EVALUATION OF THE SUSTAINABLE PERFORMANCE OF NATIVE AND INTENSIVE SILVOPASTORAL SYSTEMS IN THE MEXICAN TROPICS USING THE MESMIS FRAMEWORK † MARCO MESMIS]

a tool to assess sustainability criteria in agroecosystems. Objective. To use MESMIS to compare the sustainable performance of NS, IS and MS and determine the system with the best sustainable performance in the Mexican Tropics. Methodology. One MS IS and NS per municipality (Tizimin, Merida and Tzucacab) were evaluated in the state of Yucatán, Mexico. Based on the MESMIS approach, the evaluation of the critical points of sustainability resulted in the selection of 19 indicators classified according to the attributes also defined by MESMIS (production, adaptability, stability-resilience, equity and self-management) and by sustainability dimensions (environmental, animal welfare, economic and social). After evaluation, indicator scores were obtained and integrated into attributes and dimensions through the assignation of equitable, balanced weights (W). Finally, attribute and dimension scores were aggregated in amoeba graphs to facilitate visual interpretation. Results. NS were better for the dimensions ‘Environmental’ and ‘Economic’ and the attributes ‘Stability, ‘Reliability’and ‘Resilience,’ and ‘Production’. IS were best for the dimension ‘Animal Welfare’ and attributes ‘Adaptability’ and ‘Self-reliance’. MS were better for the ‘Social’ dimension and the ‘Equity’ attribute. Implications. The fact that IS appeared to be more sustainable than MS does not leave out the idea of considering NS as a better option for some criteria such as the biodiversity conservation and the prevention of disease outbreaks, than IS. We suggest that more studies are carried on areas of potential improvement for IS as well as NS. Conclusions. This information will be useful to continue working on the parametrization of sustainability criteria of cattle extensive systems to be used for more efficient policies. fuera la idea de considerar a los NS como una mejor opción en algunos aspectos en comparación de IS; por ende, sugerimos que se lleven a cabo más estudios en áreas de mejora potencial para los IS y para los NS. Conclusiones. Esta información será útil para continuar trabajando en la parametrización de criterios de sostenibilidad para el ganado en sistemas extensivos, con el objetivo de ser usado en políticas más eficientes.


INTRODUCTION
In the last 30 years, livestock production has transitioned from complex natural systems to monoculture systems (MS) and it occupies 70% of all agricultural land and 30% of the land surface of the planet (Steinfeld et al., 2006). MS represents a threat to multiple ecosystem services and contributes to global environmental problems such as greenhouse gases emissions (Gerber et al., 2010), deforestation and deterioration of water sources (Herrero et al., 2010). Likewise, MS reduces genetic diversity, induce soil degradation and are susceptible to agricultural plagues (Gliessman, 2014), making them incapable of maintaining complex ecological interactions (Lamb, Erskine et al., 2005). Furthermore, MS are associated with economic disadvantage, as they require great external input (Gliessman, 2014, Steinfeld, 2006 and animal welfare problems common for large scale production (Broom, 2016). Therefore, livestock production needs to develop toward sustainability.
Sustainable livestock production has been recognized as a continuing and complex process characterized by the attempt to reduce negative effects on the environment and the increase in the provision of environmental services (Milera, 2013). In this sense, silvopastoral systems (associations of trees and shrubs) represent a good alternative, as they are associated with increased photosynthetic rates, nitrogen fixation, nutrient recycling, biomass production and organic matter in soil , as well as with the provision of animal welfare and the continuation of environmental services, such as carbon sequestration, water preservation, soil rehabilitation and biodiversity conservation (Broom et al., 2013, Murgueitio et al., 2011. In Yucatan, Mexico, although livestock extensive systems are mainly based on MS, native silvopastoral systems (NS), which are pastures with unmanaged native shrubs and trees (Gómez-Cifuentes et al., 2019) are still used to feed cattle during the dry season and represent the main resource for many conventional cattle smallholders (Ramírez-Cancino and Rivera-Lorca, 2010). Likewise, intensive silvopastoral systems, based on the integration of fodder shrubs at high densities (> 10000 plants ha −1 ) and productive pastures and trees (Murgueitio et al., 2011) are also present in the region (Améndola et al., 2016;Mohammed et al., 2016). IS integrate technical knowledge, such as the inclusion of specific plant species to increase the benefits associated with silvopastoralism, whereas NS are not actively managed and more likely to include native flora and fauna (Murgueitio et al. 2011;Nahed-Toral et al., 2013) and may offer more advantages in terms of the provision of environmental services than IS. To support the implementation and preservation of silvopastoral systems, it is necessary to generate information on their strengths and weaknesses by comparing the sustainable performance of NS, IS and MS.
One of the methodologies used to evaluate the sustainable performance of productive systems is the Framework for Assessing the Sustainability of Natural Resource Management Systems (MESMIS; Masera et al., 2000a). MESMIS is a methodology for the cyclic evaluation of sustainability in systems managing natural resources. It was designed as a multidisciplinary, flexible and adaptable tool that encourages the analytic process and provides reliable elements that users can use to give recommendations for the improvement of the evaluated systems. MESMIS comprises the spatial and temporal context of productive systems and the limitations that arise from the interactions between environmental, economic and social elements (Masera et al., 2000b). This framework also allows the comparison of different production systems or systems over time by integrating indicators and putting their evaluated values on a standardized scale and comparing (Masera et al., 2000a). MESMIS is recognized as a framework that offers a good overview of agroecosystems (Gliessman, 2014). Despite its usefulness, MESMIS has never been used to compare the sustainability performance of NS, IS and MS.
We evaluated nine farms in Yucatan, Mexico using the MESMIS framework. Our study aimed to compare the sustainable performance of NS, IS and MS and identify areas of improvement in terms of four sustainability dimensions (Environment, Animal Welfare, Economic and Social) and five MESMIS attributes (Sustainability, Reliability and Resilience, Production, Adaptability, Self-reliance, and Equity). This information is useful to work towards the parametrization of sustainability criteria to elaborate better policies in Mexico and in Latin America.

Study area and evaluated farms
This study was carried out in the state of Yucatán, Mexico. Average annual temperatures in Yucatan range from 24 -28 °C, with an average maximum temperature of 36°C and an average minimum temperature of 16°C. The lowest precipitation values are 500 mm, whereas the highest range from 1200 to 1500 mm (Orellana et al., 2010). The nine farms evaluated belonged to three municipalities located in the state: Tizimin, Tzucacab and Merida. Tizimin (07° 58' N latitude and, 88º 09' 04' W longitude) is in the northeast region of the state, with sub-humid climate, an annual mean temperature of 25.8°C and annual average precipitation of 1084 mm. Tzucacab (19º 38' and 20º 09' N latitude,and 88º 59' and 89º 14' W longitude) is in the south region of the state, with warm sub-humid climate, an annual mean temperature of 25.8°C and annual average precipitation of 1084 mm. Merida (20° 45' y 21° 15' N latitude and 89° 30' y 89° 45' W longitude) is in the northeast region of the state, with warm sub-humid climate, an annual mean temperature of 26.2°C and annual precipitation in the range of 470-930 mm (IFAED, 2016).
In each municipality, three farms were chosen in a 25 km radio to represent one of the following production systems: • Native Silvopastoral System (NS): pastures with unmanaged native shrubs and trees (Gómez-Cifuentes, Gómez et al., 2019). These units were characterized by continuous forest fragments of at least 10 km 2 . The NS farms were: Roble in Tzucacab (NS1); Santa Teresa in Merida (NS2); Xhopel in Tizimin (NS3).
• Monoculture System (MS): conventional grazing system based on monoculture of grass (Mancera et al., 2018). These units were characterized by a homogeneous landscape with induced grass and less than 5% of its surface occupied by forest coverage. The MS farms were: Ramonal in Tzucacab (MS1); UADY in Merida (MS2); Escalera in Tizimin (MS3).
Farms NS1, IS2 and MS2 are double-purpose cattle systems (production of milk and meat). All farms belonged to smallholders working with limited resources in rural areas.

MESMIS methodology description
The MESMIS framework is based in four premises (Masera et al., 2000b): Sustainability is defined by seven general attributes which are measured by indicators: • Production: the system's capacity to provide the required levels of goods and services • Stability: the system's capacity to maintain a constant production • Resilience: the system's ability of returning to its original levels after normal perturbing events • Reliability: system's ability to keep equilibrium during normal perturbing events • Adaptability: system's ability to find new levels of equilibrium after perturbing events. • Equity: system's ability to distribute the generated benefits and responsibilities in a fair manner • Self-reliance: system's ability to control its interactions with its surroundings. The evaluation is performed for a specific system, under specific conditions and scales, being valid only under these conditions: • The evaluation process is cyclic and multidisciplinary, and; • The evaluation needs to be comparative, either between two systems (traditional or alternative) or the same systems across time.
The methodology consists of six main steps (Masera et al., 2000b): 1) Determination of study object, 2) determination of systems' strengths and weaknesses, 3) selection of indicators, measuring and monitoring of indicators, 4) integration of results, 5) conclusions, and 6) recommendations. The determination of the study object consists in the identification of systems' principal components through the socio-environmental context, biophysical components, the spatial and temporal scale and the identification of production systems. The determination of strengths and weaknesses corresponds to the identification of critical points that interact positively or negatively by facilitating or obstructing systems' capacity to withhold through time. The selection of indicators is then performed according to the critical points identified and need to be selected to provide information easily understandable; after selection, they need to be integrated in a pertinent, robust, sensitive and reliable matrix. For measuring and monitoring of indicators, a direct or indirect methodology, which needs to be accessible and replicable, needs to be selected. After measuring, the integration of results compares the outcomes obtained between systems, through the determination of optimal values for each indicator, which needs to be expressed in the scale selected by the evaluator and in the same direction. Results are further integrated in a matrix (by dimension, attribute, and system). Is recommended to express results in radial diagrams known as AMEBA, were each dimension or attribute represent an axis and each value obtained in the matrix, a percentage. Finally, recommendations on the faults and good decisions of systems as well as an opinion on them is given through the visual interpretation of results.

Selection and evaluation of indicators
After identifying strengths and weaknesses in Yucatán livestock systems, nineteen indicators were selected to evaluate sustainability performance in NS, IS and MS. Indicators were positive (+) or negative (-) if the best outcome for sustainability was associated with the maximum or minimum value, respectively. For instance, the indicator 'Productive Diversification' is classified as positive, as the maximum value achievable corresponds to the maximum number of products generated in a single farm. In contrast, the indicator 'Use of Agrochemicals' is negative, as the maximum value corresponds to the absence of agrochemical substances use.

Indicators 'Electric Energy Consumption' 'Fossil Fuel
Consumption' and 'Governmental subsidies and assistance' were calculated as investment per hectare to account for different farm sizes. The indicator species richness was obtained from direct wildlife monitoring performed by research partners, through the capture-recapture of birds, bats and small rodents. Mist nets were used for bats and birds, and Sherman traps were used for small rodents (Domínguez-Meneses, 2018;Dominguez-Hernandez et al., 2018). The indicators Good Feeding, Good Health and Appropriate Behaviour were evaluated using the Welfare Quality® (WQ) protocol for the evaluation of dairy cattle (WQ, 2009). The evaluation of 'Good Housing' was excluded from this evaluation, as this indicator is highly tailored for intensive conditions and does not consider factors such as shaded area and solar radiation, which may be more influential for cattle raised extensively (Mancera, 2011). WQ is recognized as a multidimensional, animal-based welfare assessment tool composed of four principles, twelve criteria and thirty indicators. The indicators used in this study are WQ principles based on the following measurements (WQ, 2009):

Good Feeding
• Body condition: focal sampling based on visual criteria with rear and flanks of animals as reference. Ranks were assigned as follows: 0 = very lean, 1 = regular body condition, 2 = very fat. • Water: evaluation of water provision, cleanliness of water points, water flow, functioning of water points. Good Health • Lameness: evaluated when individuals left the milking parlor using weight bearing, timing and rhythm of steps as indicators. It was ranked as follows: 0 = not lame, 1 = lame, 2 = severely lame. • Integument alterations (injuries, inflammation and alopecia): observation of five areas from of a randomly selected side of the body from 2 m. The areas were: hindquarter, tarsus, flank/side/udder, carpus and neck/shoulder, back. The presence or absence of integument alterations was evaluated in each of these areas. • Presence or absence of each health indicator: nasal discharge, ocular discharge, hampered respiration, diarrhea, vulvar discharge and ectoparasites. • Coughing and sneezing: mean number of coughs/sneezes per animal per 15 min. • Disbudding/dehorning, tail docking: focal sampling of animals. Ranks were assigned as follows: 0 = no dehorning/disbudding, 1 = disbudding of calves using thermocautery, 2 = disbudding of calves using caustic paste, 3 = dehorning of fattening cattle, and; 0 = use of anesthetics, 2 = No use of anesthetics, and; 0 = use of post-surgery analgesics, 2 = no use of analgesics.

Appropriate Behaviour
• Agonistic behaviours: continuous recording of the number of head butts and displacements observed in the herd from a high point of the pen or grazing paddock. • Access to pasture: hours of the day herd spent in the paddock. • Observation of herds with the assignation of qualitative behavioural traits by the observer avoidance distance: Number of cows touched or flight distance when observers approached cows slowly (one step/2 s) with one arm stretched forward at a 45°degree angle to attempt to touch the muzzle.
Obtained data was processed with the software program Welfare Quality® scoring system (available at: http://www1.clermont.inra.fr/wq/index.php?id=simul &new=1), which uses weighted sums, decision trees and Choquet integrals to obtain principles scores (WQ 2009). Indicators, measurement method and source of information for their assessment are summarized in Table 1.

Scoring standardization of qualitative and quantitative variables
Indicators were a combination of quantitative and qualitative variables. Qualitative variables were 'Use of Agrochemicals', 'Workers Development and Training' and 'Organization and Participation'. These indicators were measured by assigning a descriptive state of the indicator to each farm. Each descriptor was assigning to one of five scores (100, 75, 50, 25 and 0), being 100 the most desirable description for sustainable development and 0 the least desirable.
For quantitative indicators evaluated with different measuring scales, a methodology to obtain standardized scoring values was established. Maximum values for each indicator were obtained after evaluation and divided by five. The resulting value was multiplied by factors 2, 3 and 4 to obtain the numerical limits of five intervals between 0 and the maximum value measured for the indicator. These were later associated with one of five intervals for a scoring system ranging from 0 to 100: ( Questionnaires/ records kept in the farm 100: farm has a written business plan, productive records and production costs calculations, as well as positive annual performance and achievement of all set goals 75: farm has at least one of the following: business plan, productive records or production cost calculations. It has a positive annual performance and/or achieved all set goals 50: farm has at least one of the following: business plan, productive records or production costs calculations. It has a positive annual performance or have achieved one of the set goals 25: Farm has written at least one of the following: business plan, productive records and/or production costs calculations. It has a negative annual performance and/or has achieved one of the set goals. 0: No written business plan, productive records or calculations. It has a negative annual performance and none of the set goals has been achieved.  Table 3.

Integration of indicators by MESMIS attribute and sustainability dimension
The sustainability performance of NS, IS, and MS was evaluated through the integration of indicators by sustainability dimension and MESMIS attributes. The sustainability dimensions evaluated were Environment (5 indicators), Animal Welfare (4 indicators), Economic (5 indicators) and Social (5 indicators). The environmental, economic, and social dimensions are included as recognized pillars of sustainability (Hansmann, Mieg et al., 2012). Animal Welfare is considered in this study due to its notable correlation with environmental impact mitigation in the development of sustainable dairy systems (Herzog et al., 2018) and the proven effect that silvopastoral systems have on this attribute (Broom et al., 2013). For the evaluation of MESMIS attributes, Stability, Reliability and Resilience were jointly evaluated as one attribute (5 indicators), as previously done by Nahed et al., (2019) and Dominguez-Hernandez et al., when the same indicators reflect the main characteristics of these traits in evaluated systems. The attributes Production (6 indicators), Adaptability (1 indicator), Self-reliance (4 indicators) and Equity (3 indicators) were also evaluated ( Table 2). For the integration of indicators, equitable, balanced weights (W) were assigned. When integrated by attribute, W values were assigned as follows: Stability, Reliability and Resilience = 0.2; Production = 0.17; Adaptability = 1; Self-reliance = 0.25, and; Equity = 0.33. When integrated by dimension, W values were assigned as follows: Environmental = 0.2; Animal Welfare = 0.25; Economic = 0.2, and, Social = 0.2. After weighing, scores were integrated with the following formula: Attribute/Dimension Score = ∑ (xi * w) Where Xi = indicator assigned to attribute/dimension W = corresponding weight per attribute/dimension Finally, attribute and dimension scores were aggregated in amoeba graphs to facilitate visual interpretation. Figure 1 is a schematic representation of the MESMIS system.

Evaluation of indicators
Standardized scores for farms and scores per type of system for all indicators are summarized in Table 3.

Use of agrochemicals
55.5% of farms obtained moderate scores, as units appropriately used the tick-killing agent Amitraz, classified as an III agrochemical (WHO, 2004). IS farms had the best score per system at 66.7 (good).

Electric energy consumption
Electric energy in Yucatan is generated by thermoelectric plants. Therefore, farms using alternative energy sources were given the best score in terms of sustainability. 77% of farms had excellent scores. All NS farms had scoring values of 100, whereas IS farms scored 91.3 and MS 54.

Fossil fuel consumption
44.4% of farms had excellent scores. IS had the best score between systems, with a score of 73.33 (good).

Care and use of water
All NS farms had scores of 100, whereas only and IS and one MS farms had excellent scores. Thus, NS farms had a score of 100 (excellent), whereas IS and MS farms had moderate scores (59.9 and 45.2, respectively).

Species richness
All NS farms presented scores above 70. IS2 scored 57.57 for this indicator, whereas MS2 had the lowest value at 30.3. Consequently, NS systems had the highest score between all systems (85.9; excellent), whereas IS and MS reached only scores classified as good (72.7 and 70.7, respectively)

Good feeding
For this indicator, there were no farms reaching excellent or good scores, as all farms had values below 41, with the lowest at 9.95 for farms IS2 and MS1 and the highest at 40.94 for farm IS3. IS farms had the best valuation (26.9; limited), whereas NS and MS reached 24 and 24.5, respectively.

Good Health
The best scores for this indicator was given to farms NS2 and IS3 (100; excellent). IS farms had the best score amongst systems (78.3; good), whereas NS and MS had scores of 71.3 and 77, respectively.

Appropriate behaviour
44.4% of farms had scores classified as good. When evaluated per type of system, all had scores considered as moderate, with IS systems presenting the highest score (61.6).

Production cost
All NS and IS farms had excellent scores, with 97.1 and 84.8, respectively. MS farms had an average score of 62.7, classified as good.

Cost/benefit relation
For this indicator, only farm NS1 had a 100 score (excellent). 66-6% of the farms had scores classified as not classified, with the lowest value at 5.66 for IS2 farm. Scores per system were also low for IS and MS systems, with not classified values of 12.8 and 12.6. NS farms obtained an average score of 51.3 (moderate).

Productive diversification
55.5% of farms were classified as not limited, with the lowest score for MS1 farm (10). The highest score was given to IS2 farm (100; excellent). When evaluated per type of system, IS farms had the best score at 48.3 (moderate), whereas NS and MS were valuated at 28.3 (limited) and 15 (not classified).

Independence of external supply
NS1 and IS2 farms were scored as excellent, whereas farm MS1 was scored as not classified with a value of 0. NS farms had the best scores with 79 (good), followed by IS with 70.2 (good) and MS with 50.5 (moderate).

Business plan
33.3% of farms had moderate scores. IS2 and IS3 farms received excellent scores (100 for both), whereas farm MS1 had the lowest score at 0 (not classified). IS farms were score an average of 75 (good), whereas NS and MS farms had moderate scores (58.3 and 41.7, respectively).

Salary level
Only one farm had an excellent score for this indicator (MS2; 100; excellent). The farm worst valuated was MS1, with a moderate score of 47.11. MS farms were the best valuated on average, with a score of 72.3 (good), whereas NS and IS farms had scores of 60.9 (good) and 63 (good), respectively. Table 4 and Figure 2 summarize the sustainability performance of NS, IS and MS systems per MESMIS attribute. NS farms had the best scores between type of system for the attributes Stability, Reliability and Resilience and Production, scoring as excellent (81.02) and good (64.66), respectively. Meanwhile, IS had the best scores for the attributes Adaptability and Selfreliance, with scores of 48.33 (moderate) and 63.38 (good). MS were the best ranked for Equity, with a score of 68.8 (good).

DISCUSSION
Livestock production in extensive systems is one of the most important economic activities in Yucatan, occupying 30% of the state's territory (INEGI, 2017). The implementation of alternative systems is paramount to preserve natural resources in the area. This study demonstrated that silvopastoral systems whether native or intensive, are a good alternative to improve the sustainability performance of livestock systems in the tropics, especially for indicators related to the environmental dimension, and that both types of silvopastoral systems are an option to transition to sustainable animal production. Likewise, areas of improvement were identified.
For the evaluation of the dimension Environment and the attribute Stability, Reliability and Resilience, the same indicators were used, and NS systems were scored as excellent, IS as good and MS as moderate.
The better performance of NS and IS systems is related to the different use of natural resources in silvopastoral systems compared to monocultures. For instance, electric energy consumption in NS and IS farms was rated as excellent, in great part due to the use of alternative sources of energy (solar and wind; NS3), units not using electric energy at all (NS1, NS2 and IS3) or using less than 74 kWh/ha/bimester (IS1 and IS2). The coast of the Yucatan Peninsula has wind potential because of its excellent wind flows (Alemán-Nava et al., 2014), thus making it a potential source of energy for all types of systems. Similar tendencies were observed for fossil fuel consumption, where NS and IS systems had average scores classified as good. The use of fossil fuel is not a sustainable option; the global reserve/production ratio estimated in 2012 was 54.2 years and its use is associated with environmental deterioration, a rapid growth in the level of greenhouse gas concentration and an increase in fuel prices (Dudley, 2012;Hernandez-Escobedo et al., 2011). NS and IS systems were not highly dependent on heavy machinery, as observed while describing current traditional agricultural practices in Mesoamerica (Chazdon et al., 2011); thus, silvopastoral systems were more sustainable in their energy use than MS.  Better environmental sustainability was also clear for the indicator 'Care and Use of Water', as NS were excellently scored. Systems with high tree coverage increase water retention , whereas the deforestation caused by the creation of MS is associated with a decrease in water infiltration (Martínez et al., 1992). This made MS farms increase their irrigation time and water consumption. Additionally, IS systems had moderate ratings for this indicator despite the presence of trees and shrubs. This could be related to the type of trees associated with the farms, as some are better at improving water retention (Esperschuetz et al., 2017); thus, in addition to the presence of tree coverage, increased knowledge on plants associations is vital to improve water retention and reduce irrigation times.
Finally, the indicator 'Species Richness' was also scored as excellent for NS systems, demonstrating the known relationship between improved biodiversity and the presence of trees and shrubs in silvopastoral systems (Broom et al., 2013). IS and MS systems had scores classified as good, indicating that other factors in addition to the presence of trees could also benefit biodiversity, such as the surrounding landscape of MS systems, which needs to be further investigated. For the Animal Welfare dimension, all types of systems were classified as good and IS systems had the highest score amongst systems, which could be attributed to the results obtained for the indicators 'Good Feeding', 'Good Health' and 'Appropriate Behaviour'. The indicator 'Good Feeding' was evaluated as limited for all systems but slightly higher for IS. Limited scores reflect severe setbacks on the provision of food and water. In particular, the evaluation of body condition for the 'Good Feeding' component 'absence of hunger' reflects fat content of individuals, which could relate to reproductive and immune alterations (Broring et al., 2003). The slightly higher valuation of IS for 'Good Feeding' is consistent with previous results were IS systems were superior than MS for this indicator (Tarazona et al., 2013); however, this previous study also showed scores of 98 for the 'absence of prolonged hunger' component, which were attributed to the presence of high protein grasses in the tropical grassing context, such as C. plectostachyus, P. maximum, with crude protein percentages of 8 -10.8 % and 5.5 -7.4%, respectively (Lagunes et al., 1999;Molina et al., 2016). In Yucatan, there is documented the presence of C. plectostachyus and P. maximum (INEGI, 2017), however, their exact contribution to animal feeding in the systems evaluated is unknown, and a much lower 'absence of hunger' score of 15.8 was found in the IS evaluated. Thus, it is necessary to perform botanical census of systems to better interpret the low scores obtained for this welfare indicator. Meanwhile, the 'Good Health' indicator was good for all systems but higher in IS. Tree coverage can improve health indicators and body condition (Mancera et al., 2018), influencing also reproductive and immune function (Broring et al., 2003). As 'Good Health' is necessary to maintain system production, MS were also expected to have good values for this indicator, as shown here. The main difference between MS and silvopastoral systems could be in the methodologies used to maintain herd health, which could or not implicate the use of antibiotics or other substances. Thus, further research needs to evaluate health protocols in farms.
For 'Appropriate Behaviour' IS was classified as good an NS and MS were classified as moderate. In particular, the component flight distance of this indicator (distance at which animals flee when humans attempt to touch them) was smaller for IS and NS animals compared to MS, as previously seen when comparing IS with MS alone (Tarazona et al., 2013). It has been observed before those greater percentages of tree coverage are correlated with reduced flight distances; whether these reductions are related to the inability to flee due to trees preventing animals to move more freely, or due to a decrease in glucocorticoids as a result of reduced heat stress still remains to be proven (Mancera et al., 2018).
For the MESMIS attribute Production, NS systems had scores classified as good for this attribute, whereas IS and NS were classified as moderate. These differences can be attributed to the scores for 'Production Cost' and 'Cost/benefit Relation', which are part of the evaluation of this attribute along with Animal Welfare indicators. For 'Production Costs', NS and IS had excellent scores, whereas MS had only good scores. Silvopastoral systems are recognized for reducing production costs by increased utilization of local resources (Cuartas et al., 2014). Additionally, cost/benefit relation was better for NS, but classified as moderate, whereas IS and MS were not classified. This indicates that, although NS had better performance for this indicator, costs are still outstanding in relation to benefits. Cost/benefit analyses in silvopastoral systems have demonstrated that they can be economically productive when product quality and new commercial innovations are integrated, such as product certifications (Esquivel et al., 2004;Fassola et al., 2004). Certification is the process of identifying through labelling that products comply with a set of regulations governing the production process. As a market tool it creates niches, product recognition and/or secures price premiums (Taylor, 2005). Certification is considered the next step for the payment of environmental services (Ghazoul et al., 2009) and could improve cost/benefit relations in silvopastoral systems. This kind of scheme need to be further investigated for implementation in the future.
For the Economic Dimension, in addition to 'Production Cost' and 'Cost/benefit Relation', the indicators 'Productive Diversification', 'Independence of External Supply' and 'Business Plan' were also considered. NS had the highest score which was classified as good, followed by IS with moderate and MS with limited. This can be attributed to the better performance that NS had on Production Cost', 'Cost/benefit Relation' and 'Independence of External Supply'. Nonetheless, IS had better ratings for 'Productive Diversification' and 'Business Plan', which make silvopastoral systems better than MS. Similar results have been found by Chagoya (2004) and Chaparro (2005), who found better rentability and economic efficiency in multistrata agroforestry systems.
In this regard, IS had the highest score for the 'Productive diversification' indicator, which also reflects the Adaptability attribute. IS reached a moderate classification, whereas NS and MS were limited and not classified. It is known that silvopastoral systems provide diversification of products to farms in the form of timber, forage and other services (Villamil, 2017). Better scores for IS systems are consistent with the better resource management and the continuous development of producers' capabilities that intensification of silvopastoral systems implies , which would imply a better resource management for IS compared to NS. Nonetheless, as adequate production diversification requires technical knowledge to be successful (Calle et al., 2009), moderate scores for IS indicate that more technical knowledge is required in order to improve scores for this indicator and the Adaptability attribute.
'Independence of External Supply' was best rated for NS, although both NS and IS had good ratings and MS had a moderate score. As mentioned before, a great part of the low production costs of silvopastoral systems is their ability to rely on local resources (Cuartas et al., 2014). Likewise, the use of external inputs such as fertilizer and agrochemicals are replaced by natural processes such as natural soil fertility and biological control (Altieri and Nicholls, 2012) unlike MS systems, where the use of external supply is more prevalent. Finally, IS had good scores for 'Business Plan', which is also related to the improved resource management that intensification of silvopastoral systems imply which point toward the generation and implementation of an holistic business vision that includes management strategies such as careful land-use planning, rotational grazing, diversified forage, and diminished use of purchased inputs (Ferguson et al., 201;Nahed-Toral et al., 2013), which were characteristics found in silvopastoral systems in this study. 'Independence of External Supply' and 'Business Plan' were also included in the evaluation of the attribute Self-reliance, along with indicators 'Workers' development and training', and 'Organization and participation'.
Self-Reliance was valuated with a good score for IS systems, whereas NS and MS classified as moderate and limited, respectively. IS had better ratings not only for 'Business Plan', but also 'Workers' development and training' and 'Organization and participation', which account for the better outcomes compared with the other systems. 'Workers' development and training' had moderate scores for both IS and MS, meaning that more than half of workers had finished secondary education but did not received training courses in the farm, or that all workers had concluded primary education and received training in the farm or independently. Calle et al. (2009) demonstrated the importance of training in silvopastoral systems, as it increases worker satisfaction, promotes commitment with the job and increases positive production outcomes. Moderate scores in training-related outcomes were also found in cattle systems evaluated with the Sustainability Assessment for Food and Agriculture (Gayatri et al., 2016). Additionally, training also relates with poor health and safety in the workplace (Iunes, 2002). Therefore, the implementation of training courses in all systems could also help improve job satisfaction, security and improve workers' livelihood. Lack of training and participation in silvopastoral systems has been previously observed in South-eastern Mexico, where producers have a low level of participation in agricultural organizations and few receive training, technical assistance, or financial support .
This relates with outcomes encountered in 'Organization and participation', where IS systems had the highest scores but moderate outcomes, whereas NS and MS were not classified. The presence of farmer groups and organizations is essential; for instance commodity roundtables, a form of non-state marketdriven governance system, include the creation of working groups that address specific plans and actions towards sustainable development (Buckley et al., 2018); group participation, for instance in local markets, also increases good access to information networks and provides a better understanding of implemented policies and regulations (Gayatri et al., 2016). Thus, participation needs to be improved for the implementation and effective management of silvopastoral systems. 'Workers' development and training' and 'Organization and participation' were also indicators for the Social dimension, which additionally comprised the attribute Equity with the indicators 'Transmissibility and succession', 'Governmental subsides and assistance' and 'Salary level'. MS had the highest Social dimension and Equity ratings, with moderate and good scores, respectively. Such results are related to MS owning the best outcomes for 'Transmissibility and succession', 'Governmental subsidies and assistance' and 'Salary level'. However, for this indicators, good ratings were not necessarily related to good outcomes in Yucatan's context. 'Transmissibility and succession' estimates the possibility of passing on a determined amount of financial resources to allow the continuance of the system for the next generation and is a measure of sustainability for livestock systems (Tommasino et al., 2012). MS had excellent transmissibility, whereas silvopastoral systems were scored as limited, which implies that monocultures have a better chance to remain over generations. In addition, for 'Governmental subsidies and assistance' MS had moderate scores, whereas silvopastoral systems were classified as limited. In Latin American, the policy has encouraged deforestation for timber extraction and conversion of forest areas to monocultures. Such encouragement has come from subsidized credit, guaranteed prices, and other incentives and large-scale ranches continue to practice large-scale deforestation in many areas (Ramírez-Cancino and Rivera-Lorca, 2010;Villamil, 2017). Therefore, the better MS score on this indicator reflects governmental policies that need further improvement to promote alternative livestock systems. In this sense, technical assistance associated with the payment for environmental services has shown to have a positive influence on adoption rates of silvopastoral systems, particularly for practices with private benefits of improving rangeland production (Garbach et al., 2012). Finally, 'Salary level' was rated as good for all types of systems, but MS systems had higher scores for this indicator. Despite scores rated as good, this indicator, as well as 'Governmental subsidies and assistance', is not a reflection of good wages in the Mexican context. One of the most important problems in Yucatán is salary levels (INEGI, 2017). In our evaluation, only farm MS2 paid the minimum wage. Therefore, it is necessary that all systems improve salaries for workers regardless of the type of system. IS farms were determined as more sustainable than NS and MS farms regarding the following criteria: 'Animal Welfare' and attributes 'Adaptability' and 'Self-reliance'. Nonetheless, it is important to highlight the better performance of NS on indicators related to the Environment and Economic dimensions, such as 'Care and Use of Water', 'Species Richness', 'Productive Diversification' and 'Independence of External Supply'. Since intensive and native silvopastoral systems are already present in the Mexican tropics, it is important to carry our more studies in both types of silvopastoral systems to identify potential improvements and support the progress of farmers already engaged with silvopastoral farming, in order to foster the transition to alternative farming systems in the region.

CONCLUSION
To date, this is the first study comparing NS, IS and MS using the MESMIS methodology. It was demonstrated that NS and IS systems had a better sustainability performance than MS per MESMIS attribute and sustainability dimension; however, our evaluation also highlighted areas of improvement that need to be considered for the successful implementation and preservation of silvopastoral systems in the area. When evaluated through MESMIS attributes, NS systems had better ratings for the 'Stability, Reliability and Resilience' and 'Production', as NS scored better for environmental indictors, 'Production Cost' and 'Cost/benefit relation'. IS were better scored for the attributes 'Adaptability' and 'Self-reliance', as all related indicators, excluding 'Independence of external supply' was the highest rated for IS. Nonetheless, it was observed that, despite a clear business plan, aspects such as technical knowledge to improve benefits, training to increase workers' knowledge and better conditions for participation and engagement were necessary to increase sustainability performance. Finally, good Equity ratings for MS were reached due to the presence of good heritable capital for the continuance of these systems, a good access to government subsidies and the best salary level encountered in an MS farm.
For sustainability dimensions, Environment and Economic were best rated for NS systems. The Environment dimension included those indicators that reflected the direct benefits that the presence of trees has on farms, such as better water retention and increased species richness. Meanwhile, the Economic dimension reflected the good scores obtained for NS in the indicators 'Production Cost' and 'Independence of External Supply', which are related to the ability of silvopastoral systems to supply their own resources locally and from the same farm. Meanwhile, IS had the best Animal Welfare scores, in great part due to 'Good Feeding', 'Good Health' and 'Appropriate behaviour' scores for this type of system. Finally, MS had the best Social dimension score, as it included the Equity attribute indicators, which, as mentioned before, were highly rated but not excellent in the bigger context of social sustainability.
The MESMIS framework was successful in identifying an improved sustainability performance in NS and IS systems, as well as highlighting those issues that need to be addressed to contribute to the development of alternative farming systems in the state of Yucatan. Further research on aspects such as assessing objective criteria for all dimensions is still needed. This information will be useful to design better policies to improve the management of silvopastoral systems and contribute to the preservation of environmental services.