Introduction

Bovine brucellosis, primarily caused by Brucella abortus, is one of the most important zoonotic diseases worldwide, with significant implications for livestock productivity, animal health, and human well-being [1,2]. In cattle, the infection manifests predominantly through reproductive losses, including abortion, retained placenta, and infertility, which collectively result in substantial economic losses for farmers, particularly in resource-limited production systems [2–4]. In humans, the disease can cause undulant fever, arthritis, and chronic complications, highlighting its zoonotic significance and reinforcing the need for integrated control under the One Health framework [2,5,6].

In South Africa, bovine brucellosis is a controlled disease under the Animal Diseases Act (Act No. 35 of 1984) [7], with interventions focused on statutory surveillance, farmer education, vaccination, testing, and culling [8–10]. Historically, surveillance has relied heavily on serological assays, particularly the Rose Bengal Test (RBT) as a screening tool and the Complement Fixation Test (CFT) as a confirmatory assay [11]. These conventional methods remain widely used because of their practicality and affordability within provincial veterinary laboratory systems (MPVL) [12]. However, limitations in sensitivity and specificity have been noted, particularly in chronically infected or vaccinated animals [13]. Recent advances in diagnostics, including enzyme-linked immunosorbent assays (ELISAs), fluorescence polarization assays, and molecular techniques such as PCR, offer improved sensitivity, specificity, and standardisation, and are increasingly advocated for integration into surveillance programmes [14,15]

Vaccination strategies have also evolved, with the classical B. abortus S19 vaccine being gradually replaced or complemented by RB51. While RB51 offers advantages for Differentiating Infected from Vaccinated Animals (DIVA), concerns remain about its efficacy and safety, particularly under field conditions [16]. These debates reinforce the need for context-specific vaccine policies aligned with regional epidemiology.

Despite diagnostic and vaccinology advances, provincial and subnational-level prevalence data in South Africa remain scarce. Past studies have reported Seroprevalence of around 2% in North West Province [17] and approximately 1.5% in KwaZulu-Natal communal [11,12]. More recent analyses highlight persistent spatial heterogeneity in prevalence across South African provinces and uneven implementation of control measures [10]. Similar findings from East and southern Africa reinforce the influence of herd structure, communal grazing, and biosecurity on disease persistence [4,18]. For example, serological surveys in Namibia and Ethiopia revealed higher prevalence in communal and smallholder systems compared to commercial herds, and molecular testing has increasingly been used to validate serological findings [19,20].

Mpumalanga Province, with its dense cattle population, diverse production systems ranging from commercial enterprises to communal grazing, and varied ecological conditions, remains understudied. Its proximity to international borders further raises the risk of transboundary spread of Brucella and complicates control efforts. Yet, to date, no comprehensive provincial-level analyses have been undertaken to explore temporal, seasonal, and spatial trends in Seroprevalence.

To fill this gap, the present study utilised a retrospective cross-sectional design based on diagnostic records from the MPVL over the period January 2021 to December 2024. Routine RBT results from cattle herds across 17 Local Municipality Areas (LMAs) were analysed. The objectives were to estimate the overall and annual Seroprevalence of B. abortus, to assess seasonal variations in Seropositivity across climatic periods, and to describe spatial heterogeneity between LMAs. By quantifying the independent effects of year, season, and locality, the study provides novel insights into the epidemiology of bovine brucellosis in Mpumalanga Province. These findings will inform targeted vaccination, improved surveillance allocation, and enhanced biosecurity, while contributing to South Africa’s national goal of progressive brucellosis control and eradication within the One Health framework.

Materials and methods

Results

Descriptive statistics

Annual trends in brucellosis seropositivity.

Between 2021 and 2024, a total of 568 batches comprising 67,974 cattle serum samples were tested for brucellosis at the MPVL. Of these, 6,182 (9.1%) tested positive on the RBT test. The annual distribution of batches and samples is presented in Table 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Annual trends in brucellosis Seropositivity.

https://doi.org/10.1371/journal.pone.0344037.t001

The number of batches submitted each year ranged from 76 in 2023–215 in 2022, with corresponding sample submissions varying from 11,707 (2023) to 25,283 (2022). The overall proportion of RBT-positive samples fluctuated across the years. In 2021, 6.94% (95% CI: 6.58–7.31) of samples tested positive, increasing slightly to 7.42% (95% CI: 7.11–7.74) in 2022. A notable spike occurred in 2023, when 13.05% (95% CI: 12.48–13.63) of samples tested positive, representing nearly a twofold increase relative to prior years. But in 2024, the proportion of positive cases declined to 7.50% (95% CI: 7.08–7.94), returning to levels consistent with 2021–2022.

Seasonal variation in brucellosis Seropositivity.

Table 2 summarises the distribution of brucellosis Seropositivity by season. A total of 568 batches comprising 67,974 cattle serum samples were tested across the four seasonal periods.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Seasonal variation in brucellosis Seropositivity.

https://doi.org/10.1371/journal.pone.0344037.t002

Marked seasonal differences were observed and the highest proportions of RBT-positive samples recorded in spring (10.57%, 95% CI: 10.08–11.09) and summer (10.24%, 95% CI: 9.77–10.71). Together, these two seasons accounted for more than half of the total seropositive cases. In contrast, lower positivity rates were documented in autumn (6.79%, 95% CI: 6.38–7.22) and winter (6.94%, 95% CI: 6.65–7.23), despite larger sample submissions during winter (n = 27,344).

Spatial distribution of brucellosis Seropositivity.

Table 3 presents the distribution of brucellosis testing outcomes by Local Municipality Area (LMA). A total of 568 batches comprising 67,974 cattle serum samples were analysed across 17 LMAs, revealing marked spatial heterogeneity in RBT seropositivity. In terms of absolute number of RBT-positive samples, the highest counts were observed in Msukaligwa (n = 1,457; 9.82%, 95% CI: 9.35–10.31), followed by Emalahleni (n = 856; 13.31%, 95% CI: 12.49–14.17), Govan Mbeki (n = 721; 8.73%, 95% CI: 8.13–9.36), and Steve Tshwete (n = 667; 8.40%, 95% CI: 7.80–9.03). These municipalities accounted for the majority of seropositive detections in the province.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. The distribution of brucellosis Seropositivity by Local Municipality Area (LMA).

https://doi.org/10.1371/journal.pone.0344037.t003

When examined by seroprevalence, Emalahleni recorded the highest proportion of RBT-positive samples (13.31%, 95% CI: 12.49–14.17), closely followed by Victor Khanye (12.99%, 95% CI: 11.43–14.69), Mbombela (12.03%, 95% CI: 10.86–13.28), and Dipaleseng (10.90%, 95% CI: 9.68–12.22), identifying these municipalities as areas of comparatively higher burden. Conversely, Mkhondo (1.69%, 95% CI: 1.28–2.21) and Albert Luthuli (2.72%, 95% CI: 2.22–3.28) recorded the lowest seroprevalence, suggesting relatively lower transmission intensity or more effective local control.

Intermediate levels of seropositivity were observed in Bushbuckridge (5.39%), Nkomazi (5.89%), Lekwa (6.63%), and Thaba Chweu (5.92%), representing municipalities with a moderate burden of infection, while Thembisile Hani (4.54%) and Emakhazeni (4.82%) fell toward the lower end of the distribution.

Inferential statistics

Univariate logistic regression.

Table 4 presents the unadjusted ORs for predictors of brucellosis Seropositivity based on year of sampling, season, and LMA. Compared with the reference year 2021, the odds of RBT Seropositivity were significantly higher in 2023 (OR = 2.00, 95% CI: 1.87–2.14, p < 0.001), indicating that cattle tested in this year had nearly twice the odds of being seropositive. In 2024, the odds of animals testing positive (OR = 1.09, 95% CI: 1.01–1.19, p = 0.036), was still statically higher compared to the reference year. Modest statistically significant increases were also observed in 2022 (OR = 1.08, 95% CI: 1.00–1.16, p = 0.048).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Results of the univariate logistic regression analysis.

https://doi.org/10.1371/journal.pone.0344037.t004

Relative to winter, both spring (OR = 1.61, 95% CI: 1.50–1.72, p < 0.001) and summer (OR = 1.57, 95% CI: 1.47–1.68, p < 0.001) were associated with substantially increased odds of Seropositivity. No significant difference was observed for autumn compared to winter (OR = 0.97, 95% CI: 0.89–1.05, p = 0.420).

Using Mkhondo (lowest prevalence) as the reference, cattle from Emalahleni (OR = 8.84, 95% CI: 6.60–11.8, p < 0.001), Victor Khanye (OR = 8.74, 95% CI: 6.21–12.3, p < 0.001), and Mbombela (OR = 7.82, 95% CI: 5.80–10.6, p < 0.001) exhibited markedly higher odds of brucellosis Seropositivity, suggesting these municipalities are high-burden hotspots. Albert Luthuli also showed a significantly elevated OR (OR = 1.64, 95% CI: 1.23–2.18, p < 0.001), though the magnitude of effect was smaller.

Multivariate logistic regression analysis.

Table 5 presents the adjusted odds ratios (AORs) for predictors of brucellosis seropositivity after simultaneous adjustment for year, season, and Local Municipality Area (LMA). A pronounced temporal effect was observed, with the odds of RBT seropositivity nearly doubling in 2023 compared with 2021 (AOR = 1.95, 95% CI: 1.81–2.11; p < 0.001). In contrast, 2022 showed only a borderline increase (AOR = 1.10, 95% CI: 0.99–1.22; p = 0.081), while 2024 was statistically indistinguishable from 2021 (AOR = 0.95, 95% CI: 0.86–1.05; p = 0.280), indicating a marked but temporally confined surge in seropositivity during 2023 that was not sustained thereafter. Seasonal associations differed from those observed in univariate analyses.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. Multivariate logistic regression results.

https://doi.org/10.1371/journal.pone.0344037.t005

With autumn as the reference category, spring exhibited significantly lower odds of RBT seropositivity after adjustment (AOR = 0.41, 95% CI: 0.24–0.70; p = 0.001). In contrast, summer (AOR = 0.66, 95% CI: 0.26–1.63; p = 0.363) and winter (AOR = 0.82, 95% CI: 0.39–1.76; p = 0.619) were not statistically significant predictors in the adjusted model. The attenuation and change in direction of seasonal effects relative to univariate analyses likely reflect correlations between season, testing year, and sample submission patterns rather than a biological reversal of seasonal risk.

Seasonal sample sizes in the adjusted model were unequal, winter (n = 27,344), autumn (n = 13,203), spring (n = 12,966), and summer (n = 14,461), which may have contributed to imprecision in adjusted estimates and warrants cautious interpretation.

In contrast, strong and consistent spatial effects were observed: compared with Mkhondo, the odds of seropositivity were approximately six- to seven-fold higher in Emalahleni (AOR = 6.90, 95% CI: 5.12–9.28; p < 0.001), Victor Khanye (AOR = 6.75, 95% CI: 4.77–9.57; p < 0.001), and Mbombela (AOR = 6.21, 95% CI: 4.53–8.50; p < 0.001), indicating persistent municipality-level heterogeneity in brucellosis burden.

Discussion

This study provides the first province-wide assessment of bovine brucellosis seropositivity in Mpumalanga Province using routine laboratory surveillance data over a four-year period. The findings demonstrate pronounced temporal variability, evidence of seasonal patterning, and substantial heterogeneity between local municipalities, underscoring the complex epidemiology of brucellosis in this setting and the importance of subnational analyses for informing control strategies [10,21,22].

The temporal pattern observed, characterised by a marked increase in seropositivity during 2023 followed by a decline to levels comparable with earlier years, suggests episodic amplification of transmission rather than a sustained upward trend. Similar transient increases have been reported in other endemic settings and are often associated with changes in surveillance intensity, outbreak-driven testing, or disruptions in control measures such as vaccination or movement regulation [21,23]. In the present study, these explanations should be interpreted cautiously and regarded as hypotheses, as routine surveillance data do not allow direct assessment of vaccination coverage, animal movement, or biosecurity practices at herd level [24,25].

Seasonal variation in seropositivity is biologically plausible in the South African context. Warm and wet months are associated with increased cattle aggregation at shared grazing areas and water points, creating conditions conducive to pathogen transmission [26]. These periods also overlap with calving and abortion events, during which Brucella organisms are shed in high concentrations into the environment, increasing opportunities for transmission [27,28]. In contrast, cooler and drier months are typically characterised by reduced animal contact intensity and lower environmental persistence of the organism, which may contribute to comparatively lower transmission risk [29,30]. However, the attenuation and change in direction of seasonal effects observed after multivariable adjustment indicate that apparent seasonal patterns are influenced not only by biological processes but also by correlated factors such as testing year, submission practices, and uneven seasonal sampling. Routine surveillance data are particularly susceptible to such effects, as testing is often reactive rather than uniformly distributed throughout the year [25,30]. Consequently, the adjusted findings do not imply a biological reduction in transmission risk during warmer months but instead highlight the challenges of disentangling true seasonal drivers from surveillance artefacts in retrospective datasets.

Marked spatial heterogeneity was observed across municipalities, with some areas consistently exhibiting higher seroprevalence and adjusted odds of seropositivity than others. These findings indicate areas of relatively higher burden rather than confirmed transmission “hotspots,” as formal spatial modelling and fine-scale geolocation data were not available. Spatial heterogeneity in bovine brucellosis has been widely documented in South Africa and elsewhere in sub-Saharan Africa and is often linked to differences in herd structure, production systems, communal grazing practices, and access to veterinary services [22,31–33]. These patterns reinforce the need for locally tailored approaches to surveillance and control rather than uniform provincial strategies.

From a programmatic perspective, the findings support the adoption of risk-based surveillance and control measures. Prioritising municipalities with consistently higher seroprevalence for enhanced testing, vaccination, and farmer engagement may improve the efficiency of resource allocation and strengthen provincial brucellosis control efforts [10,34]. In addition, although seasonal effects were attenuated after adjustment, the convergence of biological plausibility and unadjusted seasonal patterns suggests that scheduling surveillance and preventive interventions ahead of traditionally high-risk periods may still be operationally advantageous, provided such strategies are evaluated using prospectively designed data [30].

Several limitations should be considered when interpreting these findings. The analysis relied on routine laboratory submissions and on the Rose Bengal Test without systematic confirmatory testing, which may have introduced some degree of misclassification [11,35]. The retrospective nature of the dataset and the absence of herd-level information, including vaccination history, animal movement patterns, and management practices, limited causal inference and the ability to explore hierarchical or network-based transmission dynamics [24,31]. In addition, although spatial heterogeneity was evident, the lack of formal spatial modelling and detailed geospatial visualisation constrained interpretation of geographic clustering. Despite these limitations, the large sample size, multi-year coverage, and consistent analytical approach provide robust descriptive insights into the epidemiology of bovine brucellosis in Mpumalanga Province and offer a strong foundation for future prospective, spatially explicit, and One Health–oriented studies [21,22,34].

This study has several methodological strengths. First, it utilised a large provincial dataset comprising more than 67 000 cattle samples collected over four consecutive years, providing substantial statistical power and temporal coverage. Second, the use of routine surveillance data enhances the operational relevance of the findings, as the results reflect real-world testing patterns within the provincial veterinary system. Third, the analysis incorporated both descriptive and multivariable regression approaches, allowing the independent effects of year, season, and locality to be examined while controlling for potential confounding.

However, several limitations should be considered. The study relied exclusively on routine laboratory submissions, which are not based on a probability sampling design and may therefore be influenced by outbreak-driven testing or other operational factors. The use of the Rose Bengal Test without systematic confirmatory testing may have introduced some degree of misclassification. In addition, the absence of herd-level identifiers and management data, including vaccination history, animal movements, and biosecurity practices, precluded multilevel modelling and limited causal inference. Seasonal sample sizes were also unequal, which may have affected the precision of adjusted seasonal estimates. Finally, although spatial heterogeneity was observed, the absence of fine-scale geolocation data and formal spatial modelling constrained the interpretation of geographic clustering.

Despite these limitations, the large sample size, multi-year coverage, and consistent analytical approach provide robust descriptive insights into the epidemiology of bovine brucellosis in Mpumalanga Province and offer a strong foundation for future prospective and spatially explicit studies.

Conclusions

This study provides a province-wide description of bovine brucellosis seropositivity in Mpumalanga Province based on routine laboratory surveillance data collected between 2021 and 2024. The findings demonstrate notable temporal variability, evidence of seasonal patterning, and marked heterogeneity between local municipalities. These patterns indicate associations between seropositivity and time, season, and place, rather than causal relationships.

The transient increase in seropositivity observed during 2023, followed by a return to earlier levels, suggests episodic fluctuations in detected infection burden rather than a sustained increase over time. Seasonal differences observed in unadjusted analyses should be interpreted cautiously, as adjusted models indicate that these patterns are influenced by correlated factors such as year of testing and submission practices. Similarly, municipalities with consistently higher seroprevalence are best interpreted as areas of relatively higher burden rather than confirmed transmission hotspots, given the absence of formal spatial modelling and herd-level data.

Overall, these findings highlight the value of routine laboratory data for identifying temporal and geographic patterns that can inform risk-based surveillance and prioritisation of control activities. Future prospective studies incorporating balanced seasonal sampling, confirmatory diagnostics, herd-level metadata, and spatially explicit analyses are needed to more robustly assess drivers of transmission within a One Health framework.

Acknowledgments

We thank the Mpumalanga Department of Agriculture, Rural Development, Land and Environmental Affairs (DARDLEA) for granting formal permission to access and analyse the dataset, and the Mpumalanga Provincial Veterinary Laboratory (MPVL) team for providing the data used in this study. We also thank the College of Agriculture and Environmental Sciences Animal Research Ethics Committee, University of South Africa (UNISA), for reviewing the study protocol and granting ethical clearance (AREC-100818–024).