Introduction

Food and land-use systems are unsustainably managed in most regions of the world (Dirzo and Raven 2003; Foley et al. 2005; Metzger et al. 2006; Newbold et al. 2015; Tilman et al. 2017). In particular, these systems fail to deliver decent livelihoods to many poor farmers; drive poor health outcomes, including persistent hunger, obesity, and malnutrition; account for over a quarter of global greenhouse gas (GHG) emissions (IPCC 2019); cause unprecedented biodiversity and forest loss; drive over 90% of water scarcity globally; and threaten the health of oceans and freshwater ecosystems through chemicals (Navarro et al. 2021; Stehle and Schulz 2015) and fertilizer runoff and percolation. Food and land-use systems are highly inefficient with up to one-third of food lost or wasted (Aguilar Gutiérrez 2017). Furthermore, these systems are vulnerable to climate change and extreme events (Mbow et al. 2019). Under business as usual, pressures will increase as the growing world population and rising per capita demand for food, fiber, and energy feedstocks risk making land-use and food systems even more unsustainable.

Mexico faces all of the challenges of unsustainable land use. Agriculture and livestock production have been and remain the main causes of deforestation (CONAFOR 2020; Ellis et al. 2017; Mendoza-Ponce et al. 2018). Based on Mexico’s National Institute of Statistics and Geography (INEGI by its Spanish acronym) data on land use and vegetation (series I and VI, INEGI 1992, 2017 respectively), we estimate that this phenomenon has resulted in the loss of almost 10% of the forests in the country since the 1970s (SEMARNAT 2019). Most of the country’s drylands have been used for cattle ranching for centuries, and many of them have suffered intense biological changes as a result. During the 1970s and 1980s, several Mexican federal government programs intended to increase the production of beef in the country (e.g., “Alianza para el campo”, “Estímulos a la producción ganadera” and “Programa nacional de ganaderización”) were responsible for massive deforestation (Bravo Peña et al. 2010) and the introduction of an exotic invasive grass (Cenchrus ciliaris) that now grows rampantly across the country’s drylands (Bravo Peña et al. 2010).

Meanwhile, the food system is not delivering healthy diets. According to the Latin American and Caribbean Food Security Scale, an indicator to assess household perception of inadequate food access in the previous 3 months, 43% of the Mexican population experiences moderate to severe food insecurity, while seven out of ten adults are either overweight or obese (Mundo-Rosas et al. 2019). Changes in the diets of Mexicans, particularly among the urban population, tend to involve a lower consumption of fruits and vegetables and a growing demand for processed and ultraprocessed products (Rivera et al. 2004; Reyes-Garcia et al. 2021). This dynamic triggers changes in production chains, especially in the use of resources (e.g., energy, soil, water, and biodiversity), by encouraging the expansion of livestock and crop production, particularly grains and industrial crops, further intensifying the pressure on Mexico’s agricultural systems and ecosystems (Ibarrola-Rivas and Granados-Ramirez 2017).

Mexico has made climate and biodiversity commitments within the United Nations Framework Convention on Climate Change (UNFCCC) and the United Nations Convention on Biological Diversity (CBD), respectively. Forest conversion to agriculture and other land-use conversions are key drivers of both GHG emissions and biodiversity loss. In contrast, nature-based solutions such as improved land management, restoration, and protection can meet up to a third of emission reduction needs of the Paris Agreement, also contributing toward meeting the sustainable development goals (SDGs) (Griscom et al. 2020; Roe et al. 2019). One of Mexico’s challenges is fulfilling these international commitments while ensuring sufficient and healthy food production for national needs.

This paper employs an integrated land-use modeling tool—the FABLE Calculator—to estimate the degree of policy ambition required for the country to meet mid-century climate, conservation, and production goals in the land-use sector. The Calculator produces long-term pathways for land use in Mexico under alternative policy scenarios, allowing us to quantitatively explore the degree to which various policy and technological drivers of land-use change, including agricultural and livestock productivity, protected area expansion, reforestation, and dietary change, come together to put the country on a path to sustainability. The coordination of national efforts to achieve multiple targets in the long term requires modeling flexibility that this tool can accomplish for complex systems like food and land systems. Furthermore, an integrated model allows robust and transparent analysis of trade-offs and opportunities of competing pathways representing national strategies and policies related to land-use and food systems.

Like most other countries, Mexican policymakers would benefit from tools for integrated land-use planning that can address the complex synergies and trade-offs between agriculture, water, land use, biodiversity, healthy diets, and GHG emissions. Just like it is impossible to design and implement economic policies without sound macroeconomic models, countries cannot make their land-use and food systems sustainable without robust analyses and tools to help design policies and model their impacts.

Our approach is predicated on the long-term pathways as an organizing principle for stakeholders and knowledge communities related to land use. The transformation of land-use and food systems requires long-term strategies, as called for in the Paris Agreement (Sachs et al. 2019). Long-term pathways help in three critical ways: (1) they provide a framework for engaging stakeholders (governments, businesses, civil society, and the scientific community) to review and suggest improvements on how to achieve national and international targets, which can build a societal consensus for transformations; (2) they encourage a long-term perspective to prevent countries locking themselves into unsustainable land-use systems; and (3) they help identify mid-term technology benchmarks needed to achieve the targets, such as increases in agricultural productivity or efficiency gains in cattle ranching, which can then guide private sector action and innovation (FABLE 2020).

Methods

An interdisciplinary team of researchers identified long-term objectives for Mexico consistent with achieving the SDGs, the objectives of the Paris agreement, and national targets published in the National Development Plan 2019–2024 and in the Sectoral Programs 2019–2024. On the basis of current international commitments and published Mexican public policies, two pathways were created, which vary in terms of assumptions about future characteristics of Mexican land-use and food systems. The Current Trends (CT) pathway captures present-day public policy commitments, while the Sustainable (SUST) pathway depicts a higher level of ambition and models policies that would move the country toward sustainable land and food systems as well as meeting international commitments. Each parameter for the pathways was developed on the basis of existing national and international scientific literature supplemented by government expert opinion and by the experience of the author team.

Table 1 shows the differences in assumptions across the CT and SUST pathways. The CT pathway is characterized by relatively low population growth (from 128 million in 2020 to 148 million in 2050) (CONAPO 2020), significant constraints on the expansion of agricultural frontier reflecting current Mexican policy (SAGARPA 2017), a low afforestation target of 4.7 Mha, an increase of up to 17% in the extent of protected areas, the same livestock productivity growth as over 2000–2010 (SIAP 2020), a continuation in diets similar to the 2000–2010 trend (high in cereals and sugar, increased intake in oils and fats, roots, nuts and red meat) combined with low physical activity, and increased exports and imports compared to 2010. This pathway corresponds to a future based on current policy and historical trends; in addition, it points to considerable progress with regard to halting agricultural expansion and a reconversion of cropland toward cultivation of high-value exports (SAGARPA 2017). Moreover, this CT pathway assumes a global GHG concentration trajectory that would lead to a radiative forcing level of 6 W/m2 (RCP 6.0), or a global mean warming increase likely between 2 and 3 °C above pre-industrial temperatures, by 2100. The CT pathway also includes the corresponding climate change impacts on globally important crop yields by 2050 for maize, rice, wheat, and soybeans (Arneth et al. 2017).

Table 1 Parameter values, underlying assumptions and rationale for each pathway
Full size table

The SUST pathway models significantly increased efforts to adopt sustainable policies and practices, while staying within the bounds of plausibility as gauged by experts on the author team. Compared to the CT pathway, the SUST pathway assumes progress toward healthier diets that rely less on cereals and more on high intake of fruit, vegetables, and pulses as well as animal protein in healthy quantities. The SUST pathway assumes a high afforestation target of 5.8 Mha, 30% of the total land included in protected areas, and no expansion of the agricultural area. The CT pathway assumes that productivity for crops will follow the growth trend from 2000 to 2010 (as the model uses 2000, 2005, and 2010 as historical values). However, the SUST pathway assumes an increase in productivity levels for crops and livestock as a result of improved agricultural practices as well as the introduction of improved germplasm and silvopastoral practices. The SUST pathway includes a global GHG concentration trajectory that would lead to a lower radiative forcing level of 2.6 W/m2 by 2100 (RCP 2.6), in line with limiting warming to 2 °C.

Description of pathway assumptions

Diets

Diets in the CT pathway approximate the current average Mexican diet in the FABLE Calculator, using information from the 2016 Midway National Health and Nutrition Survey (Castellanos-Gutiérrez et al. 2021). It is important to note from the outset that Mexican diets vary considerably across the country, such that the “average Mexican diet” is a construct for the purposes of the modeling tool that does not reflect the diet of any particular region of Mexico. The diet was calculated for a daily energy requirement of 2288 kcal, corresponding to a male person with a sedentary level of physical activity and considering Mexican median male weight and height. The diet in the SUST pathway is based on a published diet (Castellanos-Gutiérrez et al. 2021) that provides daily energy intake by food groups by adapting the EAT-Lancet Commission (Willett et al. 2019) recommendations of a well-balanced and sustainable diet to the Mexican context. Furthermore, in the case of some food groups, less ambitious goals were considered taking into account current dietary patterns (this was the case for red meat and nuts). This diet considers the energy distribution of each food group with low environmental impact and is also healthier and culturally adapted for Mexicans. The diet was calculated for a daily energy requirement of 2288 kcal, corresponding to a person with a moderate level of physical activity and considering Mexican median male weight and height.

Crop productivity

The CT pathway assumes that crop productivity for the four most important crops—maize, beans, wheat, and sorghum—continues 2000–2010 productivity trends. An annual average yield increment was estimated for these crops for 2000–2010 based on data from the Mexican Agricultural and Fisheries Information Service (SIAP by its Spanish acronym) (SIAP 2020). Each crop’s yield up to 2050 is the result of a linear extrapolation from the 2015 value using the 2000–2010 trend. In the case of maize, we calculate separately the mean annual yield increment for rainfed and irrigated maize. The overall maize yield in 2050 is the average of irrigated and rainfed yields in 2050 weighted by their proportion of harvested land. The resulting 2050 yield for maize, beans, wheat, and sorghum is 5.98 t/ha, 1.05 t/ha, 5.7 t/ha, and 4.3 t/ha, respectively.

The SUST pathway incorporates the expected yield improvement of the MasAgro program in maize, currently implemented in 19% of the national area cultivated with maize (SAGARPA 2017; Silva Hinojosa 2017). This government program consists of using improved seeds and efficient crop management practices. Since MasAgro projections run to 2024, we calculate expected maize yield in 2024 by incorporating yield improvement in rainfed areas under this program, and assuming that yield remains constant in the rest of the rainfed area at 4.3 t/ha (SDAyR and UACh 2019). Specifically, we estimate that the improved seeds generate a mean annual increase of 0.088 t/ha between 2016 and 2024, and then for the area not under MasAgro program, we calculated the mean annual increase of 0.083 t/ha from 2006 to 2016 and assume an increased 30% in rainfed yield due to improvement of farming practices (Duvick 1984), resulting in a mean annual increase of 0.025 t/ha. Once the two practices (improved seeds and farming practices) were added, the resulting mean annual yield increment was 0.113 t/ha. Then this value was extrapolated to 2050, resulting in a new weighted average of rainfed yield of 7.3 t/ha. For irrigated maize, we use the annual mean increases from 2006 to 2016 (0.254 t/ha per year) and extrapolate it to 2050. Finally, we add the two yield increases to obtain an area weighted average 2050 yield of 10.2 t/ha, assuming that the proportion of the harvested areas under rainfed and irrigation is maintained until 2050.

For sorghum, beans and wheat, the projections for 2030 generated by the National Agricultural Plan (Planeación Nacional Agrícola) were used to assign productivity increases by extrapolating to 2050 (SAGARPA 2017). The resulting yield for beans, wheat, and sorghum in 2050 is 2.3 t/ha, 6.2 t/ha, and 6.5 t/ha, respectively.

Livestock productivity

The CT pathway assumes a continuation of the 2000–2010 livestock productivity trend to 2050. Using data from SIAP (SIAP 2020), the average annual growth for livestock productivity was calculated (e.g., cattle value was 80 kg/head per year) and extrapolated to 2050 assuming that a continuation of practices is maintained.

In the case of the SUST pathway, pork, eggs, and chicken follow the same productivity trend as 2000–2010. For each product, its annual average livestock productivity growth was calculated using data from SIAP (SIAP 2020) from 2000 to 2010 and the resulting value was extrapolated to 2050. However, we used a different approach for beef and milk, given that agricultural expansion in Mexico is mostly due to extensive cattle ranching (Anta Fonseca et al. 2008). Research on high-productivity livestock systems based on modern silvopastoral systems (Flores-Estrada 2014; Guevara Sanginés et al. 2020; SIAP 2020) was used to calculate cattle productivity for Mexican biomes. Modern silvopastoral systems have been extensively researched in Mexico. This livestock production system is effective in providing more favorable conditions for livestock, resulting in increased survival, higher gains in body weight and milk production, and reduction of the length of the production cycle when using an effective silvopastoral system and sufficient resources (irrigation, labor and agricultural inputs) (Álvarez et al. 2021). The silvopastoral systems are mostly managed through rotational grazing with high livestock densities and brief grazing periods interspersed with long recovery periods for the protein sources (Montagnini et al. 2013). Although these systems can be found widely around the world in different biomes, they are most common in ecosystems with high levels of precipitation (tropical and temperate), as humidity is the most frequent limiting factor at introducing these production systems. Using data from Guevara Sanginés et al. (2020) and Lara et al. (2021), we calculated the mean productivity for milk and beef production for each Mexican biome, for extensive and modern silvopastoral systems. We estimated the average livestock yield as a weighted average composed of 70% traditional extensive systems (SAGARPA 2017), 10% modern silvopastoral, and 20% intensive systems (Chauvet-Sánchez 1999). Considering that adopting silvopastoral systems is a slow process due to the magnitude of required investment that each producer has to undertake (Apan-Salcedo et al. 2021), we used the weighted average productivity for cattle beef (79 kg/head per year) and for milk (7718 L/head per year) as the SUST pathway assumption for 2050.

Afforestation/reforestation

Reforestation and afforestation efforts in Mexico have varied significantly in the last two decades, and that variability was included in both pathways. The CT pathway uses data from a combination of past and current reforestation efforts, and assumes that the rate of deforestation from 2020 to 2050 will decrease due to a Mexican public policy of no agricultural expansion starting in 2015 (SAGARPA 2017). The past and current programs include the reforestation that occurred from 2015 to 2018 (CONAFOR 2021), 60% of the area planned to be implemented by the program “Sembrando Vida” that corresponds to reforestation from agroforestry practices from 2018 to 2024 (BIENESTAR 2020) and the National Forestry Program (PRONAFOR) intended goals (SEMARNAT 2020a) from 2018 to 2024 (Table 1). For the period 2024–2050, there is no declared reforestation goal from the Mexican government. We therefore assume that an effort similar to the one from 2013 to 2020 will continue under the current trends. The annual average reforestation growth was calculated for 2013–2020 and used to calculate the total reforested area for the period 2014–2050. The total reforestation area by 2050 for CT is 4,728,395 ha.

The SUST pathway uses the same criteria as the CT pathway for the 2015–2024 period. For the remaining years (2024–2050), the National Commission on Biodiversity (CONABIO by its Spanish acronym) high priority areas for restoration in temperate and tropical forest biomes were used to assign reforestation potential (Tobón et al. 2017). This is based on the Mexican goal to achieve net zero deforestation by 2030, committed in the Nationally Determined Contributions (NDC) (SEMARNAT 2020b), and the 30% reduction of deforestation by 2024 included in the PRONAFOR and the Sectoral Program for Environment and Natural Resources (PROMARNAT by its Spanish acronym) programs (SEMARNAT 2020a; c). The total area reforested by 2050 under the SUST pathway would be 5,886,104 ha. The establishment of forest plantations was considered marginal in both CT and SUST.

Integrated assessment modeling using the FABLE Calculator

The FABLE (Food, Agriculture, Biodiversity, Land Use and Energy) Consortium has developed the “FABLE Calculator”, an Excel-based modeling tool for food system and land-use change over 2000–2050 (Mosnier et al. 2019). The impact of different policies as well as changes in the drivers of these systems can be tested by altering model parameters; for example, agricultural intensification, crop and livestock productivity, dietary shifts, deforestation or reforestation, to highlight policy drivers of land-use change.

The FABLE Calculator is an accounting tool and not an optimization tool, such that price changes are not modeled endogenously as in economic models (prices are only used ex-post to compute production and trade value). The FABLE Calculator focuses on agriculture as the main driver of land-use change. It includes 76 agricultural raw and processed products from the crop and livestock sectors and relies extensively on the FAOSTAT database for input data. In each 5-year time step over 2000–2050, the level of agricultural activities, land use, food consumption, trade, and GHG emissions are computed, as well as the total area where natural processes predominate as areas with low human impact because they are not managed primarily for human needs (Jacobson et al. 2019; Mosnier et al. 2019). The FABLE Calculator is currently being used by members of the Consortium in 19 countries around the world (Mosnier et al. 2019).

We adapt the FABLE Calculator to model Mexican land-use and food systems and to construct the CT and SUST pathways. Each pathway was created with settings and parameter values described above. The following output indicators for 2050 were used to identify whether each pathway achieved long-term objectives: feasible calorie production, land area where natural processes predominate, and total CO2e emission reduction from the land sector. During the entire process of capturing existing and potential future policies, the author team was in communication with experts at the Ministry of Agriculture (SADER by its Spanish acronym) and the National Institute of Public Health (INSP by its Spanish acronym) to iteratively improve our pathway assumptions and include public policies that were being considered for adoption.

Sensitivity analysis

A sensitivity analysis was performed to identify which model parameters significantly affect the outcomes of the SUST pathway. However, since modeling diets at population level provides the information of the minimum kcal needed to be produced, we decided to measure the effect of the evolution of diets from 2000 to 2050. To do this, we created three diet phases with the same kcal amount of 2288 kcal, but gradually reduced kcal portion in cereals, red meat and sugar, and steadily increased kcal portions in food groups such as fruits and vegetables, oils and fats, pulses, and nuts. These three diet stages are the current Mexican diet (used in the CT pathway), the healthy diet in the SUST pathway, and a diet halfway between the two (“Middle Diet”), representing an incomplete transition from the current to the healthy diet. Six parameters were chosen for the sensitivity analysis of the SUST pathway: population, food waste, post-harvest loss, livestock productivity, ruminant density, and maize productivity. These were sequentially tested by increasing or decreasing their parameter values by 10%. For example, the SUST pathway assumption for population in 2050 is 145.73 million, which was changed by a 10% increase to 160.3 million or a 10% decrease to 131.3 million. For each parameter test, we used three output variables: GHG emissions in CO2e, percentage of land where natural processes predominate and support biodiversity, and food production measured in feasible kcal.

Results

Figure 1 compares the 2010–2050 change in diets in the CT and SUST pathways. For the CT pathway, the current average intake mostly consists of eggs, red meat, roots, and sugars, with cereals representing 60% of the total calorie intake. Feasible food consumption per person ranges from 2524 kcal per day in 2030 to 2288 kcal per day in 2050, indicating that this pathway provides enough food to meet the daily caloric requirements of the population. The minimum dietary energy requirements (MDER) of the average Mexican are 1858 kcal in 2030 and 1864 kcal in 2050. As mentioned earlier, Mexican diets vary considerably across the country, such that these results on the “average diet” do not imply that all Mexicans will be consuming a diet approximating the one here.

Fig. 1
figure 1

Diet composition under Current Trends and SUST pathways. Comparison of 2010–2050 diet changes in the CT and SUST pathways

Full size image

Despite dietary shifts and improved environmental outcomes under the SUST pathway, Mexico could provide enough food to meet caloric requirements for its population (Fig. 1). Feasible kcal production reaches 2542 kcal per day per person in 2030 and 2288 kcal per day in 2050, always exceeding the 2090 kcal MDER of the average Mexican. As shown in Fig. 1, while the feasible kcal of the CT and SUST diets are very similar, the dietary composition differs. Specifically, red meat, pork, alcohol, and sugar are a significantly smaller proportion of the 2050 diet in the SUST than in the CT pathway, while pulses, nuts, fruits, and vegetables are larger in SUST than in CT.

Figure 2 shows the evolution of area by land cover type in the FABLE Calculator. The top figure shows the CT pathway, in which most land-use categories remain roughly constant over the 2000–2050 period, with some increases in urban area. Deforestation is offset by new forest cover due to reforestation efforts to accumulate a total of 4.73 Mha reforested by 2050. Cropland decreases from 25.4 Mha in 2000 to 23.4 Mha in 2050, while pastureland increases from 80.7 Mha in 2000 to 81.3 Mha in 2050 due to a higher amount of animal protein consumption and limited increases in cattle productivity.

Fig. 2
figure 2

Evolution of area by land cover type in the Current Trends pathway and in the Sustainable pathway. This figure shows the evolution of area by land cover type in the FABLE Calculator. There is a significant expansion of land in the “Other Land” category that is removed from productive use and can return to sustaining biodiversity

Full size image

The bottom portion of Fig. 2 shows that, under the SUST pathway, Mexico experiences a cropland reduction of 15% from 2000 to 2050, decreasing to 24.68 Mha in 2030 and to 21.59 Mha in 2050. Pastureland reaches 67.34 Mha in 2030 and 42.39 Mha in 2050, almost half of its 2000 area. The reduction in pastureland is due to the reduction of animal protein and the increase in cattle productivity. This change in pasture area dramatically reduces land pressures and results in a significant expansion of land in the “Other Land” category (i.e., all other vegetation different from forest) that is removed from productive use and can return to sustaining biodiversity.

Figure 3 shows GHG emissions from the land sector from 2000 to 2050, with the total net emissions from the CT pathway in the solid black line. Livestock remains one of the main generators of GHG emissions; in addition, there is no GHG sequestration due to pastureland reverting to other types of vegetation (“Other Land” category). Total GHG emissions increase from 66 MtCO2e to 71 MtCO2e in 2050, reaching a maximum of 73 MtCO2e in 2030. With 55 MtCO2e in 2030 and 59 MtCO2e in 2050, the livestock sector accounts for 75% of total GHG emissions in 2030 and 83% in 2050. The large increase in emissions from 2010 to 2015 is due to land-use conversion as the main emitter in 2015 with 58 MtCO2e, a product of a large amount of forest land converted into pastureland. In 2020, there is a significant decrease in emissions because land-use conversion becomes a net sink mainly due to the policy of no expansion of agricultural land beyond the 2015 area (SAGARPA 2017).

Fig. 3
figure 3

Projected land sector emissions and removals between 2010 and 2050 by main sources and sinks. This figure shows net greenhouse gas (GHG) emissions from the land sector from 2000 to 2050, with the total net emissions from the Current Trends pathway in the solid black line

Full size image

Under the SUST pathway, there is a reduction in total GHG emissions of 29 MtCO2e in 2030 and 78 MtCO2e in 2050 compared to the CT pathway. Although livestock is the main emitter of CO2e, the progressive conversion of pastureland to other types of vegetation (Other Land Category) would compensate those emissions through sequestration, to the point that in 2050 CO2e sequestration would be larger than emissions from the entire AFOLU sectors.

Figure 4 shows the evolution of land area where natural processes predominate from 2010 to 2050, and shows that the CT pathway results in a 2050 land area where natural processes predominate that falls below the initial area in 2010. While the 2010 share of total land supporting biodiversity is 28%, that share falls to 27% in 2030 and remains stable until 2050. Meanwhile, under the SUST pathway, the area where natural processes predominate begins increasing in 2020 to reach 46% by 2050.

Fig. 4
figure 4

Evolution of area where natural processes predominate. This figure shows the evolution of land area where natural processes predominate from 2010 to 2050, and shows that the Current Trends pathway results in a 2050 land area where natural processes predominate that falls below the initial area in 2010

Full size image

Sensitivity analysis

Table 2 shows the results of the sensitivity analysis, where key SUST pathway parameters (population, food wasted, post-harvest loss, livestock productivity, ruminant density, and maize productivity) were modified by ± 10% of its value to evaluate the effects on output variables including total GHG emissions, land supporting biodiversity, and feasible kcal production. These sensitivity analyses are shown separately when the diet is kept as the Current Diet until 2050, when diets move somewhat toward becoming healthier (“Middle Diet”) and when the Healthy Diet is reached by 2050. Values in bold indicate the original parameter without ± 10% modification.

Table 2 Sensitivity analysis results
Full size table

Several patterns emerge consistently across the sensitivity analysis. While continuing with the Current Diet until 2050 (keeping all other parameters equal to those for the SUST pathway) results in 41.95 MtCO2e in net annual emissions, altering the diet toward the Middle Diet results in 22.73 MtCO2e, and the healthy diet yields the − 8.01 MtCO2e that is reported in Fig. 3. Meanwhile, the land supporting biodiversity is 53% in 2050 if the Current Diet is maintained, whereas it is a much greater 65% under the healthy diet assumption. The effects of these changes in diet are larger in magnitude than the variation of model parameters by 10%.

Of the parameters besides diets examined in this sensitivity analysis, model outputs are most sensitive to changes in the population growth assumption. If population growth is slower such that the 2050 population is 131 million (instead of the 146 million in the SUST pathway assumptions), the net annual GHG emissions under the healthy diet scenario decreases to − 21.24 MtCO2e, a significant change from the model output of − 8.01 MtCO2e. This is a larger change than a 10% reduction in food waste or harvest loss, or than a 10% increase in the 2050 value for cattle productivity, ruminant density, or maize yield. Of these remaining variables, the change in cattle productivity yields the largest change in net emissions, followed by ruminant density and then maize yield. Comparable ± 10% changes in the food waste and crop harvest loss parameters result in the smallest change in 2050 net emissions. These same patterns are observed when looking at parameter change effects on the 2050 land area supporting biodiversity, or if we make the comparisons in the “Middle Diet” results or if the Current Diet is maintained until 2050.

With regard to food production and self-sufficiency as a model output, the sensitivity analysis indicates that small changes in parameters never result in an inability to produce enough food to satisfy national demand. The changes in feasible kcal across the diet scenarios are a function of kcal demand changes across those dietary regimes, but the 10% variation in any of the other parameter values keep the kcal production constant. This means that, under the parametrization of the Sustainable Pathway, food security at a national level is not compromised by these small parameter changes.

Discussion

Our pathways analysis suggests that Mexico can meet its mid-century international climate and biodiversity commitments as well as its production and healthy diet goals by making ambitious and yet feasible policy decisions. Our results suggest that Mexico can adopt a feasible land-use pathway (as presented in Table 1 SUST pathway) that would ensure adequate nutrition for the population, set a limit to further agricultural expansion, and expand natural habitats. Mexico needs to promote highly productive and sustainable food systems that will increase its self-sufficiency in key product groups (animal protein, pulses, and cereals). Mexico also needs to implement existing policies to ensure that pressure for land conversion toward pasture and agriculture is reduced.

The model results highlight the role that diet plays in reducing land-use change and GHG emissions. A change toward a healthier diet implies a reduction in the intake of refined cereals and sugar but also an increase in whole cereals, fruits, vegetables, and nuts, and a healthy consumption of animal protein, which includes less red meat and dairy products (EAT-Lancet Commission 2019). These changes would affect what is produced in Mexico and what needs to be imported, and a combination of strategies would result in decreased pressures on land systems. To promote a necessary shift in diets, it is necessary to implement measures that encourage consumers to make healthier food choices. Placing nutrition labels on the front of the food packages (Jáuregui et al. 2020), including a sales tax to reduce sales of sugar-sweetened beverages and energy dense ultra-processed products, and increasing consumption of untaxed beverages (Colchero et al. 2017) are some of the general strategies that have been proposed.

However, the most important policies need to address the obesity epidemic for school-age children, such as a policy proposed in 2010, which included the development of general guidelines for the sale and distribution of food and beverages in elementary schools, with the objective of facilitating an adequate diet for children in schools and having a structured and unified regulation among states (SEP and SSA 2014; SSA 2010). It would be desirable that policies such as these be implemented and monitored to ensure compliance. This policy was created in 2010 and includes the development of general guidelines for the sale and distribution of food and beverages in elementary schools; the goal was to promote an adequate diet for children, and to establish structured and unified regulation among states (SEP and SSA 2014; SSA 2010).

Importantly, existing Mexican programs could be strengthened and scaled up to the necessary ambition to move the country toward sustainable land-use and food systems. For example, agricultural productivity increases can be achieved by improving the genetic base and updating agricultural practices to produce maize and other grains (Licker et al. 2010; Mueller et al. 2012). The MasAgro program led by the Maize and Wheat Improvement Center (CIMMYT by its Spanish acronym) has been shown to be efficient in reaching this goal by introducing a strong capacity-building program based on productivity gains as well as by adopting genetically improved seeds and cultural practices adapted to each municipality (CIMMYT and SADER 2018). For cattle ranching systems, increases in productivity could be achieved through the adoption of silvopastoral practices, rotating pastures, fenced paddocks and reticulated water systems, which improve the menu of feeding components of traditional cattle diets, increasing weight gains and the herd carrying capacity per unit of area (Guevara Sanginés et al. 2020). Furthermore, GHG sequestration can be enhanced by implementing programs that promote vegetation restoration and reforestation linked to agroforestry practices such as the “Sembrando Vida” program (BIENESTAR 2020), the National Restoration Program (SEMARNAT 2020a) and a myriad of private and civil society initiatives (e.g., Reforestamos Mexico, Reforestación Extrema, among others) which add up to the restoration and protection commitments of the country. In addition, initiatives aimed to reduce pressure on forested lands include: (a) the restoration of traditional systems of cattle ranching through the introduction of silvopastoral systems, (b) the introduction of high-productivity agroforestry systems with the use of high-value crops in the agriculture–forest interface, and (c) support to different demand-driven mechanisms to increase demand for products with a deforestation-free supply chain, contributing to the recovery of low agricultural productivity areas into natural vegetation lands. These measures could be particularly important when considering options for NDC enhancement.

Tools such as the FABLE Calculator and its pathways can serve as a common modeling framework for reflecting on trade-offs and synergies as policymakers debate the way forward in climate policy, environmental policy, and agricultural policy, as well as in nutrition guidelines and campaigns. This analysis provides an overview of how the land-use and food systems may evolve to 2050, and what drivers, policy challenges, and policy opportunities are most salient in Mexico. In particular, an integrated pathway for sustainable land use can inform Mexico’s long-term strategy and the land-use component of the country’s NDC.

Future analysis can continuously update pathway assumptions by incorporating more details of government policies as they evolve. In 2019, the Mexican government started the implementation of its National Development Plan for 2019–2024 (Presidencia de la República 2019), with the goal of improving the well-being of Mexicans through sustainable development. The CT and SUST pathway assumptions reflect our best understanding of the programs and operating rules that key federal agencies are considering (e.g., Sustainable Forest Development programs from CONAFOR), or updates on Mexico’s international commitments (e.g., Mexico’s latest NDC). Moreover, the federal government has recently created an intersectoral group called “Health, Agriculture, Environment and Competitiveness” (GISAMAC by its Spanish acronym). GISAMAC aims to support new forms of agricultural and forestry production to reduce the negative effects on human health and natural environment (SEMARNAT 2020d). Under the coordination of the Ministry of Environment, and in collaboration with 18 working groups from more than 10 government agencies and research institutes, GISAMAC focuses on articulating public policies to promote sufficient and sustainable production of healthy foods, prioritizing production from family farm producers and medium-sized producers as well as the protection and restoration of ecosystem services. These kinds of policy initiatives could play an important role in improving policy coordination on land-use and food systems across government ministries and agencies.

The FABLE Calculator has limitations, including a lack of prices to model market adjustments in food production or in international trade, and lacking geospatial resolution to resolve whether and how land-use changes would occur. In addition it is not yet allowing for land-use categorization that allows dual use (such as agroforestry, agroecology, or low-intensity livestock production that maintains natural land cover and biodiversity). However, the tool has proven to be an excellent departing point to identify the likely effects of different pathways of land use and diets, and has helped detect information and methodological gaps for future analyses. Future work should aim to enhance the FABLE Calculator to more fully represent Mexican strategic policies in the land sector and their potential trade-offs or synergies. For example, the GHG sequestration effects of sustainable forest management would be an important addition to model alongside policies such as halting agricultural expansion and avoiding deforestation, as well as to include alternative agricultural practices to represent current policy interest in a transition toward agroecological systems.

Conclusion

Pathways analysis for long-term planning is an important tool for policymakers considering trade-offs and synergies for land use. This paper deploys an integrated national-level land-use accounting tool that serves as a starting point to model the challenges and opportunities of sustainable land use in Mexico. Comparing Sustainable and Current Trends pathways suggests that expanding sustainable agricultural practices can increase productivity, reduce pressure for agricultural extensification, and generate an important reduction in GHG emissions. Dietary transitions play a crucial role in reaching SDGs; our results show that Mexico can adopt a feasible sustainable land-use pathway and ensure adequate nutrition for its projected population in 2050. The sensitivity analysis indicates that policies targeting reduction in population growth and adoption of a Healthy Diet at the population level are the more efficient ways to control land-use change and to limit greenhouse gas emissions. The transparency and relative simplicity of this modeling framework offers potential to improve communication and engagement across multiple sectors from academia and the policymaking community as they evaluate the future implications of alternative policy scenarios in land-use and food systems.