was also utilised on samples from the same geographic region, is the higher detection of T. vivax (87%) and T. brucei sp (18%) cases. Faye and colleagues demonstrated a much lower prevalence of T. vivax (21%) compared to our results, despite using the same oligonucleotide primers, although it must be pointed out that their sample number was comparatively low (11 horses and 29 donkeys). In our study the use of WGA, which increases overall sensitivity by replicating the genomes present in a particular sample [11], may explain the increase in sensitivity by increasing the detection level of trypanosomes at low parasitaemia above the threshold of PCR, thereby explaining the discrepancy in prevalence between the two studies. In conclusion, the prevalence of both T. vivax and T. brucei sp reported in this study are much higher than previously reported, and the use of WGA followed by PCR is likely to be the explanation for this.

From our study, T. congolense Savannah is associated with a lowered PCV (Table 3), similar to previous studies. For example, Dhollander et al. [4] reported a prevalence rate of 63%, of which 64% were infected with T. congolense and 32% T. vivax. The majority of these animals were also anaemic, and the correlation between an animal being microscopically positive and anaemic has also been demonstrated [2]. Therefore the results of Dhollander and others correlate with our results, in that the microscopically positive cases had more severe anaemia. PCR allows the detection of subpatent parasitaemias, such as those characteristically resulting from T. vivax and T. brucei sp, and therefore results in a much more comprehensive and realistic picture of what is present. However, microscopic positivity is also associated with the acute stage of disease where parasite burden is greater, and it may be that this stage is more acute in T. congolense, resulting in a greater degree of presentation at clinics. Our results raise an important issue however, as they suggest that the association of particular parasite species with particular pathology may only be accurately determined in field studies if the true parasite prevalence's are known. The assumption that T. congolense causes the more severe disease may be correct, but as is indicated by our results there are added complexities, such as the pathogenic influence of T. brucei sp, and perhaps most intriguingly, the ameliorative effect of concurrent T. vivax infection. However, a more comprehensive and controlled study is required to fully understand the epidemiology of the disease, and the roles of different trypanosome species, mixed infections, and different host species in the holistic picture of trypanosomosis.

In the case of T. brucei sp, it has been reported that infection in equidae is severe and often fatal [12], and this has been suggested as a reason for the few T. brucei sp cases observed, due to the rapid death of the host [4]. If this scenario is correct, then our prevalence of T. brucei sp infection of 18% is likely to be an underestimate and the clinical effect may be more severe. Alternatively T. brucei sp is not as virulent in equidae as previously thought, or there are variations in pathogenesis induced by different strains of T. brucei sp. In the equidae examined in our study, infection with T. brucei sp was associated with a significantly lower PCV, although this effect was not as great as with T. congolense infections. In contrast, although the prevalence of T. vivax was very high, it did not appear to be associated with a decreased PCV, which seems to confirm the notion of T. vivax being relatively apathogenic in equines [12–14]. However, experimental infection in horses with subsequent disease has been demonstrated [15], and in cattle T. vivax infections have been reported as both severe, causing pyrexia, and of a more chronic nature with progressive anaemia and weakness [14]. Certainly strain-dependent pathology does occur in T. vivax, with a haemorrhagic strain reported in cattle [16]. Interestingly, the multivariable model indicated that concurrent infection with T. congolense and T. vivax is associated with a reduced expression of pathology (in terms of PCV) relative to infections with T. congolense alone, and this may be due to competition between the trypanosome species. The phenomenon of competition was initially identified in superinfections of T. congolense [17, 18], where primary infections prevented establishment of a second strain. It was also investigated between species [19], where goats already infected with T. congolense delayed the establishment of a T. brucei secondary challenge, but not that of T. vivax. Therefore, one could postulate from our data that prior or co-infection with T. vivax ameliorates the pathology due to T. congolense by a similar interference phenomenon. Obviously, more detailed analysis of populations, and controlled surveys, ideally using naïve young animals, are desirable to analyse whether this phenomenon is occurring, and if it is, what implications this has for the epidemiology of the disease.

Previous studies have demonstrated that trypanosome prevalence is higher in horses then in donkeys [4, 5], although Faye et al. [5] included only a small numbers of horses. To explain these findings it has been suggested that donkeys are inherently resistant to trypanosomes, that there is a feeding preference of tsetse flies for horses, and that donkeys have a better ability to deter flies from feeding by skin rippling, head movements and other behavioural avoidance mechanisms [2]. In the current study there was a slight overall difference in prevalence between horses and donkeys, although T. vivax was more prevalent in horses, and T. brucei and mixed infections were more prevalent in donkeys. These results, assuming that a single species of tsetse fly is involved in transmission, argue against feeding preference or behavioural differences. In the present study donkeys had a small but significantly lower PCV than horses. There are few published reference values for PCV in working horses and donkeys, although Pritchard et al. [20] found mean values for PCV were low, but not significantly different in both (uninfected) working horses and donkeys. This may be due to mal- and under-nutrition, as well as infectious diseases.

In summary, trypanosome infection is highly prevalent in equidae in the Central River Division of The Gambia, and appears to be a clinically significant condition associated with anaemia. There were not any reliable, pathognomic clinical signs to definitively aid diagnosis. The sensitivity of microscopy for diagnosis was poor, and analysis by PCR following Whole Genome Amplification revealed the prevalence of trypanosome infection to be startlingly high and much higher than previously reported. The increase in detection seen in our study is very likely due to the combination of WGA and PCR. The infections were predominated by T. vivax (87%), followed by T. congolense Savannah (31%) and T. brucei sp (18%). There were also a high proportion of mixed infections. PCV was significantly lower in animals infected with T. congolense or T. brucei sp and, interestingly, there appeared to be clinical differences between those animals concurrently infected with T. congolense and T. vivax compared to those infected with T. congolense alone in terms of the effect on PCV. Therefore, the epidemiology of trypanosomosis in the field with interactions between different trypanosome species and strains in different species of host is obviously complex, but the use of tools such as WGA and PCR allow deeper analysis by allowing detection of subpatent infections, and therefore a more comprehensive picture of the dynamics of parasite populations in the host. However, in order to fully analyse the situation in the case of African trypanosomes, a comprehensive and controlled longitudinal study is necessary in order to fully examine the influence of different host species, different trypanosome species and environmental factors. Furthermore, data on these aspects are particularly required in order to qualitatively inform control and treatment strategies.