The field investigation was conducted over a continuous 12-month period spanning from January 2025 to December 2025 within communal grazing areas managed by the Raymond Mhlaba Local Municipality, Eastern Cape Province, South Africa (32°47′S, 26°50′E) (Figure 1). The municipality is situated within a semi-arid agroecological zone characterized by smallholder pastoral operations where cattle, sheep, and goats share open rangeland resources under a continuous, unrestricted grazing management framework. Annual precipitation across the cycle ranges from approximately 500–850 mm, concentrated heavily during the wet summer months (October to March). Mean annual ambient temperatures fluctuate between 12 °C and 27 °C, though acute microclimatic heatwaves regularly exceed these means during the peak summer index windows. The local rangeland vegetation community is dominated by mixed grassland and woody savanna canopy species, including Themeda triandra, Panicum maximum, Eragrostis curvula, Cynodon dactylon, Acacia karroo, and Vachellia species, which provide both low-lying grazing swards and elevated browsing canopy throughout the year (Mucina and Rutherford, 2006).

Ethical approval for all live-animal field operations was reviewed and granted by the University of Fort Hare Animal Research Ethics Committee (Approval No.: MAP021SSLA01). Animal care, seasonal handling protocols, and non-invasive field measurement procedures complied strictly with the South African National Standard for the Care and Use of Animals for Scientific Purposes. Prior to initiating any experimental animal monitoring, community informational workshops were held, and formal written informed consent was executed by all participating communal livestock owners.

A total of 120 clinically healthy, communally raised beef cattle aged between 3 and 6 years were selected for monitoring from participating village herds. Tracking growth performance in semi-mature and mature animals allowed for an evaluation of how thermal environmental stress maps onto body reserve maintenance and condition retention in a pastoral context. Because these test animals were sourced directly from multi-owner communal herds, they comprised indigenous crossbred genotypes (predominantly Nguni, Afrikaner, and Brahman crosses) of mixed sexes (comprising 74 multiparous cows and 46 castrated oxen). Due to the open, observational nature of communal pastoral systems, breed composition and sex distribution could not be balanced or artificially standardized; these are explicitly addressed as potential confounding factors within the analysis limitations. To ensure uniform environmental and nutritional exposure across the entire 12-month cycle, all 120 experimental animals were maintained together as a single, cohesive herd grazing within the same communal rangeland boundary. The experimental grazing area comprised a 450-ha unimproved natural rangeland pasture. This boundary contained zero artificial cooling structures, but featured natural shade resources provided by patches of mature Acacia karroo canopies, which were freely accessible to the cattle. For individual tracking and rapid behavioral verification by field teams, each animal was fitted with a unique, high-visibility, dual-sided numbered ear tag at the beginning of the study.

Cattle phenotypic coat characteristics were evaluated at the start of the study using standardized phenotypic assessment protocols (FAO, 2015; McManus et al., 2011). To verify hair structure parameters beyond visual appraisal, physical hair samples were collected from the mid-side flank region of each animal. Samples were secured by shaving a 5 × 5 cm patch down to skin level using specialized veterinary clippers. Collected strands were stored in sealed, desiccated containers for laboratory analysis. Hair length was quantified by sampling 20 individual strands per animal at random and measuring them under a digital caliper to determine a mean value (mm). Hair density (strands per cm²) was evaluated by counting all roots present within a standardized 1 mm² sub-grid magnified under an optical stereomicroscope. Hair texture was logged qualitatively as smooth or coarse based on tactile evaluation. For the primary statistical modeling, cattle were split into two operational groups matching their biophysical short-wave solar radiation absorptance and convective heat loss traits:

Light coat phenotype group (n = 39): Animals with solid white or light-cream/light-mixed coats, characterized by short (mean length <8 mm), smooth, and flat-lying hair profiles.

Because color, length, and density are physically linked within these phenotypes, treating them as separate variables can create multicollinearity issues in statistical models. We resolved this by calculating Variance Inflation Factors (VIF); testing them as completely independent effects yielded high VIF markers (>5.0), which confirms multicollinearity. Therefore, the integrated operational coat group term was utilized as the primary structural effect, while hair length and density were restricted to baseline descriptive covariates within the final mixed model framework.

Microclimatic conditions were recorded at 30-min intervals during all field observation sessions using portable digital weather stations positioned at mean animal height within the active grazing pasture zone. Recorded metrics included ambient air temperature (T, °C), relative humidity (RH, %), wind speed (m/s), and total solar radiation (W/m²). Black globe temperature (Tg, °C) was recorded using a standard 15-cm copper globe thermometer painted matte black to account for the combined forces of air temperature, radiant solar energy, and wind speed. The Temperature–Humidity Index (THI) was computed using the standard biometeorological equation from Mader et al. (2006): THI = 0.8 × T + R H 100 T − 14.4 + 46.4

Thermal environments were classified as: Thermoneutral (THI <72), Mild Heat Stress (THI 72–78), Moderate Heat Stress (THI 79–88), and Severe Heat Stress (THI >88). While the mean midday summer THI reached 84.7 ± 0.5, acute peak climatic shifts during extreme heatwaves reached a maximum THI value of 88.4 during summer midday intervals, providing empirical observations within the severe heat stress bracket (THI >88).

Daytime behavioral time budgets were documented twice monthly over the 12-month study using focal animal sampling methods (Altmann, 1974; Martin and Bateson, 2007). Each observation session lasted exactly 8 h, restricted to the daytime window from 08:00 to 16:00 h. Consequently, nocturnal thermal profiles, early-morning transitions, and nighttime compensatory grazing patterns were not captured. To systematically observe the 120 tagged animals across the pasture within this 8-h window without creating observer fatigue, field data collection relied on a team of four trained observers working concurrently. Each observer was assigned a specific sub-cohort of 30 animals per session, tracking individual animals via ear-tag identifiers at 10-min intervals. To maintain high inter-observer reliability, observers were rotated across sub-cohorts between sessions, and consistency was verified prior to study kickoff via independent parallel logs (yielding a Cohen’s kappa coefficient of k = 0.88).

To ensure behavioral budget classifications were mutually exclusive and cumulative (summing to a clean 100% daytime budget), a priority hierarchy matrix was enforced:

Grazing/Browsing: Active forage pulling, processing, or browsing within the low sward or woody shrub layers.

Initial tracking of low-frequency variables (such as explicit separations of standing vs. lying postures, fine-grained panting scores, and rare agonistic encounters) yielded highly zero-inflated datasets that destabilized model variance. These were collapsed into the broader, robust baseline categories above to preserve statistical power.

Cattle body weights were recorded monthly before morning grazing using a calibrated electronic livestock scale. Average daily gain (ADG; kg day^-1) was calculated as the total net body weight change over the entire 365-day experimental cycle divided by 365 days. Body condition score (BCS) was assessed monthly by a single evaluator using a standard 1–5 scale (Nicholson and Butterworth, 1986). Core physiological parameters were recorded monthly at 08:00 h and 14:00 h to capture diurnal variation. Rectal temperature (°C) was recorded using a digital veterinary thermometer, and respiration rate (breaths min^-1) was determined by counting flank movements for one full minute using a stopwatch (Gaughan et al., 2008; Sejian et al., 2018).

Data analysis was performed using R software (Version 4.3.2) (R Core Team, 2024). Behavioural allocations, physiological metrics, and performance outputs were evaluated using linear mixed-effects models via the lme4 package. Fixed effects included coat group, hair length covariate, hair density covariate, THI category, and the Coat group × THI category interaction. Individual animal identifier tags were specified as a random intercept effect to fully control for the repeated-measures covariance structure. The formal model was structured as: Yijkmn = μ + Ci + Lj + Dk + Tm + C × T im + An + ε ijkmn

Where Y is the response variable, μ is the grand mean, Ci is the fixed effect of the coat phenotype group, Lj is the hair length covariate, Dk is the hair density covariate, Tm is the fixed effect of the THI category, (C × T)im is the interaction terms, An is the random intercept effect for individual animal, and ε is the residual error. Denominator degrees of freedom and corresponding P-values were calculated using the Satterthwaite approximation via the lmer Test package. Percentage time budget inputs were arcsine-square root transformed prior to analysis to fulfill normality criteria. Model assumption compliance was confirmed through graphical inspections of residual normality plots and tests for homoscedasticity using Levene’s diagnostic test. Pearson correlation coefficients were calculated using data aggregated at the individual animal level (averages across the study cycle) to eliminate potential pseudoreplication from sequential repeated measures. Significance was declared at P < 0.05.

Midday observations regularly exposed cattle to elevated thermal loads, with peak ambient air temperatures reaching 32.8 °C ± 0.7 °C and mean midday THI values reaching 84.7 ± 0.5 (Table 1). However, during localized heatwave events in February, peak indices reached a maximum THI value of 88.4, providing empirical observations within the severe heat stress bracket (THI >88). Within the monitored herd, brown coats were the most frequent (25.8%), followed by black (23.3%), mixed (18.4%), red (18.3%), and white (14.2%). Short-haired phenotypes constituted 60% of the animals.

Daytime behavioral time budgets varied between phenotype groups (Table 2). Light-coated cattle spent a significantly higher proportion of daytime grazing (54.8% ± 1.2%) than dark-coated cattle (46.9% ± 1.1%; P < 0.001). Conversely, dark-coated cattle exhibited higher frequencies of walking, resting, drinking behavior, and shade-seeking (P < 0.001). Across all animals, elevated THI categories were associated with a progressive decline in grazing activity, falling from 59.8% ± 1.1% at THI <72 down to 40.6% ± 0.9% during severe heat stress windows (THI >88, Table 3). Shade-seeking expanded more than ten-fold across the same gradient (P < 0.001).

Dark-coated cattle exhibited higher mean rectal temperatures (39.4 °C ± 0.1 °C vs. 38.7 °C ± 0.1 °C; P < 0.001) and higher respiration rates (49.3 ± 1.4 vs. 37.2 ± 1.2 breaths min^-1; P < 0.001) than light-coated cattle. Regarding productive output (Table 4), initial body weights were similar between groups (P = 0.641). Over the full 12-month cycle, light-coated cattle achieved a higher final body weight (425.6 kg) and body condition score (3.41) than dark-coated cattle (P < 0.01). The actual net weight gain over 365 days corresponded to an ADG of 0.26 kg day^-1 for the light coat group and 0.20 kg day^-1 for the dark coat group (P < 0.001).

The analysis of variance confirmed that the thermal environment explained the largest component of variation in grazing activity (F = 42.17) and rectal temperature (F = 61.55, Table 5). Significant Coat × THI interactions were verified for all major traits (P < 0.01), indicating that the magnitude of behavioural modifications under heat stress varied depending on the coat group. At the individual animal aggregation level (Table 6), mean grazing activity shared a strong positive correlation with 12-month weight gain (r = 0.75; P < 0.001), but was negatively correlated with average core body temperature (r = −0.67; P < 0.001).

The progressive decline in daytime grazing activity from 59.8% under thermoneutral conditions (THI <72) to 40.6% under severe thermal stress (THI >88) underscores the sensitivity of foraging behavior to microclimatic stress. Under high thermal loads, grazing cattle face a dual heat burden: external radiant heat from solar radiation and internal metabolic heat generated by microbial fermentation and physical locomotion. This observed decline is directly supported by foundational heat-stress literature. For example, Polsky and von Keyserlingk (2017) and Cooke et al. (2023) established that extensive grazing beef cattle alter their feeding schedules to limit internal heat production during peak environmental heat loads. The dramatic reduction in grazing time, coupled with a corresponding increase in resting (11.5%–22.1%) and shade-seeking (0.4%–4.8%), represents a highly conserved adaptive strategy to minimize total heat gain. When environmental temperatures exceed an animal’s upper critical threshold, seeking natural or artificial shade serves as the primary behavioral mechanism to block short-wave solar radiation (Schütz et al., 2014; Lees et al., 2018). However, this safety strategy introduces a strict ecological trade-off. Because high-quality forage resources in extensive communal rangelands are spatially distributed away from shaded canopies, animals seeking microclimatic relief must abandon active grazing patches. This explains why grazing and resting behaviours were strongly negatively correlated (r = −0.79) in this study. These findings mirror those of Idris et al. (2024), who reported that beef cattle under extensive hot conditions systematically trade off foraging time for standing or resting under shade canopy to mitigate extreme solar loads.

The observed increase in drinking frequency (1.4%–3.3%) further corroborates the intensification of evaporative cooling mechanisms. Water consumption acts as a heat sink, lowering core body temperature through conduction while replacing fluids lost via cutaneous evaporation (sweating) and respiratory dissipation (panting). As documented by Sejian et al. (2018), access to adequate water is a critical asset for enabling livestock to maintain homeostatic balance under escalating THI thresholds. It is worth noting that low-frequency or transitional variables, such as explicit separations of standing vs. lying postures and detailed panting scores, were originally tracked but ultimately collapsed into broader baseline categories due to high zero-inflation. For instance, severe open-mouth panting (scores≥3) occurred only during peak midday summer hours, while fine-grained postural transitions were highly variable across multi-owner herds. Consolidating these data into robust, mutually exclusive time budgets ensured sufficient statistical power to resolve the primary effects of thermal stress on macro-level behavioral allocation.

A major finding of this study is that cattle possessing light-coloured, short-haired, and smooth coats maintained higher daytime grazing budgets (54.8%) and lower resting requirements (12.8%) than dark-coated cohorts with longer or denser hair. This divergence is directly grounded in the biophysics of heat exchange between an animal and its microclimate. Coat colour is the primary determinant of short-wave solar radiation absorptance. Finch (1986) demonstrated that dark pigments (melanin) absorb a vast majority of the visible and near-infrared spectra, converting solar energy into sensible heat at the coat surface. Lighter coats, particularly solid white or light cream, reflect a substantial portion of this incident radiant energy, preventing it from penetrating the hair fleece. This biophysical rule is well-supported by Silva et al. (2022a), who reported that light-coated dairy cows in tropical regions maintained significantly lower surface temperatures and higher behavioral comfort than dark-coated herdmates under identical environmental exposures.

However, color cannot be evaluated in isolation from hair morphology. Hair length, density, and texture dictate the thickness and insulation capacity of the boundary layer trapped against the skin. Short, smooth, and flat coats minimize boundary layer resistance, allowing convective air currents to strip heat away from the skin surface. Conversely, long or dense coats act as a highly effective barrier that traps warm air and impedes cutaneous evaporative cooling. This dynamic is validated by Silva et al. (2022b), who showed that under tropical conditions, a dense hair coat severely reduces the efficiency of sweating. The primary pathway for heat dissipation in cattle, by preventing moisture from evaporating directly at the skin surface. Consequently, dark-coated and dense-haired animals accumulate a higher net thermal load during daytime grazing, forcing them to activate behavioral avoidance strategies (such as seeking shade and halting grazing) much earlier in the diurnal cycle than their light-coated peers.

The physiological data reinforce the biophysical mechanisms observed behaviorally. Dark-coated cattle exhibited significantly higher mean rectal temperatures (39.4 °C) and elevated respiration rates (49.3 breaths min^-1) compared to light-coated animals. Core body temperature increases when net heat storage becomes positive, indicating that heat dissipation pathways are overwhelmed by the combined influx of environmental and metabolic heat. To counteract this thermal imbalance, the animal initiates homeostatic respiratory reflexes. The observed increase in respiration rate serves to maximize latent heat loss via the evaporation of moisture from the mucosal surfaces of the respiratory tract. While effective in the short term, sustained tachypnea and panting carry heavy metabolic consequences. As detailed by Gaughan et al. (2008) and Sejian et al. (2018), respiratory muscle activity requires considerable muscular work, shifting valuable energy away from growth and maintenance toward basic homeostatic survival.

Additionally, prolonged thermal stress elevates blood flow to the periphery (vasodilation) to maximize sensible heat loss, which simultaneously reduces blood flow to the gastrointestinal tract (Soliman et al., 2025). This splanchnic vasoconstriction can compromise gut integrity, induce mild systemic inflammation, and lower nutrient absorption efficiency. The strong negative correlation verified between individual mean rectal temperature and average daily gain (r = −0.61) highlights this systemic biological strain. Physiological heat stress does not merely alter behavior; it actively drains the animal’s net energy reserves through increased maintenance requirements and reduced digestive efficiency.

The corrected average daily gain values over the 12-month period (0.26 kg day^-1 for light coats and 0.20 kg day^-1 for dark coats) reflect the realistic production constraints of extensive communal rangelands in semi-arid zones. It is critical to interpret these growth metrics within the context of the age structure of the cohort (3–6 years old). Mature or semi-mature beef cattle managed under natural continuous grazing systems do not exhibit the rapid, linear weight accumulation seen in juvenile feedlot animals. Instead, as discussed by Mapiye et al. (2009), their weight profiles follow a cyclical trajectory, capturing periods of moderate gain during the wet, high-forage summer months, balanced by periods of maintenance or nutritional weight loss during the dry winter season. The significantly superior final body weight (425.6 kg) and body condition score (3.41) observed in the light coat group are directly explained by their sustained grazing budgets. Because light-coated cattle maintained more stable daytime foraging times across hot seasons, their cumulative annual dry matter intake may have been higher than that of the dark-coated animals, which frequently halted grazing to seek shade. We offer this strictly as a hypothesis, as volumetric dry matter intake was not directly quantified in this study. The strong positive correlation between aggregated individual grazing time and ADG (r = 0.75) confirms that behavioral resilience under heat load directly preserves long-term energy balances. Under extensive communal management, where nutritional supplementation is absent, maximizing daytime grazing efficiency is a primary indicator of weight maintenance and body condition retention.

The highly significant Coat × THI interactions verified for grazing activity, resting, and rectal temperature demonstrate that cattle responses to microclimatic stress are non-linear and depend heavily on phenotype-by-environment dynamics. Light-coated animals demonstrated a wider thermoneutral buffer, maintaining relatively stable grazing activity and core body temperatures up into mild and moderate THI categories. Dark-coated animals, by contrast, exhibited sharper behavioral drops and steeper physiological escalations as the environmental heat load intensified. This indicates that thermally adaptive coat characteristics increase the specific microclimatic threshold at which disruptive homeostatic adjustments become necessary. This threshold dynamic must be interpreted alongside broader macro-environmental variables like seasonality and changing forage availability. It provides an instructive point of comparison against recent findings from other extensive settings. In an evaluation of beef cattle behavioral plasticity on pasture, Paixão et al. (2026) observed that grazing and other active daytime behaviors were not significantly suppressed up to moderately elevated THI values, demonstrating a non-linear threshold buffer where cattle successfully navigated moderate heat loads before altering their time budgets. In contrast, our dataset demonstrated a steady, monotonic decline in grazing activity across the entire THI spectrum, with no evidence of a moderate plateau (Table 3). This divergence likely reflects the heightened environmental vulnerability of our specific communal rangeland context. Under the semi-arid, high-radiation conditions of the Eastern Cape, the lack of intensive infrastructure and the sparse distribution of shade canopies elevate radiant heat loads acutely. This environmental load overpowers minor behavioral adjustments, forcing an immediate reduction in active foraging to control metabolic heat generation.

To ensure the scientific validity of these findings, several constraints inherent to field-based observational research must be addressed:

Because coat phenotypes were naturally occurring within existing herds rather than randomly assigned via experimental treatments, this study cannot definitively isolate coat characteristics as an independent causal source. The identified patterns represent robust statistical associations rather than proof of direct causality.