2.1. Study Design

Whole blood and serum samples were collected during May 2018 to September 2019 from two national parks, KNP and ENP, and the host species targeted in this study included free roaming greater kudu (n = 72; 32 from KNP and 40 from ENP), plains zebra (n = 65; 39 from KNP and 26 from ENP), impala (n = 21 from KNP), and blue wildebeest (n = 30 from ENP) (Figure 1). These samples were originally tested for the presence of antibodies against Bacillus anthracis (causal agent of anthrax) [47]. The sample size was small due to budget constraints as the animals were chemically immobilized and collared to monitor their movement and exposure to B. anthracis in KNP and ENP [48]. In the framework of the present work, the same samples were also screened using serology to detect FMDV, Brucella spp. and C. burnetii, and DNA from blood were analyzed using a molecular reverse line blot (RLB) method to detect Anaplasma, Ehrlichia, Theileria, and Babesia spp.

Each animal was selected randomly from different herds. All animals were adults or subadults as was required for the collaring study. Each sample was assigned a unique identification number. Supporting information data on sampling date and GPS location were recorded.

2.2. Study Area

KNP is situated in the Limpopo and Mpumalanga provinces of South Africa. It is regarded as one of the largest and most important national parks in Africa, hosting a total of 148 wild mammal species, including the big five (i.e., lion, leopard, elephant, rhino, and buffalo), in a 19,485 km² fenced conservation area situated in the FMD infected zone [49]. Population estimates for the selected wildlife species in KNP include 11,200–17,300 greater kudus, 132,300–176,400 impalas, and 23,700–35,300 plains zebras (https://www.sanparks.org).

ENP, also situated in the FMD protected zone [50], is an almost 23,000 km² wildlife reserve located in northern Namibia. ENP is home to 114 mammal species, but it is not considered a big five reserve as African buffaloes are not present in the park [51]. Aerial estimates of selected wildlife include 2822–5592 blue wildebeests, 11,338–17,126 plains zebras [51], and 394–580 greater kudus [52].

2.3. Laboratory Protocols

2.4. Data Analysis and Reporting

Data were analyzed in R programming language (version 4.2.1) using the R studio IDE software (RStudio Team, 2021). To account for our small sample sizes, confidence intervals and hypothesis testing were estimated employing exact/nonparametric methods, and the results were interpreted with great caution. The 95% confidence intervals (CIs) were calculated to measure variability and error of our estimated point prevalences by species. Because of small sample sizes, we opted for the more conservative Clopper–Pearson method [58] using the R function “exactci” from the “PropCIs” package.

To determine which infections were most likely to co-occur in hosts, we used the Spearman’s correlation coefficient (r_s) using function “cor” (with method = “Spearman”) from package “stats” in R. Coefficient (r_s) values from 0 to 0.25 or from 0 to −0.25 indicate the absence of correlation, whereas values from 0.25 to 0.50 or from −0.25 to −0.50 point to poor correlation between variables; values ranging from 0.50 to 0.75 or −0.50 to −0.75 are regarded as moderate to good correlation, and r values from 0.75 to 1 or from −0.75 to −1 indicate very good to excellent correlation between variables [59]. This correlation was considered significant if the t test for Spearman rank correlation indicated a p-value <0.05 under the null hypothesis of no correlation [58]. When performing multiple comparisons, the family-wise error rate increases, hence the probability of finding at least one false positive (type I error) [60]. To yield conservative results, p-values were adjusted using the Bonferroni correction in which the p-values are multiplied by the number of comparisons [61]. This was achieved by applying function “p.adjust” (method “bonferroni”) from package “stats.”

To assess correlation between the prevalence and independent variables (i.e., animal species, sex, and sampling park), we employed the chi-squared test. An alternative when the conditions for a chi-squared test are not met (i.e., no cells with expected values <1, and no more than 20% of cells with values <5), is a Monte Carlo simulation [62] performed with the option “simulate.p.value = TRUE” in the function “chisq.test.” We set the number of replicates in the simulation to B = 2000. Again, p-values were adjusted using the Bonferroni correction, and statistical level was set at α = 0.05.

4.1. Anaplasma/Ehrlichia and Theileria/Babesia Prevalences are Higher in KNP Compared to ENP

As highlighted by the comparison of the 95% CI and the chi-squared statistics, infection prevalences of Anaplasma/Ehrlichia and Theileria/Babesia genera were significantly higher in KNP compared to ENP in both kudu and zebra. This may be due to the relative diversity and abundance of ticks inhabiting the parks. Indeed, the prevalence of tick infestation in ENP wildlife is reportedly well below those reported in other parts of southern Africa [63–65]. Tick distribution and ultimately the survival of pathogens in ticks and animal hosts are, in turn, affected by abiotic factors. Indeed, hot dry conditions and desiccating winds adversely affect the population of questing ticks by imposing mortality on unfed ticks [66]. Moisture-related indices significantly affect the presence of ticks and TBDs, with wetter conditions almost always beneficial [66]. ENP is located in a semiarid region of Namibia characterized by a large salt pan, which may be dry for extended periods of the year, especially during the dry season [67]. On the other hand, KNP is situated in northeastern South Africa and has a more diverse climate with a greater availability of water throughout the year compared to ENP. Overall, ENP is considerably drier than KNP and therefore a less suitable region than KNP for tick proliferation, infestation and transmission of TBDs. For instance, Amblyomma hebraeum, Amblyomma variegatum (vectors of E. ruminantium), Rhipicephalus decoloratus (vector of Babesia bigemina and A. marginale), and Rhipicephalus appendiculatus (vector of Theileria parva and A. bovis) are present mainly or only in KNP, whereas Hyalomma rufipes (vector of Babesia occultans), Hyalomma truncatum (vector of several Anaplasma/Ehrlichia spp.), and Rhipicephalus evertsi (vector of T. equi and B.caballi) are found in both parks [68–70].

4.2. High Prevalence and Coinfection of Theileria spp. in Kudu and Impala From KNP

In the present study, we report extremely high prevalence of T. buffeli and T. bicornis in 27/28 kudus (96%; CI: 82%–100%) and 19/19 impalas (100%; CI: 82%–100%) from KNP. In addition, in KNP kudu, there was high prevalence (90%–100%) and significantly associated coinfections of pathogens from the genera Theileria, including T. taurotragi, T. buffeli, Theileria sp. (kudu), and Theileria sp. (sable) (Table 1). Theileria spp. (sable) was also detected in 5/19 impalas (26%; CI: 9%–51%) from KNP. None of the 40 kudus from ENP tested positive for any of the tested Theileria species.

Theileria taurotragi and T. buffeli are “schizont non transforming” Theileria spp. and therefore classified as benign parasites, with rare clinical signs that mainly occur due to piroplasm-induced acute hemolytic anemia [71]. Indeed, T. taurotragi caused bovine cerebral theileriosis in young African shorthorn cattle [71] and theileriosis in eland (Tragelaphus oryx) [71]. Theileria sp. (sable) and Theileria sp. (kudu) [56] are regarded as pathogenic species in African wild artiodactyls. Mortalities in roan antelope (Hippotragus equinus) due to Theileria sp. (Sable) have been reported after translocation [56]. Infection with Theileria sp. (sable) negatively affects attempts to establish breeding herds and reintroduction efforts into the wild due to calf mortalities [72]. Theileria bicornis has not been found to cause mortality but has been reported in free-ranging white and black rhinoceroses in South Africa and Kenya [55, 73, 74], as well as from apparently healthy nyalas (Tragelaphus angasii) [75], impalas, eland (Taurotragus oryx), and sable antelope (Hippotragus niger) in South Africa [76]. The very high T. bicornis prevalences obtained in this study in kudu and impala from KNP (Table 1) might raise concerns for the rhino populations as they are already suffering from poaching and stress induced by unavoidable translocations [77, 78].

Further studies may assist in determining the health effects of the above-mentioned Theileria infections in wildlife species. Coinfections may alter virulence of pathogens and subsequent disease outcomes in the hosts [79–81]. As a general rule, coinfections may lead to worse health outcomes for hosts and increase within host pathogen titers, altering transmission ecologies. Nevertheless, the impact on animal fitness due to coinfections between pathogenic and benign Theileria species appears to be intricate. For instance, apathogenic T. mutans and T. velifera seem to protect cattle from the detrimental consequences of T. parva infection [82]. This could also be our case, with the benign T. taurotragi, T. bicornis, and T. buffeli protecting wild antelopes from the adverse effects of pathogenic Theileria sp. (sable) and Theileria sp. (kudu), but this hypothesis needs further investigation. The occurrence and effects of coinfection of multiple pathogen species within wildlife populations remains largely unknown. Indeed, understanding dynamics of coinfection or coexposure to different pathogens is useful in improving our knowledge of pathogen epidemiology in wildlife and in the development of risk models for diseases in various epidemiological contexts.

4.3. Anaplasma centrale in Impala and Zebra From KNP and Wildebeest From ENP

Anaplasma centrale and A. marginale are closely related species that cause bovine anaplasmosis in cattle [83]. A. centrale is known to be less pathogenic than A. marginale in domestic animals as it induces a low degree of anemia, with rare clinical outbreaks [84], but it confers immunity against infection by A. marginale. Nonetheless, a clinical case of bovine anaplasmosis caused by A. centrale was reported in Europe in 2008 [85]. A. centrale seems to be largely subclinical in wildlife [38] where it occurs with moderate prevalences (10%–30%), especially in African buffalo, impala, eland, waterbuck (Kobus ellipsiprymnus), and blue and black wildebeest (Connochaetes gnou) [37–39, 76, 86]. These wild animal species may be able to maintain A. centrale much more efficiently than tick vectors. In fact, although experimental transmission of A. centrale by ticks (e.g., Rhipicephalus simus and Dermacentor andersoni) has been proven [87, 88], secretion of this pathogen into tick saliva occurs at a much lower rate than A. marginale, and hence, transmission is achieved only when tick numbers are dramatically increased to compensate for the low pathogen load [88]. In addition, A. centrale prevalence in ticks is very low in all tick species considered [89], making them an inefficient reservoir for A. centrale. In support of this hypothesis, we report infection with A. centrale in 12 impalas (63%; 12/19) and two zebra (5%; 2/39) from KNP and in two wildebeest (7%; 2/30) from ENP. The occurrence of A. centrale in impala from KNP is not surprising as the pathogen was already reported in the same species and in buffalo, black wildebeest, common eland, and waterbuck from South Africa [37–39, 76, 86], while the occurrence of A. centrale in zebra from KNP and wildebeest from ENP is a new finding that sheds light on the geographic and host range of the pathogen.

4.4. Anaplasma platys in Kudu and Impala From KNP

Anaplasma platys is the etiologic agent of thrombocytic anaplasmosis in dogs and is the only recognized Rickettsiales species known to infect platelets [90]. After the first description, A. platys has been reported worldwide, including the Americas, Eurasia, Africa, and Australia, mainly in tropical and subtropical areas [91–93]. For a long time, A. platys was considered only a canine pathogen, but a wider host tropism for A. platys has been demonstrated in recent decades. Cases of A. platys infection have been reported in cats, goats, cattle, Bactrian camels (Camelus bactrianus), red deer (Cervus elaphus), sika deer (Cervus nippon), and sable antelope [94–101]. Occurrences in atypical hosts have been attributed to A. platys-like bacteria [102, 103]. However, A. platys-like species cannot be distinguished from A. platys based on 16S rRNA as they are very closely related. These A. platys-like species in atypical hosts are considered the probable cause of human infections [104], with clinical signs varying from chronic and nonspecific, including headaches and muscle pains [105] to migraines and seizures due to mixed A. platys, Bartonella henselae, and “Candidatus Mycoplasma haematoparvum” infection [106].

Rhipicephalus sanguineus is considered the primary vector for A. platys [98, 107, 108] which rarely infests impala and kudu. The agent has also been detected in Haemaphysalis longicornis and Ixodes persulcatus in Korea, Rhipicephalus turanicus in Israel, and Rhipicephalus spp. in China [98, 109–111].

Here, we found three kudus (11%; 3/28) and one impala (5%; 1/19) positive to A. platys by means of RLB hybridization. Given the limited information available on A. platys infections in Africa, it is of particular interest to understand the sylvatic cycle of A. platys in kudu and impala and which tick vector (if any) is involved in pathogen transmission.

4.5. Babesia occultans in Wildebeest From ENP

Babesia occultans is considered less pathogenic than other Babesia species [112]. Observable clinical signs due to infection with B. occultans in cows include anorexia, weakness, fever (≤40°C), anemia, and pale mucous membranes. However, unlike B. bigemina, B. bovis, and B. divergens infections, no jaundice, hemoglobinuria, gastrointestinal disorders, and nervous symptoms have been found in cows infected with B. occultans [113, 114]. In this study, we identified nine B. occultans-positive wildebeest (30%; 9/30). Since its clinical signs are nearly identical to those of piroplasm infections, it is important for local animal health officers and veterinarians to acknowledge the presence of the pathogen and consider it in diagnoses and treatment strategies.

4.6. Ehrlichia ruminantium in KNP Zebra

Reports of E. ruminantium in African nonruminant wildlife are rare and controversial. For instance, E. ruminantium-like colonies were detected in brain endothelial cells of a Nigerian African elephant (Loxodonta africana) that reportedly died of anthrax [115]. This report requires verification due to the unusual nature of the case and the possible presence of pathogens similar to E. ruminantium. Black and white rhinoceroses from Zimbabwe tested serologically positive to E. ruminantium using a MAP1 competitive ELISA [116]. However, this technique is known to cross-react with other Anaplasmataceae [117], and therefore, no confirmation can be drawn from these findings.

In our study, two plains zebra from KNP tested positive to E. ruminantium with RLB. The occurrence of the pathogen in a wild equid could be most likely incidental, but it may still be of epidemiological importance to understand the source of infection and transmission dynamics, for which further molecular characterization of the pathogen may provide significant insights.

4.7. Seropositivity to FMDV in Greater Kudu in KNP

A total of 13 greater kudus (41%; 13/32) from KNP sampled in October 2018, South Africa, were found seropositive to FMD by means of NSPCE. While natural infection with FMD has already been reported in greater kudu from Botswana by means of reverse-transcriptase PCR [118, 119], the present study represents the first report of FMD based on NSPCE in greater kudu in South Africa using serology. This test has not been validated for wildlife. Risk factor analysis (Table 2) indicates that greater kudu has significantly higher prevalence of FMD among the affected animal species investigated. The location (sampling park) was a significant predictor of infection. Antibodies against 3ABC complex of FMDV can be detected in a window of between 1 week to 6 months after exposure to the pathogen [120]. These observations point to circulation of FMD in kudu population from the northern area of KNP that were exposed to the pathogen anytime during April to October 2018. Interestingly, this event might have occurred in proximity and just a few months before the January 2019 outbreak in Vhembe district, Limpopo, South Africa in cattle. Greater kudu has been reported to shed the virus up to 160 days after experimental infection, more than any other African non-buffalo bovid (“antelope”), and clinical signs have been reported from this species without mortality [118, 119, 121]. Nonetheless, the role of kudu in maintaining and spreading FMDV is still to be investigated and clarified. This report underscores the importance of further investigation into the role of kudu in the epidemiology of FMD in KNP and validation of FMD serological tests for wildlife. The lack of seropositive kudu from ENP—where buffalo populations are absent—may indicate that the source of infection for kudu in KNP was most likely the contact with FMD-infected buffaloes. As highlighted by Thomson, Vosloo, and Bastos [19] and Hargreaves et al. [122], antelope species (like kudu and impala) infected through contact with buffalo herds within the park have the potential to jump over the fences and transmit the virus to the cattle living in adjacent communal farms. SPCE is the official screening test in South Africa and Namibia for livestock, which is not validated for wildlife. In this study, the SPCE for SAT-1, SAT-2, and SAT-3 was negative in KNP and all serotypes in ENP. However, to our knowledge, this work represents the first attempt of FMD SAT serotyping in African non-buffalo species by SPCE [6]; hence, the sensitivity of the technique in these animals is not known as there has been no report, to our knowledge of SPCE for SAT in kudu and wildebeest. SPCE is serotype-specific meaning that it targets the structural proteins whose aminoacidic variability is per definition the highest among all viral proteins [119]. Antigenic variation is considered more common in wild animal populations, due to repeated exposure and immune selective pressure of a highly diverse population of infected host species [120, 121]. The strains of the serotypes (SAT-1, SAT-2, and SAT-3) coated to the plate of the SPCE may be significantly different than the ones circulating in KNP wildlife, as the SPCE is validated for livestock animals. Hence, the sensitivity of the SPCE might be mildly to markedly lower than the NSPCE, which on the other hand targets a highly conserved component of the FMDV capsid, that is, the 3ABC complex. Alternatively, positive reactions in kudu by NSPCE might be considered as false-positive results, although this is very unlikely due to the high specificity of the test (>99%) which does not depend on a species-specific conjugate (being a competitive ELISA) and also due to the high titers observed in 12 kudus from KNP (38%; 12/32). Additional research and characterization (using VNT or other tests) are strongly expected to shed light on this phenomenon and could be investigated in the future using available samples.

4.8. Seropositivity to FMDV in a Blue Wildebeest From ENP

One blue wildebeest (CT05, male, adult; 3%; 1/30) from ENP, Namibia, was found seropositive for FMD by means of NSPCE but seronegative using SPCE. This finding has to be interpreted cautiously because: the positive sample had a S/N percentage close to the ELISA cutoff (Supporting Information: Figure S1); all the other animals (kudu and wildebeest) from the same park, area and sampling period, tested negative by the assay; buffalo, considered the main maintenance host for FMD in wildlife, is not present in ENP [67]. FMD infection in blue wildebeest from Tanzania, Botswana, and Kenya has been reported by means of RT-PCR with serotypes O and A, SAT-1, and SAT-2 [121, 123]. Blue wildebeest may also suffer the clinical disease, developing oral and foot lesions associated with lameness, fever, and inappetence [123]. However, the NSPCE results were not confirmed with SPCE and thus require further investigation using a larger samples size and alternative techniques such as RT-PCR on oropharyngeal lymph nodes.

4.9. Confirmed Brucella Exposure in KNP Kudu, Questionable for Plains Zebra, Blue Wildebeest, and Impala

Three kudus (10%; 3/29) in KNP could be considered seropositive for Brucella spp. These animals reacted to two serological tests, and additional five kudus were positive to only one serological technique. Numerous studies conducted in southern Africa could not find any serological response in greater kudu, although sample sizes were often small (<30) and used serological tests validated for livestock [124–128]. In this study, seropositivity means that kudu were exposed to Brucella spp. and it remains unknown whether they are incidental hosts or part of the maintenance host community for Brucella spp. in wildlife. Three plains zebra from KNP (9%; 3/35) tested positive either with RBT (two animals) or iELISA (one animal) and were regarded as suspect cases. This is an area for additional research as agglutination reaction to Brucella spp. in zebra has been reported by a previous study [129]. The domestic horse, which is evolutionarily related to zebra, has been demonstrated to harbor different Brucella spp. (i.e., B. abortus and B. suis under natural circumstances and B. canis after experimental challenge) and may eventually experience clinical signs (fistulous withers, abortion, and other reproductive problems) [130]. Moreover, a study from Nigeria conducted by Bertu et al. [131] isolated B. abortus from asymptomatic horses living in a multispecies farm in Nigeria. However, the risk of transmission of brucellosis from equids is still to be clarified as horses have been indicated as dead-end host [132].

4.10. Widespread Exposure to Coxiella burnetii in KNP and ENP

In this study, a remarkably high number of individuals (57%; 106/185) across all evaluated wild animal species (44/65 zebras, 22/69 kudus, 14/21 impalas, 26/30 wildebeests) tested positive to the C. burnetii iELISA (Table 1). We also obtained many strong positive reactions (19%; 35/185) in any species considered (33/65 zebras, 9/69 kudus, 9/21 impalas, 18/30 blue wildebeests). Finally, our seroprevalence estimates were significantly different than those reported by Gakuya et al. [133], where similar wildlife species were investigated in Kenya using the same serological technique (iELISA). These findings led us to assume that C. burnetii is ubiquitous in both KNP and ENP and might have a predilection for southern Africa’s ecosystems and/or soils. A significantly higher seroprevalence was registered in animals from ENP. Q fever seroprevalence was especially higher in blue wildebeest, plains zebra, and impala. However, the multispecies C. burnetii iELISA has only been validated for use in domestic animals and not wildlife and has not been validated for wildlife species as iELISA tests are designed to be host-specific. The use of inaccurate tests could overestimate the prevalence of disease. In multiple species iELISA assays, IgG-binding proteins (such as protein A, protein G, and protein A/G) are suggested and used as conjugates [134–137], but it is not known how these react with every wildlife host species. According to Kelly et al. [135] and Stöbel, Schönberg, and Staak [137], impala, wildebeest, greater kudu, and zebra react weakly with protein A and strongly with protein A/G, while binding affinity with protein G varies; for impala and wildebeest, reactivity is weak, whereas for kudu, it is moderate and for zebra is strong. The binding affinity with protein A/G is particularly strong for kudu [135]. The Q fever iELISA kit employed in this study used protein A/G. Considering all the facts discussed above, additional investigation may determine if kudu is less affected/exposed to C. burnetii than the other species.

Further testing on tissues of wild animals matched with investigation in feeding ticks may provide important details for the clarification of Q fever epidemiology in African wildlife. Also, the expansion of C. burnetii investigations in predator animals may provide further information on the sylvatic cycle of the pathogen.

4.11. Limitations of the Study and Suggestions

We could detect reactions to nonspecific probes for Anaplasma/Ehrlichia and Theileria/Babesia in ENP but not too many of the species-specific probes investigated. This suggests that the strains present in ENP may not be detectable by the probes which were designed for strains occurring in South Africa due to the presence of local SNPs that do not allow binding with RLB probes. Sequencing data could characterize Anaplasma/Ehrlichia and Theileria/Babesia species occurring in ENP wildlife and thus design probes that can hybridize reliably also with these strains. It may also indicate the occurrence of new species that are not reported in the literature.

RLB probes cross-reactions are not infrequent, and a subset of positive samples should be sequenced to confirm specificity of the RLB probes. However, due to funding constraints, we could not sequence nor characterize any positive RLB occurrences. As a future study, it would be particularly interesting to sequence and confirm the occurrence of A. platys in kudu and impala and E. ruminantium in zebra from KNP, given their relevance for human and animal health. For serology, there is lack of known positive reference material from wild animals. Multispecies ELISA make use of conjugates that react with multispecies with cutoffs that are not animal species-specific. It is ideal to develop and validate ELISA assays specifically tailored for detecting FMDV, brucellosis, and Q fever across a range of wildlife species.

Our prevalence estimates have wide CIs due to small sample sizes and need to be interpreted cautiously. Interpretations and interventions are conducted by considering both the point estimate/prevalence as well as the entire confidence interval; that is where the true population lies with 95% confidence.

Samples used in this study were part of another project that aimed to unravel differences in exposure to anthrax in endemic and nonendemic locations. Although randomization was introduced as much as possible when selecting sampling units, a moderate–high selection bias has to be considered as it is not possible to extract a proper random sample from wildlife. Moreover, due to prior use in other research, the total number of available samples was reduced leading to a slight discrepancy in the number of animals tested for certain pathogens. For instance, out of the total 32 kudu samples collected from KNP, we had only 28 sera and 29 DNA samples available for testing. This depletion meant that for four of the 32 kudus, we had only one of the two sample types available (either DNA or sera but not both).