This study was conducted in the Salambala Conservancy in northeastern Namibia. Officially registered in 1998, Salambala is one of Namibia’s pioneer communal conservancies and has since gained recognition for its role in community-based natural resource management (NACSO, 2023). The conservancy spans approximately 9,300 ha and supports a population of around 9,193 residents distributed across 18 villages and over 290 cattle posts (Figure 1). The terrain is predominantly flat, with an elevation ranging from 928 to 983 m above sea level, and the vegetation consists of a woodland–grassland mosaic, dominated by mopane and Kalahari woodlands in the north and open grasslands in the south (Mendelsohn et al., 1997). The area experiences a semi-arid climate, with annual rainfall averaging 650–700 mm, falling mainly between October and April, and temperatures ranging from 21 °C to over 35 °C depending on the season (Krug, 2017). The agricultural system is characterised by small-scale, subsistence farming, where communal grazing is practiced year-round. Cattle are the dominant livestock, and common breeds include Sanga, Brahman-Sanga, and Simbra crosses. Livestock are typically kraaled at night and graze freely during the day with minimal supplementation. Tick infestations are common due to the extensive grazing system, and tick-borne diseases such as anaplasmosis, babesiosis, and heartwater remain persistent challenges in the region (Haikukutu et al., 2020; Githaka et al., 2022). Chemical control using acaricides is the primary strategy employed by farmers, although accessibility remains a challenge, with most products only available in the regional capital, Katima Mulilo. This logistical constraint, along with high product costs, has led some farmers to use unconventional or informal substitutes. Salambala was selected as the study site due to its ecological complexity, high livestock–wildlife interface, and documented presence of oxpeckers, making it a suitable setting for investigating tick control practices, oxpecker-host dynamics, and conservation attitudes in a real-world communal farming context.

The interviewed number of cattle keepers (n) was calculated using the formula by Thrusfield (2007): n = z 2 p 1 − p d 2 , (1) where, n is the required sample size, Z is the standard normal deviate corresponding to a 95% confidence level (Z = 1.96), p is the expected proportion of cattle farmers with experience of oxpeckers, and d is the desired absolute precision (0.05). A previous study found a total experience of 10.9% (p = 0.109) among cattle farmers in Nigeria (Abakpa et al., 2023). The initial sample size (n = 149) was adjusted for a finite population (N = 9193) using the correction formula: n a d j = N × n / N + n (2) resulting in an adjusted sample size of 146. To accout for potential non-response, the sample size was inflated by 37% (non-response rate based on preliminary field expectations and similar rural surveys), yielding a final target sample size of approximately 200 respondents.

A stratified random sampling approach was used to ensure representation across the Salambala Conservancy. A complete list of villages within the conservancy was obtained from local administrative records, and villages were treated as primary sampling units. A total of 15 villages out of 30 villages in Salambala Conservancy (50%) were randomly selected, with the village number decided on by the available survey resources.

Within each village, households owing cattle were identified with the assistance of community leaders and extension officers. From these lists, households were assigned unique identification numbers and selected using a random number generator in Microsoft Excel. Only one respondent per household was interviewed, and eligibility required that the individual be actively involved in cattle management and aged 18 years or older. In cases where a selected household was unavailable or declined participation, it was replaced by the next randomly generated household from the same village list. The 200 targeted households were distributed across the villages in proportion to the village household populations. The number of households selected per village ranged from 10 to 20.

Data were collected through structured face-to-face interviews using a pre-tested questionnaire. The questionnaire was developed based on a review of relevant literature and previous socio-ecological studies on human-wildlife interactions, and was validated by co-authors who have experience in community and oxpecker behaviours. A total of 10 pilot interviews (five in English and five in Subia) were conducted prior to the main survey to refine question clarity and translation accuracy. All interviews were conducted by a single trained interviewer to ensure consistency, and each interview lasted approximately 20–25 min. Interviews were conducted in either English or Subia, the predominant local language in the area. This approach minimised literacy-related bias, ensured consistency in data collection, and allowed real-time clarification of responses. The methodology follows established approaches in community-based conservation research in rural Africa (Roe et al., 2009; Waylen et al., 2010).

The questionnaire captured four main domains: (i) socio-demographic characteristics, (ii) tick control practices and chemical usage, (iii) knowledge and observations of oxpeckers, and (iv) perceptions and attitudes toward oxpeckers. To assess perceived oxpecker impacts in the community, responses from six variable questions were scored. The impact variables were: (1) presence of positive experiences (yes/no), (2) presence of negative experiences (yes/no), (3) frequency of observed interactions, (4) perceived severity of oxpecker-related problems, (5) reported economic losses, and (6) overall perception of oxpecker impact. Each variable was assigned a numerical score based on predefined criteria (Tables 3, 4), with higher scores indicating more positive perceived oxpecker impact and lower scores indicating more negative perceived impact. An aggregate oxpecker impact score was then calculated for each respondent by summing standardised item scores. This generated an outcome continuous variable to allow comparison across participants.

Data from the questionnaire interviews were processed in Microsoft Excel spreadsheets before descriptive, random forest and linear modeling, univariate and multivariable analyses in R statistical software 4.4.3 (R Core Team, 2024) at 5% significance level and with a two-tailed approach. The R packages “ggplott2” (Wickham, 2016), “MASS” (Venables and Ripley, 2002), “rcompanion” (Mangiafico, 2026), “rpart.plot” (Milborrow, 2024), lmtest (Zeileis and Hothorn, 2002), “randomForest” (Liaw and Wiener, 2002) and “rpart” (Therneau et al., 2025) were utilised. Descriptive analysis was conducted to determine the percentage and 95% confidence levels of responses with regard to socio-demographics, tick control practices, knowledge, attitudes and perceived or experienced impact of oxpeckers amongst the cattle farmers. An association pathway for the perceived or experienced impact of oxpeckers relative to predictor variables was hypothesised (Figure 2) and this was the basis for selection of 14 explanatory variables for analysis. A combination of random forest (RF) and linear modelling approaches (Breiman, 2001) was used to evaluate the association between the 14 socio-demographic variables, tick control practices and knowledge about oxpeckers with the outcome variable (overall perceived or experienced positive impact of oxpeckers). The non-parametric method, random forest, partitions the data into clusters in order to maximise the variance between groups while minimising the variance within a group. The basis for the method is on the classification and regression trees (CART), and this combines predictions from random subsets in the data to produce various decision trees. The random forest method is of significance when working with data that is less balanced and, in the need, to explore multiple interactions in datasets characterised by many variables. In this way, the impact of each explanatory variable is measured against the outcome variable, while measuring the relative proximity of multiple data points. The magnitude of change in response variable due to the explanatory variables is also measured. The final random forest plot revealed three explanatory variables that provided superior split to others in the dataset, in relation to the overall oxpecker impact score. The three variables were: whether respondents applied any solutions to reduce oxpecker impact (yes, no), whether the respondent wanted the population of oxpeckers to increase (yes, no), and how often the respondent saw oxpeckers on his or her farm in the past wet season (weekly, bi-weekly, monthly, every 3 months). Univariate Generalised Linear Models (GLM) with Poisson distribution were conducted to establish association of oxpecker impact scores with socio-demographic, tick control practice, and oxpecker knowledge variables. Multicollinearity between explanatory variables was assessed using the Cramer’s V test, with a threshold value of less than 0.6 for non-collinearity. Statistically significant variables (likelihood p-value <0.05) from univariate analysis were examined further in a multivariable GLM model, which controlled for confounders, in order to establish association with the outcome variable.

The majority of respondents were male and actively engaged in conservancy structures (Table 1). Employment levels were generally low, with most respondents relying on formal or non-salaried income sources. Household incomes were modest, and livelihoods were supported through a combination of government grants, wages, and family support. Farming experience was typically extensive, with many respondents engaged in livestock production for several decades. Educational attainment was generally low, with most respondents having primary or no formal education, and inly a small proportion attaining higher education (Table 1).

Tick control was widely practiced among farmers (78%; Table 2), with chemical methods being the dominant approach. Synthetic pyrethroids and organophosphates were the most commonly reported acaricides, although a range of other substances including unconventional products were also used. Application frequency varied across respondents, with some applying treatments regularly while others reported no use of tick control measures. Tick control practices were more commonly associated with the wet season (Table 2).

Awareness of oxpeckers was nearly universal among respondents (Table 3). However, the ability to distinguish between species was limited, with most respondents recognising only one type. Oxpecker sightings were frequent, particularly during the wet season, indicating regular interaction between livestock and the birds (Table 3).

Most respondents reported positive experiences with oxpeckers, primarily related to their role in removing ticks from livestock (Table 3). A smaller proportion of respondents reported negative interactions, mainly associated with wound aggravation, blood-feeding, and livestock discomfort (Table 3).

Most respondents did not consider oxpeckers to be problematic, and financial losses attributed to them were uncommon (Table 4). Perceptions of population trends were mixed, with respondents reporting both increases and declines. Reported causes of decline included reduced tick availability, chemical use, and loss of host species or nesting habitat (Table 4). Despite some negative experiences, the majority of respondents supported efforts to increase oxpecker populations and suggested conservation-related interventions (Table 4).

The random forest model explained 72.0% of the variation in oxpecker impact scores. The most influential variables were the application of control measures against oxpeckers, willingness to support population increases, and frequency of oxpecker observations during the wet season (Figure 3). The classification and regression tree indicated that higher impact scores were associated with respondents who supported increasing oxpecker populations and did not apply control measures, whereas lower scores were associated with respondents who applied such interventions (Figure 3).

Univariate Poisson regression analyses showed that all three variables from the random forest plot (whether the respondent applied any interventions to reduce the oxpecker population, their attitude towards increase or increase in the oxpecker population, and how often they saw oxpeckers in the wet season) were significantly associated with perceived oxpecker impact (Table 5). Out of the 200 cattle farmers interviewed, only 23.0% reported applying solutions to reduce oxpecker presence. Additionally, 63.5% of respondents supported an increase in oxpecker populations. Oxpecker sightings were most frequently reported on a weekly basis (35.0%), with significant variation across observation frequencies. The three variables were subsequently included in the multivariate model, which showed that respondents who applied control measures had significantly lower impact scores (more negative perception) compared to those who did not. Conversely, respondents who supported increasing oxpecker populations had significantly higher impact scores (more positive perceived impact). The frequency of oxpecker sightings during the wet season was not significantly associated with impact scores after adjusting for other variables (Table 6).

This study demonstrates valuable interactions between ticks, livestock and oxpeckers in communal rangelands and these are shaped more by socio-economic conditions and conservation attitudes than by direct ecological exposure. While oxpeckers are generally perceived as beneficial for tick removal, their acceptance depends on management practices and farmer experiences. The generally positive attitudes observed provide a strong foundation for developing community-based conservation. However, promoting sustained coexistence will require improved access to veterinary services, safer tick control practices, and targeted ecological education. Integrating local knowledge with conservation strategies will be essential for enhancing both livestock health and oxpecker conservation in communal landscapes.