In this section, I provide an overview of the two schools of thought. Eventually, I argue that we need to also take into account other crucial aspects of food production, which are still missing in the present assessments.
4.1. The Optimistic View: Agricultural Land is Still Potentially Abundant
Experts from renowned international institutions, such as the International Food Policy Research Institute and FAO (e.g., [32
]), believe that a large amount of land can still be put into production. FAO’s scenario for 2030–2050 [32
] is based on the exogenous assumption that world GDP will be 2.5-fold the present one, and per capita income will be 1.8-fold. FAO assessments are characterized for being rather optimistic concerning the possibility of meeting future land use needs.
The report by FAO [61
] (p. 41) stated that “There is widespread concern that the world may be running out of agricultural land. … Despite these losses, there is little evidence to suggest that global land scarcities lie ahead. Between the early 1960s and the late 1990s, world cropland grew by only 11 percent, while world population almost doubled. As a result, cropland per person fell by 40 percent, from 0.43 ha to only 0.26 ha. Yet, over this same period, nutrition levels improved considerably and the real price of food declined. The explanation for this paradox is that productivity growth reduced the amount of land needed to produce a given amount of food by around 56 percent over this same period. This reduction, made possible by increases in yields and cropping intensities, more than matched the decline in area per person, allowing food production to increase
”. However, the report recognized that land scarcity exists at country and local levels and may affect food security, and that action should be taken. Concerning land degradation, the report points out that the area of degraded land is not known with much precision, as its assessment is usually based on the opinion of experts. In the case of India, for example, estimates by different public authorities vary from 53 million ha right up to 239 million ha. Furthermore, FAO [60
], argues that the impact of degradation on productivity is difficult to assess, as “Its seriousness varies widely from site to site over even small distances, and at the same site according to local weather, vegetation and farming techniques” [61
] (p. 42). Although FAO reports the principal types of land degradation, it does not consider land degradation in its models: “Because it is difficult to quantify, the future progress of land degradation was not taken into account in the projections made for this study
] (p. 42).
In a successive FAO publication edited by Bruinsma [98
] (p. 136), it is stated that “Concerning the future, a number of projection studies have addressed and largely answered in the positive the issue as to whether the resource base of world agriculture, including its land component, can continue to evolve in a flexible and adaptable manner as it did in the past, and also whether it can continue to exert downward pressure on the real price of food … The largely positive answers mean essentially that for the world as a whole there is enough, or more than enough, food production potential to meet the growth of effective demand, i.e., the demand for food of those who can afford to pay farmers to produce it
”. Recent FAO reports [32
] still provide a rather optimistic assessment of the potential agricultural land that can be put into production, although the authors are very aware of the complexity of the food system. They argue that the capacity of the world as a whole to produce food is only one aspect of food security and, actually, not even the most relevant one. They [32
] (p.10) state that “there are sufficient spare food production resources in certain parts of the world, waiting to be employed if only economic and institutional frameworks would so dictate
”. According to Alexandratos and Bruinsma [32
], there are some 1.4 billion ha of prime land (classed as very suitable) and good land (classed as suitable and moderately suitable) that could be cultivated in case of need. More specifically, the scenario indicates that the following classes of land are available: 350 million ha are very suitable, 600 million ha are suitable, 450 million ha moderately suitable, 560 million ha marginally suitable and 920 million ha very marginally suitable (Table 1
). An expansion of agricultural activities in the latter classes of land, however, may come at the expense of pastures, requiring considerable development investments (e.g., infrastructures).
Alexandratos and Bruinsma [32
] warn about the difficulties of knowing the actual land in use, as data for crops and historical data for arable land for many countries are particularly unreliable. Furthermore, data on cropping intensities for most countries are non-existent; for this study, the authors derived them by comparing data on harvested land, aggregated over all crops, with data on arable land, an issue already discussed in detail by Alexandratos [145
]. Alexandratos and Bruinsma [32
], however, point out that production constraints are and will continue to be important determinants of food security. Furthermore, they argue that increasing productivity may spur population growth resulting in reducing progress and locking the system into a poverty trap. In their report, the authors also stress the fact that, notwithstanding the availability of potential suitable land for the future expansion of agriculture, such land is far from being evenly distributed among the different regions (Figure 4
). Very highly densely inhabited regions of the globe, such as East and South Asia, the Near East and North Africa, have less of a margin for manoeuvring, considering that the future population growth will take away further land, lost to urbanization and infrastructure development. According to Gerland et al.
], the human fertility rate may not slow down as expected, especially for Africa, where culturally families still wish to have a high number of children (4.6 on average), and policies for family planning and women’s education and empowerment are still limited.
Some authors [32
] also warn that the economic growth of developing countries (e.g., China and Brazil) will prospectively increase meat consumption. This in turn will exert a further pressure on agricultural resources. Biofuels may also become a serious competitor for agricultural land. Some experts believe that in many regions of the world productivity is still very low and can be substantially increased [32
]. According to scenarios from Mauser et al.
], improving crop growth management through better technology and knowledge may result in a 39% increase in estimated global production potential, while a further 30% can be achieved by the spatial reallocation of crops to profit-maximizing locations. According to the authors, the expected increase in yield will make cropland expansion redundant, nor will it be necessary to rely on GM crops. It is easy to agree that with even minimal investments the average crop yield in many developing countries may rise. The proposed scenario, however, relies on the very optimistic assumption that better technology and knowledge will be available everywhere. However, better technology and knowledge come at a cost: at present, for many developing countries such investments are out of reach (in many cases, small farmers cannot even afford to buy improved varieties or minimal inputs).
4.2. The Concerned View: Soil Quality and Soil Degradation Greatly Affect Agriculture Productivity
The limited expansion of cultivated land, notwithstanding the doubling in population, led FAO to suggest that land scarcity may not be a serious problem. Nevertheless, the conclusions of other experts (e.g., as summarized by [24
]) are different and address the fact that there is not much suitable land left for further expansion.
It has been estimated that 70% to 80% of the Earth’s land area is unsuitable for agriculture owing to poor soils, steep topography, or adverse climate [146
]. About 50% of the remaining area is already being cropped, and a large proportion of the other half is presently covered by tropical forests, which beneficially take up CO2
]. According to estimates by Ramankutty et al.
], the total global extent of suitable cropland in the current climate is 4.1 billion ha, which is roughly 120% larger than the 1992 global cropland area of 1.8 billion ha. The greatest potential for croplands in the current climate exists in tropical Africa (560 million ha) and northern South America (470 million ha), which has also been pointed out by other authors. However, displacing tropical forest will cause dramatic problems concerning CO2
emission and biodiversity loss. Furthermore, tropical soils will lose fertility rapidly once the forest cover is replaced with a crop, and will require expensive inputs to maintain the soil nutrients and conserve soil organic matter (SOM).
A number of experts challenge the optimistic findings by FAO and other authors, on the basis that the data are affected by high uncertainty and some key issues, such as soil degradation, which are not considered (e.g., [6
]). Problems with the present models are of both a technical (due to the inherent limits of the process of mensuration) and methodological nature (related to the choice of what to measure, assumptions, and boundaries). Many such limitations have also been recognized by the authors of the scenarios themselves [32
]. We concede that some of those limitations are inherently very difficult to overcome (e.g., lack of data at a small scale, proper and reliable assessment of soil degradation; great difficulties can be met just obtaining reliable data on crop productivity for many regions).
] points out that the fact that in the past 50 years the population has grown by 110% and cropland by only 10% may be telling figures pointing to the fact that there is not much land that can be easily cropped. The expansion of soybean (300%) and palm oil (700%) is presumably due to the clearing of the Cerrado in Brazil and the rain forests in many tropical countries [24
]. Young [6
] warned that the very same statistics about yields might be unreliable, as in many developing countries there is not a real measure of the areas harvested, of yield or production. Figures may then be affected by assumptions (or even conditioned by speculative forces) rather than respond to realistic measurements (an issue also recognized by [32
Already in the late 1990s, Young [6
] and Smil [15
] pointed out that in many poor countries the amount of land under cultivation is more than reported in the official statistics submitted by those countries to FAO. Therefore, data used by FAO may greatly overestimate the amount of “free land” (in some regions potential agricultural land is virtually non-existent), with the true remaining balance of cultivable land being much smaller than what is reported by these scenarios. A recent work by Lambin et al.
], based on a bottom-up approach (using direct expert knowledge), supports the claims by Young [6
]. The experts considered land availability, specific constraints and trade-offs. Their figures about the potential agricultural land for a number of regions considered are just 15% to 65% of those provided by the previous FAO assessments (only for the case of Amazonia were their estimates higher: 168%). Lambin et al.
], argue that a bottom-up approach is better able to consider more fine-grained, up-to-date, and locally relevant criteria to estimate agro-ecological suitability, current land use/cover, and the constraints and trade-offs associated with land conversion. By adopting such an approach, it is possible to provide more realistic figures compared to the global datasets. The authors point out that the drawback of the bottom-up approach is a lack of consistency in the criteria used to define the potentially available cropland, as each expert provides a judgement based on available data, current land use dynamics, and the social and political context of the region. As a result, the costs and benefits of land conversion are not strictly comparable across regions.
Bindraban et al.
] point out that although experts tend to agree on the fact that about 25% of the global land area is degraded, there are large differences concerning the estimates of the intensity and extent of soil degradation. This is due to the different definitions, methodologies applied and lack of on-the-ground validation. Furthermore, the authors argue that the large-scale assessment of the impact of degradation on plant production is also inaccurate, as it suffers from the specific opinion of the experts, or is based on statistical procedures that do not allow extrapolation in time or space.
The physical availability of arable land is only part of the story; the human role in the decision-making process has to be fully taken into account [3
]. This was a point that other authors also made for related fields such as farming system analyses and rural development (e.g., [45
The critics maintain that soil degradation reduces both actual and potential yields. In some cases, soils are already so degraded that they cannot be cropped and have to be used for pasture instead. Concerning further land expansion, critics point out that irrigation would be needed in order to achieve high productivity of the new land; however, water availability is instead becoming scarce (irrigation will also increase the operating costs of farming). Soil degradation is indeed a relevant issue because it affects land productivity directly, by reducing yields, and indirectly by increasing management costs (e.g., fertilizers, irrigation). For small and poor farmers, economic investments are coupled with indebtedness (Figure 3
). Limiting loans to a minimum is actually part of a strategy actively pursued by farmers in poor countries to reduce risks of indebtedness [6
]. This will play against farmers undertaking large investments to provide more inputs for their fields. Agricultural intensification (and the increased use of inputs), in many cases may actually indicate that farmers cannot move on new fertile land, and have to cope with soil degradation instead [6
It has to be pointed out that the African continent, where the largest share of the demographic growth is expected to take place, is also the most fragile in terms of soil composition [4
]. More than half the global population growth between now and 2050 is expected to occur in Africa. The continent is growing at a pace of 2.5% annually (in 2010–2015 figures), the highest rate of population growth among the continents [34
].The soils of Africa are derived mostly from parental materials that have been long exposed, hence are highly weathered and leached, and are characterized by inherently low productivity. The soils of Africa, therefore, may also be those most vulnerable to drought and other stressor such as those induced by intensive agriculture. Young [6
] warns that the vision of converting the Amazon and Zaire basins into Asian-type rice-lands stems from a misunderstanding of the different biophysical soil characteristics of the former in comparison with the latter.
4.3. Trends for Arable Land 1980–2010: The Complex Relation between Land, Population and Economic Growth
Arable land is regarded as the best and most productive land available; it is generally cropped with cereals or highly profitable crops. Therefore, a reduction in arable land on a per capita basis may affect food production and, most importantly for the poor countries, food security. A reduction in arable land per capita generally forces a country to increase land productivity (increasing the inputs), expand the land used for production (possible marginal land, or converting forests) and/or rely on imports.
In this section, using the data from the World Bank (WB) database [151
], I explore how the change in arable land, for the period 1980–2010 (Figure 5
), correlates with changes in GDP per capita and with population growth (Figure 6
, Figure 7
and Figure 8
). I eliminated from the country dataset a few micro countries, whose socioeconomic peculiarities made them unrepresentative. I used the data starting from 1980, to be able to include data for 111 countries of the WB database, and to be able to consider a relatively large time frame. It would have been possible to include a few more countries by taking the year 1990 or 1995 as a reference point, but that would have reduced the time frame, thus affecting the perceived level of the land change. I wish to point out that the goal of this exercise is not to provide any definitive evidence, but rather to highlight the fact that land, population and economic growth are interwoven in a complex way.
The data plotted in Figure 6
seem to indicate that the loss of arable land per capita is not related to the increase in GDP per capita. Land loss (−20% to −60% of arable land per capita) affected countries whose GDP per capita increased from five to ten times from 1980 to 2010. The loss of arable land is directly related to both demographic pressures (need for food and urbanization) and economic pressures (urbanization, industrialization, complex infrastructures, financial speculation). Therefore, while the increasing GDP per capita is usually linked to reduced demographic pressure, there was increased land use change because of the new socio-economic forces taking place (e.g., urbanization, industrial settlements, and speculation).
In the early 1990s, some environmental economics scholars suggested the existence of an environmental Kuznets curve (EKC), i.e.
, that some environmental degradation indicators tend to worsen as modern economic growth progresses, until average income reaches a certain point over the course of development, after which they tend to improve. EKCs have been found for some environmental pollutants (e.g., sulphur emissions), but not for others (e.g., energy, biodiversity). The existence of EKCs has been challenged by some recent empirical work, and also on theoretical grounds (for a discussion and criticisms see [152
]). If these criticisms were correct, we would expect that those countries that most increased their GDP per capita would present a lower loss of arable land per capita. Nevertheless, the application of this concept to land is somewhat complicated, as the reduction in the rate of land conversion may depend much more on the fact that very little land is left to be converted, or that what is left is of very poor quality and very costly to convert (decreasing the marginal return on the conversion process).
reports the change in arable land per capita and the change in the growth of GDP per capita for the years 1980–2010 for twenty of the richest countries of the database (Figure 7
a), and for twenty of the poorest countries of the database (Figure 7
b). There are no notable differences between samples.
The data plotted in Figure 8
present the relation between the change in total arable land per capita and the change in the population. From Figure 8
, it seems that there is no relation between the reduction in arable land per capita and population growth. A reduction in arable land per capita is evident both where population pressure increased as well as where it was more contained.
As has been argued by many authors (e.g., [3
]), land use change is a complex matter and cannot be simplistically attributed to a specific factor, be it population pressure or poverty. Van Vliet et al.
] made the point that scholars tend to generalize, and assume a unidirectional relationship between land use change and its impact. Nevertheless, as humans promote (and adapt to) changes in land use, a variety of consequences are possible, and the issues have to be understood within the specific environmental and socioeconomic context.
The data plotted in Figure 9
seem to indicate that there is a relation between the reduction in arable land per capita and the increase in total arable land. As previously reported, it has been estimated that about 30% of the increase in agriculture production comes from the expansion of cropping land. Of course, there is an increase in demographic pressure in some contexts (e.g., where low income makes it difficult to pay for the inputs need to intensify production, or where there are policies that support land conversion).
Countries from the dataset that lost more arable land per capita from 1980 to 2010 are (ranked in order of quantity): Nepal (−86%), Senegal (−85%), Jordan (−79%), Chile (−78%), Colombia (−71%), Botswana (−69%), Yemen (−68%), Honduras (−67%), Lebanon (−67%), Iraq (−66%), and Tonga (−66%). Countries from the sample that increased their arable land per capita from 1980 to 2010 are (ranked in order of quantity): The Netherlands (49%), Macedonia (46%), Fiji (35%), Portugal (31%), Uruguay (25%), Bolivia (15%), Nicaragua (10%), Ghana (8), and Gambia (1%).
For many low-producing countries, there may still be large margins for increasing crop productivity; nevertheless, this would require farmers to conduct a large economic investment in high quality seeds, agrochemical inputs, and possibly irrigation, which would be a difficult challenge to meet. Eventually, soil exhaustion and soil degradation may also affect those countries that cannot afford to use a large quantity of inputs. Notwithstanding such potential margins, the increasing demographic pressure in most countries will present some great challenges (see for instance the different contexts that characterize countries such as Bangladesh and the Middle East, where the expansion of agriculture is no longer an option, and countries such as Brazil and Nigeria, which can still convert their forests to agriculture or pasture land). Table 2
reports on the pressure on arable land and future demographic trends for the 10 most populous countries. As can be seen from Table 2
, all these countries have experienced a reduction in arable land per capita in 1980–2010. It can be noted that the poorest and most populous countries are also those that are already experiencing a shortage of arable land (0.2<) and a high rate of population growth.
As has been previously discussed, the figures in Table 2
may actually present an underestimate of the situation. Although we usually associate population pressure, lack of land and poor yields with poverty and hunger, paradoxically there are cases where a large part of the population can experience hunger in a country with high arable land per capita, top yields and the most advanced technology, and the highest GDP per capita in the world. The USA is a striking example of this seemingly inexplicable paradox. The USA has as much as 0.5 ha of arable land per capita, one of the highest value in the world. USA agriculture is the most advanced and productive on the planet. Large subsidies and cheap energy (compared to the EU, for instance) guarantee the lowest cost of inputs. Surplus has to be disposed of. Most farmers, in their rush to boost productivity, eventually become worse off, with only large farmers benefiting from the economies of scale and huge amount of agricultural subsidies granted by the government [40
]. The historian of US agriculture, Willard W. Cochrane [157
] talks about a continuous problem of surplus for US agriculture since the XVIII century. A surplus that brought little benefits to farmers (most of which went out of the market, indebted and unemployed), and forced government to continually intervene with policies aiming at surplus removal. Douglas R. Hurt, another renowned scholar on US agriculture, in his history of US agriculture [158
], tells of a continuous “problem of plenty”; since the 1930s, the main agricultural issue that all governments have been faced with is how to get rid of the surplus. The subsidies-surplus treadmill, typical of US agriculture, has grown so large to affect agriculture and the food system on a global level [40
Burning the surplus, producing “green fuels”, seems the final solution. Nearly half of USA maize production ends up generating ethanol [160
]. Biofuels have a low Energy Return On Investment (EROI) when compared to fossil fuels (about 10 to 30 time lower) [29
], and, what is more important, a very low power density (W.m-2
, about 1,000 to 10,000 times lower) [29
]. These characteristics render biofuels an energy carrier that is highly demanding in term of investments and labor. In turn, they have to be highly subsidised to be sold in the market at an affordable price for consumers [40
]. Koplow and Steenblik, [171
], estimate that in 2008, in the USA, total support towards ethanol production ranged between 9.0 and 11.0 billion US$. These figures are likely to be an underestimate, given the many faces economic support can take (from tax exemption to price premiums), making precise subsidy assessment a difficult task [171
] (see for instance the long list of State and Federal Laws and Incentives to support biodiesels and ethanol: U.S. Department of Energy [175
]. According to Reboredo et al.
], the present low price of oil will require “… a massive public/state subsidy flow
…” to sustain biofuels [177
] (p. 5). Concerning Genetically modified crops, which, it is claimed, should solve the problem of world hunger, according to the USDA [178
], in 2013, U.S. farmers used herbicide tolerant (HT) soybeans on 93% of all planted soybean acres, HT corn accounted for 85% of corn acreage and HT cotton constituted 82% of cotton acreage. Bt corn (which controls the European corn borer, the corn rootworm, and the corn earworm), was planted on 76% of corn acres. Other GE crops commercially grown in the United States are HT canola, HT sugar beets, HT alfalfa, virus-resistant papaya, and virus-resistant squash [178
]. According to GMO compass [179
], in 2013, HT sugar beets accounted for 95% of the acreage, HT canola for 93% and HT alfalfa for 30%. In total, in 2013, GM crops accounted for about 70% of cultivated land. Yet, notwithstanding the high-yielding strategies listed above, the USDA reports that in 2012, 14.5% of households (about 45 million people) were food insecure, meaning they had difficulty at some time during the year obtaining enough food due to a lack of resources [180
]. On the other hand, more than one-third (34.9% or 78.6 million) of U.S. adults are obese, and 68.5% of adults are overweight or obese. The future looks even darker as 16.9% of children are already obese, and 31.8% of children and adolescents are overweight or obese [181
]. Nevertheless, in the USA, when agriculture is discussed, social issues (e.g., the huge wealth inequality, the power of corporations) are never addressed. Agricultural productivity seems to be the only problem worth attention. Given the incredible performance of USA agriculture, and the vast amount of waste created by the USA food system, we may be tempted to conclude that focusing the problem on the need to produce more food serves to hide more important social issues, which are taboo, as they concern the system of power. We cannot but agree with Daly [102
] that the “growth paradigm” has come to be viewed as a solution in itself: whatever the problem, “growth” is thought to be the solution. Probably the fact that “growth” is seen as an easy solution to very complex problems (along with our natural attitude to take the easiest path) keeps our thinking locked in such a paradigm, preventing us from exploring different solutions.
4.5. The Necessity to Embrace a Precautionary Approach and to Adopt Novel Modeling Tools
As we have previously seen, the present models suffer from a number of limitation that affect their ability to provide sound and effective scenarios. Many experts argue that the present models tend to overestimate the real availability of agricultural land. These issues concern the assessment of both soil degradation and land use.
Concerning soil degradation, we have the following issues:
“soil degradation” is a broad definition, including many processes that affect the soil in different ways and to different extents; a unique definition of soil degradation is missing;
there is a lack of objective criteria to define soil degradation (soil and land degradation are often used as synonyms, although they are not); in most cases, different processes take place at the same time, making the enterprise very challenging;
it is difficult to gather basic data, and the figures provided by many local and national institutions are affected by high uncertainty and unreliability. FAO [57
] argues that the quantity and quality of information on soil degradation is very variable in different regions, and that great differences exist between countries in data and data availability on soil resources and soil change information.
Concerning the scenario analysis of land use, we have the following problems:
Uncertainty in the basic data
as FAO [57
] (p. 8) argues, “Crop models, especially when run at global scale, are highly complex models that differ widely in terms of process representations, functional implementations, data input choices and basic assumptions. Even with the same version of the same basic underlying mode, … results often differ substantially
the difficulty to know the actual land in use, its quality and the real productivity of the crops; as Alexandratos and Bruinsma [32
] warn, for many countries data are unreliable or even non-existent). Unreliable data may also concern other domains such as economics, inputs and, in many countries, the population itself;
data from different models are difficult to compare as they rely on different assumptions, boundaries and protocols;
land use is mapped at a scale that does not account for the real morphology, features and use of the land (e.g., hilly and rocky outcrops), leading to gaps in basic data;
the amount of land occupied by the people themselves is not properly accounted for;
the analysis underestimates the amount of land that is actually cultivated (e.g., illegal land occupation, forest use);
the lack of integration among the different domains that characterize food production, such as the future scenarios of water and energy, the fate of some key elements (e.g., phosphorus);
most of the land that FAO includes as potential cropland is actually represented by rain forests, grazing land and marginal land that may be providing ecosystem services;
Oversights in the description of key issues
soil degradation is not taken into account, yet it greatly affects productivity and land conversion [71
] provide a review on this issue);
the effect of climate change, the changes in water and energy supply are poorly (or not at all) included in the scenarios. It has to be stressed that the cost of inputs (and therefore the price of energy) is a key issue for maintaining high agricultural productivity;
the effects of trade and globalization bring a lot of uncertainty to the agricultural sector in different regions/countries. Other socioeconomic issues are not considered either (e.g., credit, financial speculation, conflicts);
the effects of future social and economic trends, which will pose great pressure on existing resources and on the resilience of the social fabric.
Possible approaches to provide better information and improving scenarios:
Some scholars argue that assessment of soil and land degradation is made difficult by the existence of different definitions of those processes. That makes it also difficult to develop methodologies able to provide comparable information. Therefore, it is important to have scholars working together to frame and solve this issue.
It has been argued that often the information available are based on rough estimates (at times mere guesses). An effort thus should be taken to gather more on-the-ground information upon which better models can be developed. This is much needed especially in those highly populated regions where food security is at risk, and where social conflicts may be exacerbated by lack of resources.
Gibbs and Salmon [83
] reviewed the methods presently in use to assess land degradation, namely expert opinion, satellite-derived net primary productivity, biophysical models, and abandoned cropland. The authors argue that no single estimate accurately captures all degraded lands, but each one contributes to the overall discussion. Gibbs and Salmon [83
] make the case that even if a precise map of the physical area of degraded land were to be produced, it does not suffice, on its own, to provide sound information on the actual produce potential of the land. Considering the land as detached from the other many environmental, social and political constraints may result in highly overestimating its productive potential.
It is becoming clear that in order to better assess farming and food system performances, a more complex approach to scenario analysis and land use management is needed [99
]. Steps are being taken towards a “Nexus approach”. International institutions, such as FAO (e.g., [57
]), UN (e.g., [188
]), IIDS (e.g., [189
]), along with many other scholars (e.g., [87
]) are promoting and working on this line of research, as it is believed to better respond to the complex challenges we are called to deal with. Howells et al.
] attempts to broaden the approach to integrate the effects of climate change on land, energy and water use (Climate, Land use, Energy, and Water; CLEW-model). Giampietro et al.
] are working on the nexus approach to jointly study resource use and the metabolism of societies (Multi-Scale Integrated Analysis of Societal and Ecosystem Metabolism; MuSIASEM-model). Giampietro et al.
] point out that in order to provide a better framing of the complex problems of our societies, it is necessary to integrate different narratives used in quantitative analysis. That is to say, we have to use technical, economic, demographic, social and ecological variables simultaneously, defined on different hierarchical levels and scales. In this way, it is possible to generate better and more complex quantitative representations of the viability and desirability of policies and technical solutions.
Nexus seems to be a promising approach for a better analysis of farming systems, as it allows us to study in a holistic way the interplay between the biophysical factors, socioeconomic forces and metabolic characteristics of societies, taking into account the constraints, potential, possible risks and bottlenecks. Such an approach would also be able to address the participative nature of sustainability, by involving stakeholders and raising awareness of the problems and trade-offs involved in the different solutions.
Concerning long-term scenarios, it is very difficult to know how the world will be in 2050! In the 1950s nobody could have imagined that in the 1990s we would have high-speed computer and internet, and with it a new world. Thus, for good or for bad, we must take into account the inherent uncertainty of the future due to unforeseeable problems, which are unforeseeable because we just ignore that they could exist [15
]. Of course, we may also make some novel discovery that could help us. Nevertheless, to play it safe, it is better to focus our attention on how to handle potential problems rather than hope for miracles to happen. Given the limitations of the models, the uncertainty about the real situation, and the high stakes at play, a precautionary approach must be adopted when carrying out soil assessment and producing scenarios. Enhancing the awareness of policy-makers and the public at large constitutes an essential contribution by soil and agriculture scholars.