The experiment was conducted in integrated crop-livestock system long-term protocol installed in 2013 in the municipality of Pinhais, state of Paraná, Brazil, at the Experimental Farm of the Federal University of Paraná (UFPR) (25°24’4.31’’ S; 49°7’15.02’’ W; 918 m.a.s.l.). Climate in the region is subtropical highland (Cfb), with mean temperatures ranging from 12.5 to 22.5 °C, annual average rainfall of 1400 mm, and severe and frequent frosts occurring 10 to 25 days per year on average, as described by Köppen. The study area is located within the Iraí State Environmental Protection Area (APA), where the use of agrochemicals is prohibited, in compliance with State Decree No. 1.753, dated 06 May 1996 (Paraná, 1996).

The present experiment was conducted from 31 July 2020 to 02 April 2021, totaling 245 days, the animals had 20 days of adaptation before the beginning of the evaluations in the experimental area. Two production systems were evaluated as treatments: Livestock and Livestock-forestry in an area of 10.2 ha. The systems were arranged in a randomized complete block design with three replicates, totaling 6 experimental units (paddocks) of 1.7 ha each.

The systems evaluated were integrated livestock-forestry (LF), characterized by the integration of Eucalyptus benthamii with livestock production, and specialized livestock (L) production which is not shaded. During the evaluations, the trees were seven years old and distributed along contour lines on the ground with a spacing of 28 m between single rows and 7.5 m between trees. The density was 47 trees ha-1, and the average shade percentage was 31 and 22 in winter and summer, respectively. For more information on the tree component, see Kruchelski et al. (2022).

In the summer of the 2013/14 crop year, summer perennial pasture was established in both systems using Megathyrsus (formerly Panicum) maximum, along with spontaneous species such as African stargrass (Cynodon plectostachyus), Hemarthria altissima, Urochloa (formerly Brachiaria) plantaginea, and kikuyu grass (Pennisetum clandestinum). In the winter pasture, it was overseeded black oat (Avena strigosa) intercropped with ryegrass (Lolium multiflorum) with 60 and 15 kg ha-1 of seeds, respectively, at the end of May 2020. The winter pasture also included other spontaneous forage species such as white clover (Trifolium repens) and vetch (Vicia sp.; Table 1 ).

Three crossbred Angus steers (average initial weights of 177 ± 40 kg; 12 months of age) were allocated per paddock, totaling 18 tester animals (nine per system). Pastures were managed under continuous stocking with variable stocking rate (put-and-take method; Mott and Lucas, 1952) to maintain a target sward height of 24 cm. Sward height was monitored every 15 days by measuring 150 points per paddock with a sward stick, and stocking rate was adjusted accordingly. Tester steers remained full-time in the experimental units. Put-and-take animals of similar weight and age were added or removed as needed to regulate sward height and ensure consistent grazing intensity across paddocks. The average sward height and average stocking rates were presented in Table 1 .

In all systems, animals had access to water troughs equipped with float valves for refill after consumption. The animals’ diet consisted exclusively of pasture and ad libitum mineral salt, provided in covered troughs. The internal dimensions of the water troughs in all paddocks are 1.30 m long x 1.30 m wide x 0.60 m deep, providing 0.65 m of linear water per animal. Furthermore, the maximum distance traveled by the animals to the water trough was 190 m.

This study was approved by the Ethics Committee on the Use of Animals, Agricultural Sciences Sector of UFPR, under protocol no. 076_2019.

Eight welfare criteria were selected to address the four welfare principles established by the Welfare Quality® protocol (2009 – Table 2 ). These measures were selected based on the local conditions of extensive cattle production (Huertas et al., 2009), as well as on the feasibility and relevance of the measures as reflections of the welfare promoted by the production system. Table 2 describes the adapted measures used to assess the welfare criteria for cattle grazing and local conditions.

A completely randomized design was adopted, with two treatments (L and LF) and nine replications per treatment. The experiment was conducted across 10 evaluation periods, considered as repeated measures over time. A mixed-effects model was applied for data analysis, with treatment as a fixed effect and evaluation periods as a random effect, using the MIXED procedure.

Normality of the variables was assessed using the Shapiro-Wilk test. Homogeneity of variances was evaluated using Bartlett’s test, and the independence of residuals was also verified. Only the variables ADG and the frequencies of grazing, rumination, and other activities showed a normal distribution. When significant differences were detected, the means were compared using the least-squares means (LS-means) model. The best variance structure was selected based on the lowest Akaike Information Criterion (AIC) value. Interaction between treatments and evaluation periods was further examined when statistically significant at the 5% probability level.

Other variables that did not meet the assumption of normal distribution, even after transformation, were analyzed using nonparametric analysis of variance (ANOVA) through the Kruskal-Wallis test. All statistical analyses were conducted on RStudio 4.0.4 software, and a 5% significance level (p < 0.05) was adopted.

For the BCS and ADG variables, no statistically significant differences were observed between the L and LF systems across the average of all evaluation periods (p = 0,563; p = 0,962 respectively; Table 3 ). Among evaluation periods, considering the average of both systems, the highest BCS was recorded at evaluation 10 (p < 0,001) ( Table 3 ), while the lowest ADG values were observed at evaluations 7, 8, and 9 (p < 0,001; Table 3 ). A significant interaction (p = 0,009) between livestock systems and evaluation periods was observed only for the ADG variable ( Table 3 ). At evaluations 7 and 8, animals in the LF system showed higher ADG values than those in the L system. In contrast, during evaluation 6, the L system yielded higher ADG than the LF system.

Across evaluation periods and systems, statistically significant differences were observed between time slots (p < 0,001) and activity frequencies (p = 0,009), with no interaction between time intervals and activity frequencies (p = 0,618). The highest frequency of grazing was observed during time interval 3, rumination peaked in time interval 2, and other activities (e.g., idling, social interaction, water intake, and mineral consumption) were most frequent in time intervals 1 and 2 ( Figure 3 ). Grazing and other activities were most frequent in time interval 1, rumination was less frequent than grazing in time interval 2, and grazing was more frequent than all other activities in time interval 3 ( Figure 3 ).

The body surface temperature of cattle differed between the systems (p < 0,001), being 2.5 °C lower in animals raised in the LF system compared with those in the L system, averaged across all evaluations ( Table 3 ). Averaged across both systems, body surface temperature was higher (p < 0,001) at evaluations 6, 9, and 10 and did not differ from evaluations 7 and 8 ( Table 3 ).

THI did not differ between the livestock systems (p = 0,915; Table 3 ). Across evaluation periods, averaged across systems, THI values were higher at evaluations 7, 8, 9, and 10 and lower at evaluation 1, with no difference from evaluation 2 (p < 0,001; Table 3 ).

No statistically significant differences were observed between the L and LF systems for LS (p = 0,150), FCS (p = 0,753), or ectoparasite infestation scores for botfly (p = 0,192), ticks (p = 0,570), and horn flies (p = 0,631), regardless of evaluation period ( Table 3 ). The FCS was higher at evaluation 3 and lower at evaluation 1 in both systems (p = 0,009; Table 3 ). On average, botfly infestation scores were higher at evaluations 4 to 8 (p < 0,001), tick infestation scores were lower at evaluations 1, 2, 4, and 9 (p < 0,001), and horn fly infestation scores were lower at evaluations 1, 2, and 9 (p < 0,001; Table 3 ).

Frequency of nasal discharge did not differ between the L and LF systems (p = 0,703), however, a significantly higher occurrence was recorded at evaluation 10 in both systems (p < 0,001; Table 3 ). Frequency of coughing showed an interaction between systems and evaluation periods (p < 0,001), with higher occurrence in the L system at evaluations 1 and 6, and no difference between systems at the other evaluations ( Table 3 ). No animals exhibited labored breathing during any of the evaluations.

The average flight zone was significantly greater in the L system (p=0,023), with a 63% increase compared to the LF system, regardless of evaluation period ( Table 3 ). Flight zone varied significantly between evaluation periods (p < 0,001), with greater distances at evaluations 1, 2, and 3 compared to the others in both systems ( Table 3 ). The number of vocalizations did not differ between systems (p = 0,263) or evaluation periods (p = 0,541; Table 3 ).

The frequency of RS differed significantly between systems (p = 0,036), being 85% higher in the L system than in the LF system, averaged across all evaluations ( Figure 4 ). A significant interaction was observed for RS and evaluation period (p < 0.001), with the highest scores recorded in the L system during evaluations 5 and 6 ( Figure 4 ). Escape speed was similar between systems and significantly lower at evaluations 1 and 8 ( Figure 4 ).

There were no significant differences in ADG and BCS between the systems ( Table 3 ), which can be explained by the similar pasture height in both systems ( Table 1 ). The higher BCS observed at the final evaluations ( Table 3 ) is consistent with the animals’ cumulative weight gain over the 10 months and suggests that, in both systems, the cattle had access to adequate feed, enabling comparable productivity levels.

The animals showed good performance, with an average ADG of 0.875 kg animal-1 day-1 across systems and evaluation periods, indicating that both systems were able to ensure favorable welfare conditions, as insufficient productive performance is typically associated with compromised welfare (Ritter et al., 2019; Mariottini et al., 2022). This relationship between performance and welfare is further supported by the fact that the lowest ADG values were recorded at evaluations 7 to 9, which coincided with higher parasite loads (ticks and botfly), more frequent injuries, and elevated THI values ( Table 3 ). These factors—particularly prevalent in summer—created stressful environmental conditions that likely reduced both animal welfare and graze intake (Herbut et al., 2019), thereby hindering productive performance. These findings agree with those reported by Herbut et al. (2019), who demonstrated a connection between environmental stressors and reduced productivity in dairy cows.

Cattle typically alternate between grazing, ruminating, and resting throughout the day (Souza et al., 2010), with grazing primarily concentrated during the cooler periods of the day: early morning and late afternoon (Broom and Fraser, 1997). Changes in the timing of these activities are often associated with thermal stress conditions (Kilgour et al., 2012; Cortés Fernández de Arcipreste et al., 2018). During periods of intense heat, animals modify their grazing behavior to minimize heat load (Schütz et al., 2010). Reduced forage intake also helps lower internal heat production, since fiber digestion is thermogenic (Ferreira et al., 2006). Consequently, ruminating and resting activities tend to shift to early morning and nighttime hours (Polsky and von Keyserlingk, 2017; Poulopoulou et al., 2019; Souza et al., 2019).

In this study, grazing activity peaked between 2:35 p.m. and 5:30 p.m. (time interval 3), resting and other activities were most frequent from 5:30 a.m. to 9:30 a.m. (time interval 1), and rumination was most common between 9:35 a.m. and 2:30 p.m. (time interval 2; Figure 3 ). These patterns align with the natural behavior of grazing animals (Kilgour et al., 2012). Thus, both the L and LF systems appear to have provided good housing conditions, as further indicated by the longer grazing durations relative to other activities. Kilgour et al. (2012) reported that an average grazing time of around 6.1 hours is adequate; in this study, animals grazed for an average of 6.5 hours across both systems. The availability of ample forage ( Table 1 ) allowed animals to meet their intake needs within a time frame that did not interfere with rumination, social interaction, or idling (Cortés Fernández de Arcipreste et al., 2018). Maintaining regular rumination is essential for cattle’s energy balance (Grant and Dann, 2015), and reductions in rumination can lead to acidosis (Owens et al., 1998), impairing welfare and productivity.

Thermal comfort, as indicated by THI, was similar across systems and exceeded the critical threshold of 71 only at evaluations 8 and 10 ( Table 3 ). Body surface temperature reached its peak at evaluations 6 and 10 and was significantly higher in the L system; however, the average temperature did not exceed 35 °C ( Table 3 ). This is important, as body surface temperature below 35 °C allows thermolysis to occur, establishing a thermal gradient between the body core and surface that enables effective heat dissipation (Collier et al., 2006). Lower ambient temperatures, as provided by the LF system ( Figure 2 ), facilitate this process. Although surface temperatures remained within physiological limits in both systems, animals in the LF system had significantly lower body surface temperature, by about 2.5 °C ( Table 3 ), suggesting better heat dissipation and thermal adaptation in systems with tree cover. Broom (2017) also highlighted the benefits of shade in silvopastoral systems, reporting reductions of up to 4 °C in body surface temperature compared to systems without tree cover.

Throughout the evaluation period, locomotion score in both systems remained at 1, indicating no signs of lameness or mobility issues ( Table 3 ). This is a positive welfare indicator, as movement problems cause pain, discomfort, and fear (Leach et al., 2009b; Bautista-Fernández et al., 2021). Fecal consistency score were close to 2 in both systems while nasal discharge and coughing were infrequent ( Table 3 ), and no animals showed signs of labored breathing during any of the assessments.

Ectoparasite infestation scores for botfly, horn flies, and ticks did not differ between systems but were higher during warmer months ( Table 3 ), which is expected given seasonal parasite dynamics. A corresponding increase in the occurrence of injuries was observed during periods with higher infestation scores ( Table 3 ), likely due to direct damage from the parasites (e.g., myiasis) or self-inflicted trauma from scratching (Morel et al., 2017; Rashid et al., 2018).

Contrary to findings by Murgueitio and Giraldo (2009), who reported lower parasite loads in silvopastoral systems because of increased biodiversity and the presence of natural predators, this study found no such advantage in the LF system. This suggests that other local or system-specific factors may influence ectoparasite dynamics more strongly than tree cover alone.

Handlers’ actions influence animal behavior during handling. It is well established that aggressive handling negatively affects animal welfare, increases fear toward humans, and makes animals more reactive (Grandin, 1997; Hemsworth, 2007; Hemsworth and Coleman, 2010). Flight zone and escape speed are indicators of the quality of the human-animal relationship: shorter flight zone and lower escape speed reflect better interaction between handlers and animals (Hemsworth et al., 2000; Welfare Quality, 2009; Grajales-Cedeño and Da Costa, 2024).

In this study, flight zone was 63% greater in the L system compared to the LF system ( Table 3 ), suggesting that the presence of trees contributed to a better human-animal relationship. Mancera and Galindo (2011) and Ocampo et al. (2011) reported that fear of humans may decrease when animals are partially concealed, which helps improve human-animal interaction. In this context, animals in the LF system—exposed to an environment that allowed partial concealment—appeared to show reduced fear responses toward humans. Additionally, the reactivity score was 85% higher in the L system compared to the LF system ( Figure 4 ), indicating that animals in the latter system were better able to cope with and adapt to their environment. It is important to highlight that animals in both systems were handled calmly and quietly, in accordance with the understanding that handler attitudes are the primary determinant of handling quality (Grandin, 2016).

There is a direct relationship between reactivity score and escape speed: stressed cattle typically show higher reactivity and leave the chute more rapidly (Grajales-Cedeño and Da Costa, 2024). However, as shown in Figure 4 , instances in the P system where a reactivity score of 5 was recorded coincided with slower escape speed. This may be explained by the intense tension experienced by highly reactive animals, which can lead to paralysis and muscle tremors (Da Silva et al., 2024), ultimately resulting in slower escape speed.

The results related to appropriate behavior and good housing (in terms of body surface temperature) show that integrating livestock-forestry facilitated the expression of more adaptive behaviors. This highlights the understanding that domestication deprived animals of resources beneficial to their behavior, and environmental enrichment increased welfare indices. The LF system reduced flight zone by 63%, suggesting greater tolerance to human presence, and decreased reactivity score by 85% compared to the L system. These findings highlight the behavioral plasticity of cattle fostered by domestication, enabling them to respond differently in tree-integrated systems. These results corroborate those of Améndola et al. (2016), who found that tree-covered systems enhance herd social stability and reduce aggressive behaviors, thereby improving animal welfare. Nevertheless, it is worth emphasizing that, for the principles of good feeding, good housing (in terms of rest and rumination time), and good health, both systems were effective in providing conditions that enabled animals to achieve welfare. This suggests that cattle can adapt to different environments (Pacheco et al., 2020; Martin et al., 2021; Santos et al., 2022; Da Silva et al., 2024).

Based on these findings, both systems were capable of supporting cattle welfare. However, caution is needed in generalizing, since both livestock systems in this study provided the following: appropriate handling; feed supply pasture according to demand; clean, fresh, and freely accessible water; stocking densities that allowed equitable access to pasture resources; ongoing health monitoring and veterinary supervision; facilities that reduced fear and anxiety and allowed natural behavior expression. Moreover, the experimental site’s subtropical climate—with mean temperatures ranging between 12.5 and 22.5 °C—supported thermal homeostasis. Therefore, it is emphasized that changes in production systems or the use of other technologies must be combined with meeting the basic needs of animals already known in the literature.

The growing demand for livestock systems that combine sustainability with animal welfare is evident (Nalon et al., 2021). Welfare is a fundamental component of sustainability and plays a crucial role in both productivity and consumer acceptance of animal products (Alonso et al., 2020; Del Campo et al., 2024). In this regard, the silvopastoral system provides numerous advantages: increased biodiversity (Sales-Baptista and Ferraz-de-Oliveira, 2021; Kinneen et al., 2023), improved soil fertility and hydraulic properties (Liu et al., 2024; Romanoski et al., 2025), better management of invasive plant species (Munaro et al., 2023), income diversification (Röhrig et al., 2020), lower thermal stress and improved reproductive performance (Huertas et al., 2021; Lemes et al., 2021), as well as a potential strategy to mitigate greenhouse gas emissions (Portugal et al., 2023). This study contributes new information into the interaction and adaptation of beef cattle to pasture environments with and without tree cover, and the effects on welfare and productivity.