Silvopastoral systems enhance sustainability in the Yucatán Peninsula by improving ecosystem health and animal welfare

Ruminant grazing is a predominant agricultural practice across rural landscapes worldwide, occupying nearly 60% of global agricultural land and supporting around 1.5 billion cattle and over 600 million sheep and goats, which supply approximately 10% of global beef production and about 30% of the world’s sheep and goat meat^1,2. Grazing occurs across a wide range of production systems in nearly all global ecosystems³. However, as people depend on grazing lands for essential ecosystem services and livelihoods, land degradation can lead to significant ecological and societal impacts. Due to their fragile nature, rangelands are particularly vulnerable to mismanagement, which can result in loss of biodiversity, reduced water retention capacity, elevated carbon emissions per unit of product, and diminished productivity¹.

Land degradation poses a threat to many grazing areas worldwide, particularly in tropical regions. Population pressure and practices promoted by the Green Revolution in the 1950s and 1960s have progressively increased the dependence of pastoral systems on monocultures. This has resulted in the establishment of monocropped pasture systems, which have been associated with increased greenhouse gas (GHG) emissions per unit of product, poor animal welfare and reduced productivity, loss of biodiversity, higher zoonotic risk, and decreased soil quality⁴.

In tropical Latin America, a region with one of the highest rates of deforestation, the land area allocated to extensive grazing has grown steadily over recent decades⁵. Silvopastoral systems (SPS) have been proposed as an alternative to minimize land degradation and mitigate GHG emissions⁶. SPS are agroecosystems that combine forest production (e.g., wood, fodder, fruit) with livestock production within the same farm^7,8,9. The most productive SPS are semi-intensive systems that utilize shrubs and trees with leaves that provide good nutrition as well as grasses. The potential of SPS to enhance ecosystem services is linked to environmental benefits and improved system stability in response to climate change¹⁰. It has been documented that SPS in the dry tropics enhance the provision of ecosystem services when compared with monoculture systems¹¹, promote sustainable performance¹² including increased biodiversity¹³, better animal welfare^14,15 and enhanced resilience to climate change¹⁶. However, many of these studies focus on isolated aspects, overlooking system interactions and lacking a comprehensive approach. Additionally, research is often limited to single sites, hindering the generalization of findings, which limits the development of broadly applicable recommendations, especially in a region with a great diversity of SPS.

Despite the benefits of SPS, cattle farmers are reluctant to integrate them into their operations¹⁷. Several barriers such as financial constraints, knowledge gaps, sociocultural factors, labour shortages, unclear land tenure, market dynamics and general risk aversion among farmers impede the adoption of SPS¹⁸. To facilitate the adoption of sustainable practices, more data are needed on the measurement of both positive (synergies) and negative (trade-offs) externalities at the farm level¹⁹.

Thus, this study aims to assess methane emissions, soil quality, animal welfare, and biodiversity in monoculture and silvopastoral cattle systems in the seasonally sub-humid tropics of the Yucatan Peninsula. We found that silvopastoral systems emitted less methane per unit of dry matter intake, improved animal welfare indicators, enhanced soil fertility and organic matter stabilization, and were associated with lower abundance of rodents susceptible to West Nile virus compared with monoculture systems. These findings suggest that silvopastoral systems improve ecosystem health and associated ecosystem services and support the transition towards more sustainable livestock production systems in tropical regions. This information provides a more comprehensive understanding of agroecological interactions and supports evidence-based decision-making and the development of policies aimed at promoting more sustainable livestock systems^20,21.

Externalities in the monoculture and silvopastoral systems

We found externalities in eight of the 23 indicators (p < 0.05; Fig. 1). These included greenhouse gas emissions (CH₄ kg⁻¹ DMI), soil quality (N, P, and BSR), animal welfare including health (skin temperature and THI), biodiversity status (rodent relative abundance), and relative abundance of WNV-host rodents (zoonosis risk). Student’s t-test for the indicators of ecosystem services and biodiversity in the two grazing systems is presented in Table 1.

Externalities in greenhouse gases

Cows in SPS emitted 22% less methane per kg of DMI compared to those in the monoculture system (24.0 ± 0.3 vs 30.7 ± 0.34 g CH₄ kg⁻¹ DMI; p < 0.001), consistently across seasons. In both systems, methane emissions were higher during the rainy season (p < 0.001). However, when expressed per unit of milk produced, emissions were higher in SPS than in MO (63.65 ± 23.0 vs 49.95 ± 13.39 g CH₄ L⁻¹ milk; p < 0.001), reflecting differences in productive performance between systems.

DMI was higher in SPS than in MO (15.46 ± 0.62 vs 11.97 ± 0.38 kg day⁻¹; p < 0.001), representing an increase of approximately 29%, with no seasonal variation detected. The contribution of forage to DMI was greater in SPS (75–80%) than in MO (55–60%). Crude protein intake (CPI) and metabolizable energy intake (MEI) were higher in SPS than in MO (p < 0.001), with MEI peaking during the dry season. Intake of Leucaena leucocephala in SPS was higher during the dry season (0.82 kg DM animal⁻¹) than during the rainy season (0.23 kg DM animal⁻¹). Dry matter intake (DMI), crude protein intake (CPI) and metabolizable energy intake (MEI) were estimated for each system and season based on feed intake and nutrient composition; results are presented in Supplementary Table 1.

Externalities in soil quality

According to the Official Mexican Standard²², the soil at both sites (SPS and MO) had very high C concentration (129 and 117 mg per gram, respectively), and low total N (<20 mg kg). Significantly higher soil N and P concentrations were found in SPS compared to MO (by 27% and 40%, respectively), while no significant differences were observed in SOC concentration. Soil basal respiration was 20% lower in silvopastoral systems than in monoculture (1.3 vs. 1.6 mg CO₂ g⁻¹ soil; p = 0.03), indicating greater potential CO2 emissions under MO management, likely due to a higher proportion of readily mineralizable soil organic matter derived from grasses in monoculture systems.

Externalities in animal welfare

Climatic variables differed between seasons in both systems (p < 0.001). In MO, ambient temperature was higher during the dry season (30.86 ± 5.22 °C) than in the wet season (25.68 ± 3.86 °C), whereas in SPS, temperature was slightly higher during the wet season (27.68 ± 4.12 °C) than in the dry season (25.88 ± 3.94 °C). Relative humidity increased from the dry to the wet season in both systems, with a more pronounced shift in MO.

THI values reflected these patterns. In MO, THI decreased from the dry season (81.33 ± 4.46) to the wet season (75.75 ± 4.67), whereas in SPS, THI increased from the dry (75.75 ± 4.24) to the wet season (79.15 ± 4.73). Across systems, THI values remained within the moderate–severe range, with no evidence of very severe (>89) or potentially lethal (>98) conditions.

When averaged across seasons, THI was lower in SPS (81.5 ± 0.1) than in MO (84.2 ± 0.1) (p < 0.001), although both systems remained within the severe heat stress category.

Despite these climatic differences, farm-level averages indicated that cows in SPS exhibited lower skin temperature ( − 7%; p < 0.001). No differences were detected between systems for clinical and behavioural indicators of animal welfare. The prevalence of claudication, coughing, nasal discharge and ocular discharge did not differ between MO and SPS (p > 0.05).

Externalities in biodiversity and zoonotic risk

A total of 75 bird species, 19 bat species, and 11 rodent species were recorded (see Supplementary Table 2 for the complete list of species). The Shannon–Wiener diversity index did not differ significantly between SPS and MO systems for either birds or bats. The most abundant bird species in both systems were Columbina talpacoti and Turdus grayi, representing 8.2% and 6.9% of total abundance, respectively (Table 2 and Supplementary Fig. 1a). Among bats, Artibeus jamaicensis was dominant in both systems, accounting for more than half of all captures (Supplementary Fig. 1b).

Rodent communities exhibited significant differences, with MO systems showing a slightly lower diversity index but a substantially higher total abundance (+118%; p = 0.03). Although SPS systems tended to support higher avian and bat biodiversity (α and β diversity), these differences were not statistically significant (Supplementary Fig. 1c). Notably, SPS systems also harboured a greater number of species not present in the MO systems.

Regarding the ecological indicators of zoonotic risk, SPS showed lower richness and relative abundance of bird, bat, and rodent species previously identified as WNV hosts. While most comparisons did not reach statistical significance, the relative abundance of WNV rodent hosts was significantly higher in MO, showing a 148% increase compared to SPS (p < 0.05).

Performance of indicators

Excluding avian host total abundance for WNV, ocular discharge, and bat species richness, 21 of the 23 indicators evaluated exceeded the conventional threshold for small effects (Cohen’s d > 0.2). Among these, 34% showed large effects, 26% medium effects, and 30% small effects (Fig. 2), indicating heterogeneous effect magnitudes across indicators rather than a consistent pattern across all variables.

Assessing the sustainability of livestock systems requires an integrative approach that captures processes occurring across multiple ecological scales. In our view, single indicators rarely reflect the complex interactions linking productivity, ecosystem function, and animal health. Therefore, identifying which indicators are most sensitive to management change is essential to understanding how grazing systems influence both environmental integrity and animal well-being. The following sections present and discuss the performance of key indicators grouped by thematic domain, highlighting how their nature and scale affect their responsiveness to different production contexts.

Although the expansion of commercial agriculture is considered the main driver of deforestation, subsistence agriculture still plays an important role in the loss of forest cover in many regions²³. Sustainable intensification of forage-based farming systems promises not only to improve the productivity of tropical forage-based systems and local livelihoods but also to reduce the carbon footprint of livestock production²⁴.

Under the conditions evaluated in this study, we observed that the adoption of an SPS has the potential to reduce enteric CH₄ emissions per kg of DMI by 22%, consistent with previous reports for tropical livestock systems²⁵. Studies in the region have identified that the decrease in methane emissions following intensive silvopastoral system are due to an increase in intake of forages of higher quantity and quality available to animals, and secondary metabolites (e.g. tannins, saponins) present in the shrub and tree legumes present in SPS systems¹⁶ Under controlled experimental conditions, methane mitigation of approximately 20% has been reported in cattle fed tropical grass-based diets supplemented with Leucaena leucocephala forage (~30% of DMI) in respiration chamber trials^26,27. L. leucocephala seems then a highly promising legume for a massive methane mitigation effort in cattle systems in the Global South since it is present in several agroecological regions in South America, Central America and the Caribbean, Africa and Asia. In the present study, L. leucocephala was present as part of the vegetation structure of the SPS; however, methane mitigation likely results from multiple interacting factors within the system, including forage quality, diet composition, and overall system structure.

We explicitly acknowledge that methane measurements were conducted on a single representative farm per production system. Consequently, these findings should be interpreted cautiously as indicative system-level trends rather than as fully replicated experimental comparisons. Given that enteric methane emissions are highly sensitive to animal physiology, diet composition, and specific management practices, these outcomes should be viewed as contextual baseline observations associated with the evaluated commercial systems rather than broadly generalizable regional estimates for all silvopastoral and monoculture frameworks. Nevertheless, the close alignment between our real-world commercial data and the controlled respiration chamber literature^26,27 reinforces the biological plausibility and mitigation potential of SPS under regional management conditions.

Beyond their potential effects on greenhouse gas emissions, silvopastoral systems may also influence below-ground processes and nutrient cycling, making soil quality another key component of sustainability. Because soil health underpins the long-term productivity of grazing systems, understanding how land-use practices modify soil properties is essential for identifying management strategies that enhance ecosystem functioning and productivity²⁸. Soil functioning, however, emerges from complex interactions among biological, chemical, and physical properties, as well as soil–plant interactions²⁹.

In grazing systems, production practices are an important factor in modifying soil fertility³⁰. Compared to MO, the high total N concentration in SPS soils, may be the result of changing fertilisation patterns with N sources, including urine from grazing animals³¹. In that sense, SPS showed a higher grazing intensity, resulting in a higher formation of bovine urine patches, which are characterised by high N concentrations (500-1000 kg/ha) that promote the formation of NH4 + and NO3 by mineralization and nitrification³². Furthermore, trees of Leucaena l. and Guazuma ulmifolia, produce a considerable amount of dry matter and accumulate high amounts of N in short periods of time³³, and N-fixation trees as L. leucocephala specie incorporate high amounts of atmospheric N³⁴.

An association between SPS and higher soil total P concentration was observed. Although baseline conditions prior to system establishment were not available, this pattern is consistent with previous evidence³⁵, In dry and subhumid regions, the canopy plays an important role in the capture and deposition of ash³⁶ derived from frequent forest fires in the Yucatan Peninsula for land-use-change^37,38 Unlike other nutrients (e.g. N), P cannot be replenished biologically and, therefore, the ability of the forest canopy to increase P deposition in these systems generates an important relationship between vegetation and P availability. Therefore, canopy loss in these forests could result in a permanent shift to a low vegetation cover state, whereby the reduced supply of P from atmospheric deposition inhibits the regrowth of a dense forest canopy³⁹. In this sense, it has been observed that the incorporation of trees in paddocks could improve P storage in the soil-plant system⁴⁰.

Although no statistically significant differences in soil carbon concentrations were detected between the MO and SPS systems, higher values were consistently observed in the latter. This trend likely reflects the greater standing biomass, and less decomposable forest litter (leaves and woods), and coarse and fine roots, which may enhance carbon stabilization in the soil, as was reflected by the lower C mineralization in BSR experiments. Overall, our study suggests a tendency toward increased carbon stabilization under SPS, that could increase the soil C sequestration under expected scenarios of drought intensification^41,42,43.

Finally, variation in grazing system age—from 3 to 26 years for silvopastoral farms and from 10 to 30 years for monoculture farms, should be considered when interpreting these results. Silvopastoral systems evolve gradually, as tree growth, litter deposition, and root activity progressively modify soil structure and nutrient cycling. Consequently, differences in soil quality indicators may partly reflect these temporal dynamics, with more recently established systems still in earlier stages of below-ground stabilization. Acknowledging this temporal dimension is essential for accurately interpreting soil responses under real-world management conditions.

Although soil quality contributes to the long-term functioning and productivity of grazing systems, sustainability also depends on the capacity of these systems to provide adequate living conditions for animals. Accordingly, animal welfare represents a central component of livestock production sustainability⁴⁴.

Regarding welfare assessment, the indicators recommended by the Welfare Quality® protocol (prevalence of lameness, coughing, nasal discharge, and ocular discharge) were not statistically different between groups, although the effect size (Cohen’s d) indicated a noticeable and potentially meaningful trend. The similarity in results may reflect comparable management practices across farms in the region, including the use of animal health records, vaccination programs, and routine herd management activities¹³. However, as the present study did not directly assess farmer perceptions or veterinary decision-making, these factors should be considered possible explanations rather than confirmed drivers of the observed patterns.

We observed that MO are associated with higher THI and thus higher risk heat stress. These results are consistent with previous reports⁴⁵. In SPS, cows show lower body temperature likely associated with the presence of trees providing shade. The shade from these plants lowers the temperature and wind speed, making the pasture more comfortable compared to pastures without trees. These conditions could partially explain the higher dry matter intake (DMI) observed in SPS⁴⁶, although this relationship was not directly assessed in the present study. Additionally, other environmental and management factors were not controlled in this study, and therefore the role of heat stress should be interpreted as a potential mechanism supported by previous literature rather than a direct causal effect.

While animal welfare reflects the capacity of livestock systems to provide adequate conditions for domestic animals, biodiversity indicators capture broader ecological responses to land-use management. A frequent consequence of the implementation of SPS is an increase in biodiversity compared with monoculture systems¹¹. While several studies have shown that different types of silvopastoral arrangements can increase landscape connectivity for wildlife species (e.g. birds, bats, and rodents), at the farm level the effects of SPS on biodiversity are variable⁴⁷. Although it is not clear which aspects of silvopastoral systems favour increased biodiversity, the quality of the landscape matrix has been identified as a key condition for promoting biodiversity and aiding conservation^48,49,50.

Although in our study the landscape matrix was considered in the classification of the production systems assessed, we did not include landscape metrics as an indicator of the sustainability of these systems. The reason for this limitation was that some MO farms were surrounded by mature forests. Despite a tendency for SPS to increase avian and bat richness and abundance, these differences were not statistically significant. The effect of the structure and matrix of the agricultural landscape may have influenced why, in our study, we did not find a significant difference in the richness, total abundance, and diversity of birds and bats. These results are consistent with those reported in a meta-analysis that determined that although SPS represent a clear benefit for biodiversity compared with treeless grassland, SPS benefits plants and invertebrates most, while more mobile and landscape-dependent taxa (e.g. birds, mammals) benefit less⁵¹. However, the monitoring of these taxa is important. Mathematical modelling has also indicated that changes in the bird community, especially the increase in key reservoir species, could increase the risk period for WNV transmission by up to 6-fold and increase the daily risk by 40%⁵².

We observed that MO had a significantly higher abundance of rodents overall, as well as a higher abundance of WNV-host rodents. This phenomenon, recognized as a recurring pattern in several infectious diseases, is observed in highly defaunated and transformed ecosystems (i.e., monocultural systems) and may influence ecological conditions associated with disease transmission. This may be important, given that the effects of landscape transformation, especially deforestation, can lead to altered patterns of viral transmission and spread of WNV⁵³.

Although the presence of WNV in birds, rodents, and bats has been described previously in Yucatan^54,55,56, no human cases of WNV have been reported in the Yucatan peninsula since 2007, but it should be noted that ecological conditions for the maintenance and transmission of WNV prevail⁵⁷.

Collectively, these findings underscore the multidimensional nature of livestock sustainability and the challenge of identifying metrics capable of adequately capturing system performance. Although numerous indicators have been proposed, their sensitivity and consistency often vary across studies and contexts. Integrating multiple complementary metrics may provide a more comprehensive assessment of grazing systems. However, technical and financial constraints frequently limit monitoring efforts, making it necessary to prioritize indicators that are particularly responsive to management-driven changes.

In this study, we evaluated the performance of 23 environmental sustainability indicators across two pasture-based production systems (MO and SPS). Each indicator was measured using its respective methodology and therefore carries its own limitations. While the adoption of alternative grazing systems is commonly evaluated using null hypothesis significance testing (NHST; e.g., t-tests or ANOVA), these approaches can be influenced by sample size and may not fully capture the magnitude of differences between systems. Effect size metrics provide complementary information by quantifying the magnitude of observed differences independently of statistical significance.

Using effect size estimates, we observed that 20 indicators exceeded the conventional threshold for small effects, suggesting varying degrees of sensitivity to differences between production systems. Among these, eight indicators (BSR, P, RrWn, RaWn, N, Ra, Cla, CH₄) showed comparatively larger effect sizes, indicating that they may be particularly responsive to differences between systems within the context of this study. These indicators may therefore represent promising candidates for monitoring system-level changes across the thematic dimensions evaluated.

Interpretation of these results should consider factors that commonly influence comparative studies of grazing systems, including the spatial scale of the study areas and the duration of system establishment. Indicators related to animal populations may require larger areas under consistent management to reveal system-level differences, whereas soil-related indicators often reflect cumulative processes that emerge over longer time scales.

Future research should aim to disentangle the relative contribution of specific system components (e.g., forage composition, microclimate, and management practices) to the observed patterns, as well as incorporate longitudinal and landscape-scale analyses to better capture system dynamics, particularly in relation to restoration trajectories in managed grazing landscapes.

Tropical regions account for a major share of global cattle production, yet field-based assessments of sustainability have largely focused on humid areas, overlooking seasonally sub-humid tropical regions that cover a greater land surface and are undergoing rapid ecological change. In our multidimensional analysis of grazing systems in the Yucatán Peninsula, we found system-level differences in several indicators of sustainability between monoculture and silvopastoral management under the evaluated real-world conditions. Furthermore, these systems led to significant shifts in host community composition, which may influence the ecological dynamics of potential zoonotic risks compared to conventional monocultures. Drawing on 23 integrated indicators of livestock sustainability, our analysis suggests that management decisions within tropical grazing landscapes are associated with changes in environmental, productive, and welfare indicators reflecting the inherent operational variability of commercial farms. Unlike previous studies that examined these dimensions separately, this work provides an integrated, empirical assessment of silvopastoral performance under real, seasonally sub-humid tropical conditions. Within this context, silvopastoral systems may contribute not only to production but also to the recovery of ecological structure and function in degraded systems. Taken together, our results suggest that these systems can form part of evidence-based approaches to reconcile livestock production with biodiversity conservation, climate mitigation, and landscape restoration, while acknowledging the need for further longitudinal and multi-scale analyses.

Site description and farm selection

The study was conducted in nine dual-purpose cattle pasture-based systems in the Yucatan Peninsula of Mexico (Fig. 3). The climate is warm, with a marked seasonal distribution of rainfall. Average temperatures range from 24–28 °C, with an average maximum temperature of 36 °C in the hottest quarter of the year and an average minimum of 16 °C in the coolest quarter. Mean annual precipitation ranges from 800 mm (semi-arid climate) to 1200 mm (sub-humid climate), concentrated (76% of the total annual amount) in a rainy season⁵⁸. The dominant native vegetation in this seasonally dry tropics is representative of the tropical dry forest biome⁵⁹. The main land use is agricultural, occupying 67.3% of the surface area⁶⁰.

The farms were selected to exemplify one of the following grazing systems: 1) Monoculture systems (MO), characterised by grazing on artificially created grasslands devoid of trees; and 2) Silvopastoral systems (SPS), established systems that incorporate a variety of vegetation strata (grassland, shrubs, and trees) under integrated grazing management.

Farms were selected as representative examples of regional production systems rather than through a strict matching or paired experimental design. Although all farms corresponded to dual-purpose pasture-based cattle systems operating under broadly similar regional climatic conditions, variation in structural and management characteristics among farms reflected the heterogeneity typical of commercial livestock systems in the region. To improve transparency regarding comparability among farms, descriptive variables including grazing area, livestock units, and stocking rate were incorporated into Table 3.

Conceptual framework of indicators

A total of 23 sustainability indicators were selected to evaluate: 1) the optimal maintenance of the basic components of the livestock system (enteric methane emissions and soil quality); 2) animal welfare including health; and 3) biodiversity and zoonoses risk in the grazing landscapes: species richness and relative abundance of wildlife. Indicators were chosen to represent core ecological processes operating across multiple spatial scales in livestock systems, including greenhouse gas dynamics, biogeochemical cycling, animal physiological responses, and community-level biodiversity patterns. This multidimensional framework follows ecosystem service and agroecological integration approaches that prioritize system-level externalities and cross-scale interactions rather than isolated performance metrics (Fig. 4).

The selection of indicators was based on previous work in the region^13,54,61,62 as well as expert knowledge. The complete list of variables and their abbreviations is presented in Supplementary Table 3. The following sections provide their definitions and the methods used to measure them.

This study was designed as a cross-sectional assessment of existing grazing systems, capturing system-level conditions at a specific point in time rather than temporal dynamics. The experimental unit differed according to the indicator evaluated. For soil and biodiversity indicators, the farm was considered the experimental unit (n = 9 farms). For methane emissions and animal-based welfare indicators, measurements were collected at the individual animal level, with animals nested within farms. To avoid pseudoreplication, individual measurements were aggregated at the farm level prior to statistical analyses, and farm-level means were used as the analytical unit for system-level comparisons. A completed ARRIVE 2.0 Essential 10 checklist is provided as Supplementary Table 4.

The statistical analyses were designed to compare indicators between production system types. Accordingly, farm-specific characteristics were not included as explanatory variables in the analyses. Although these structural variables are presented descriptively in Table 3, they were not included as covariates in the primary statistical models due to the limited sample size at the farm level (n = 9). Consequently, the observed differences should be interpreted within the context of real-world variability among production systems.

Sampling and analyses of environmental indicators

Enteric methane emission (CH₄)

A total of 50 milking cows were selected (MO, n = 34; SPS, n = 16) for CH₄ estimation during milking (07:00 to 09:00 h) for 10 consecutive days, as previously described in Flores-Coello et al. ¹⁶. A period of two weeks was allowed for acclimatization to the headboxes before measurements were collected. Daily methane emissions per kg of dry matter intake (DMI) (g∙kg) were calculated based on the equations reported by Garnsworthy et al. ⁶¹.

CH₄ emissions were adjusted per kg of dry matter intake per day and per litre of milk. The dry matter intake (DMI) was estimated by subtracting the initial available DM (dry matter) and residual biomass, divided by the total number of cows grazing in each paddock in both systems. Milk yield data were collected over 10 consecutive days in each season.

Forage availability and dry DM yield were estimated at the paddock level using destructive sampling after grazing. In the monoculture system, seven random quadrats (0.25 m²) were sampled, whereas in the SPS, five larger plots (2 m²) were used to account for vegetation stratification. Samples were oven-dried (55 °C, 72 h) to determine DM content. Supplementary feeds were sampled and analysed for chemical composition (CP, ash, NDF, ADF), and metabolizable energy was estimated using published equations according to Flores-Coello et al. ¹⁶. Measurements were conducted on one representative farm per production system.

Soil quality

In each farm (n = 9), three plots (10 × 50 m) were established. In each plot, five sampling points were located at 10 m intervals along a transect. Topsoil samples (0–10 cm depth) were collected at each point and composited to obtain one representative sample per plot. This sampling design allowed us to account for within-site spatial variability and provided three replicate samples per farm. This sampling design was used to account for within-site spatial variability. All soil analyses were done following the methodology described by Flores Coello et al. ¹⁶. Soil organic carbon (SOC) concentration was determined using the high-temperature incineration method in a Shimadzu 505 model C autoanalyzer. Nitrogen and total phosphorus in soil were determined by acid digestion in concentrated H₂SO₄ with the Kjeldahl method and determined colorimetrically in the Braun Luebbe automated system

Basal soil respiration (BSR) was estimated by quantifying the carbon dioxide (CO₂) released in the process during microbial respiration during 35 days of incubation. For this purpose, 50 g soil samples were placed in 600 ml glass containers with tight-fitting lids, together with a smaller flask containing 10 ml of 1 N NaOH to capture the released CO₂. CO₂ was determined by titration with 1 N HCl, after precipitation of barium carbonate formed by the addition of barium chloride (BaCl₂) in aqueous solution to the NaOH solution. Phenolphthalein diluted in 100 ml ethanol (60%, v/v) was used as an indicator.

Animal welfare

The prevalence of claudication (Cla), coughs (Cou), nasal discharges (Nad) and ocular discharges (Ocd) was evaluated in each farm (n = 9) using a modified Welfare Quality protocol⁶³. The temperature–humidity index (THI) was used as a heat stress proxy for cattle and was calculated as follows⁶⁴:

Ambient temperature and relative humidity were recorded using an automatic weather station (Davis Vantage Pro2®, Davis Instruments Corp., USA) installed on-site. The THI was categorized to interpret the risk of heat stress experienced by the animals using the scale of Moran and Doyle (2005)⁶⁵: <72 (no heat stress), 72–78 (moderate heat stress), 78–89 (severe heat stress), 89–98 (very severe heat stress), and >98 (potentially lethal heat stress conditions).

Additionally, cow skin temperature was recorded using infrared thermography as a non-invasive method to detect thermal variation associated with environmental conditions⁶⁶. Thermal images were obtained with a FLIR E5® infrared camera (Teledyne FLIR LLC, Oregon, USA). Three thermal images per animal were obtained daily at three time intervals (08:30–12:00, 13:00–16:30, and 16:50–18:40). Images were captured at an approximate distance of 1.5 m, directing the centre of the camera toward the left flank between the 6th and 9th ribs at the midpoint of the thoracic cavity. To reduce measurement variability, all images were taken by the same previously trained observer. Infrared thermography has been widely applied in animal physiology and agroecological research to characterize thermal patterns and their relationship with microclimatic conditions in tropical livestock systems⁶⁷.

Biodiversity and zoonotic risk

Resident birds, bats, and rodents were sampled in both grazing systems (n = 9). Birds were sampled using eight mist nets (12 m long × 3 m high) at each site. The mist nets were collocated in pairs at four capture points separated by at least by 50 m. Mist nets were activated before dawn for four hours, and the sampling lasted three consecutive days. Birds were identified to the species level^68,69 and were released at the site of capture after being handled. Migratory birds were excluded to avoid temporal and spatial biases and to better represent the ecological and epidemiological patterns of resident species. Although mist nets catch some species more than others, the captures provide a comparable indicator of bird biodiversity for the two conditions. Bats were mist-netted in the same way as birds. We operated the mist nets for 5 hours from sunset for three consecutive nights. Once captured, each bat was kept individually in a cloth sack to obtain the morphometric values necessary for its taxonomic identification. Bats were identified at the species level with field keys⁷⁰. Rodents were sampled using 100 Sherman traps (75 × 230 × 90 mm) placed on transects, each trap spaced at 10 m intervals. The traps were baited with a mixture of oats, peanut butter, and vanilla extract and activated at sunset the previous day and checked at sunrise. The rodent survey lasted for five consecutive nights. Captured rodents were identified at the species level based on available literature⁷¹.

For each taxonomic group, species abundance, α-diversity (species richness), γ-diversity (the total number of species found in all plots), β-diversity (β = γ/α), diversity (estimated by calculating the Shannon–Wiener index), and the number of species shared and unique to both grazing systems were also calculated.

To evaluate ecological indicators of zoonotic risk, we quantified the proportion of host species (birds, rodents, and bats) in terms of richness and relative abundance across grazing systems. West Nile virus (WNV) hosts were used as a focal model due to their sensitivity to land-use change⁵⁴.

Trapping and handling wildlife were approved by the Ministry of the Environment (SEMARNAT permission SGPA/DGVS/001355/18) and by the Subcommittee for the Care and Use of Experimental Animals of the Faculty of Veterinary Medicine at UNAM- Mexico (protocol MZ-2016/2-4). Throughout the study, cattle remained under routine farm management and no experimental manipulations beyond observational and non-invasive measurements were performed. All procedures were conducted to minimize animal disturbance. No anesthesia, sedation, analgesia, or euthanasia was required, as all cattle assessments were non-invasive.

Statistical analyses

To compare the externalities of monoculture and silvopastoral systems, the approach used by Amorim et al. and Márquez Hernández et al. ^10,72 was used. Student’s t-tests for independent samples were used to examine the differences in environmental indicators among grazing systems. Tests were performed for each indicator, taking the type of grazing system as a factor, considering the nature of the data. Normality was assessed using Shapiro–Wilk tests and homogeneity of variances using Levene’s test. When assumptions were violated, appropriate transformations were applied.

For those indicators that were statistically significant (p < 0.05) the relative change (RC) was calculated using the formula:

Positive RC values indicate higher values under monoculture relative to silvopastoral systems; interpretation as beneficial or detrimental depends on the ecological meaning of each indicator.

To characterize patterns of biodiversity within and between grazing systems, we calculated multiple diversity metrics. Alpha diversity (α) was defined as the cumulative number of species recorded within each system, while gamma diversity (γ) represented the total number of species observed across all fields. Beta diversity (β) was calculated as β = γ/α at both the field level and for the dataset⁷³. The number of species shared between systems and those unique to each system was determined. Overall diversity was further evaluated using the Shannon–Wiener index (H′). Additionally, the Olmstead–Tukey corner test⁷⁴ was used to evaluate shifts in species dominance and occurrence between systems.

Performance of indicators

Recognising that null hypothesis significance tests (NHST) p-values are strongly related to sample size and magnitude of effects, and that these procedures are insufficient to assess the results in practice, we calculated Cohens-d as method to validate the indicators by effect size. Cohens-d standardises the difference between two means and can be employed in two-group designs, where the interest is in the relative strength of the differences between the means of two populations from the sample data. Cohen’s-d magnitude is expressed as the number of standard deviations that separate the two groups⁷⁵. The values obtained for Cohen’s-d were interpreted based on the established method in which values of 0.2, 0.5, and 0.8 indicate a small, medium, or large size effect, respectively⁷⁶. All statistical analyses were conducted in R v.4.3.2⁷⁷ using the base R environment without any additional packages.