• No results found

AGROECOLOGICAL LANDSCAPES, SFRA FOR 2009 30

3. Case study

The selected area of study covers the four municipalities from chapters 2, 3 and 4. They are located in the historical county of Vallès: Sentmenat, Caldes de Montbui, Castellar del Vallès and Polinyà. In this case, the selection is conditioned by the existence of a study on the agrological

33 Other important issues such as flood control, water purification and climate regulation, which also depend on the configuration of the agroecosystem, are not considered here. This is because they do not fall within the scale or capability of the model and because they form part of the overall social objectives beyond the reproducibility or not of the agroecosystem (Bagstad et al., 2013; Vihervaara et al., 2010).

Chapter 7. Possible horizons of agroecological landscapes, SFRA for 2009

suitability for different crops (Rodríguez Valle, 2003), as well as access to cartographic information on the composition of soil uses in the mid-nineteenth century (Marull et al., 2008).

Because of the issue of limited sources, and its character of test bench, we take these four municipalities as the unit of analysis for the agroecosystem. For future approximations of the SFRA model, however, it would be desirable to define the unit of analysis more precisely, taking into account the aspects noted in the previous section. As well, in this chapter, instead of using data from 1999 we adapted the study to the next agrarian census, done in 2009.

Figure 7.2 shows the distribution of land-uses for that year. The most predominant use is woodlands and scrublands, which cover 62% of the total surface area, primarily in the northern and western region of the four municipalities. Pinus halepensis, or Aleppo pine, covers three-quarters of the woodlands and scrublands, while the remainder features Quercus ilex subs. ilex (or Holm oak), shrubs and riparian or other forests. Urban sprawl has had a severe effect on the landscape. Nearly 19% of the land is now dedicated to urban areas (16%) or to other infrastructure or unproductive agricultural uses (rocky areas, ravines and riverbeds; 3%). Between 1956 and 2009, cropland was cut practically in half as a result primarily of two forces: first, the urban area quintupled, swallowing up formerly irrigated, high-quality cropland; and second, a slight process of forest transition subsumed marginal lands that could not be mechanized or that had been abandoned. The decline in cropland has resulted in a subsequent loss of agro-silvo-pastoral mosaics, which are the typical cultural

landscape of the Mediterranean region (chapter 3). Within this landscape, grapevines represented the main crop before the advent of the Phylloxera plague, but now they occupy little more than 0.7% of land under cultivation. By contrast, the predominant use of cultivated land at present falls to crops for animal feed, which account for 75% of all crops, followed by fruit trees and olive trees.

The predominance of crops for animal consumption is consistent with the high livestock density of the four municipalities. While the livestock density has fallen 40% from the peak recorded in 1999, it continues to be extremely high at 111 LU500/km2. This figure is absolutely exorbitant in comparison with the level of 7 LU500/km2 in 1860.

As for population, the proximity of the Barcelona metropolitan area resulted in an explosive upward trend until the middle of the last decade, when residents numbered 55,433 and the population density was 462 inhab./km2, with only 0.25% of total population dedicated to farming. The high population density stands in stark contrast with Spain’s national average of 92 inhab./km2 and with the 1950 average of 100 inhab./km2 in the case study.

Obviously, the selection of this case study for an analysis of the potential development of agroecological landscape strategies has some particular characteristics relating to high population density that must be taken into account when drawing conclusions. As we propose, however, these studies should be done with local particularities taken into account so as to proceed

Figure 7.1. Land-use map for the Vallès case study in 2009.

Source: Our own, adapted from CREAF (2009).

Chapter 7. Possible horizons of agroecological landscapes, SFRA for 2009

subsequently to a networked analysis of the capacities of regions. It is necessary to take it as an example, therefore, without seeking to extrapolate the results to larger scales.

4. Methodology

Based on the structure of funds and flows defined in Figure 7.1, we put forward the main characteristics of the present-day SFRA. The modeling will be done through a mathematical approximation using non-linear programming. Below is an explanation of how the programming, the sensitivity analysis and the analysis of the results will be carried out.34

The objective will be to compare several scenarios, varying the analysis on two main axes: the type of diets provided by the agroecosystem (production strategy to meet social needs) and the amount of land for crops (whether the area under cultivation can be increased or not).

4.1 Programming the non-linear optimization model

The methodology of non-linear programming identifies the best possible result for a system based on the maximization or minimization (optimization) of a function with a finite number of variables and constraints that may be linear or non-linear. The constraints can be affected by what are known as boundary conditions, which are certain assumptions that can vary as a function of the desirability explained in chapter 5. Thus, to identify a range of scenarios under different assumptions, the boundary conditions and the function to be optimized can both be modified. We considered to run the structure through the RStudio program using NLOPT_LN_COBYLA for the different scenarios. However, while we attempted to follow this procedure, we ultimately ran the model through the GUSEK program with the SIMPLEX linear algorithm for technical reasons, adding parameters that allowed, by means of iteration, to obtain similar results to those from applying a non-linear program. In total, the model has 1,417 variables, 560 constraints and 3 different optimization functions.

4.1.1 Boundary conditions

The non-linear approximation has strong implications for the SFRA’s initial conditions.

Here, the model will define the appropriate livestock density, the structure of land-use (and, therefore, the composition of farmland), and the size of the population that can be fed by the agroecosystem itself. While the size of the domestic unit and the number of livestock were preset in the 1860 SFRA, the size and composition for the three self-reproducing funds in this case are outcomes of the model. Non-linearity also permits a reduction in the number of assumptions that in the earlier approximation in chapter 6 constrained the model’s degrees of freedom. However, several other boundary conditions are set that will affect the general structure of the new SFRA.

In relation to population, the nutritional and energy needs depend on age, sex and physical activity level. Food production must meet requirements according to the age pyramid for the set of municipalities based on census data (IDESCAT, 2009a), taking into account the current estimation of energy expenditure as a function of physical activity level (ESFA, 2017) and the average physical activity level for each age band in the Catalan population (Ministerio de Sanidad, 2014). This will be considered a stable relationship35, which makes it possible to set the average energy requirement of the population at 2,256 kcal/day per person.

In the case of livestock, we use three distinct species: pigs, chickens/hens and sheep. Each type is representative of one of the functional categories identified in section 1.2.

34 The complete model and all the associated assumptions appear in Annex III.

35 That said, if the reproduction rate is not reversed, the current ageing processes in the population could result in a shift in the proportions of the age bands and in a decline in population density, at least at a national scale (INE, 2014).

Chapter 7. Possible horizons of agroecological landscapes, SFRA for 2009

Of the limits defined regarding the soil fund36, the most clearly determinant one is the unit of analysis, that is, the amount of land that can exist for each use. Of the 11,996 ha in the four municipalities, 2,237 ha are urban or infrastructure and 30 ha are riverbeds and rocky outcrops.

These do not form part of the agricultural metabolism. By contrast, the remaining 9,729 ha do play a role. Here the boundary condition regards to forest area. We calculated that 5,528 ha have to remain as forest from figures on nature conservation land and also from stable woodland areas over the past 150 years. As a result, this land is not subject to changes of use because of its value37. For the final remaining 4,201 ha, the freedom exists to change use under various precepts of a technological and agronomic nature, which will be used to define the constraints later on.

Lastly, in the case of herbaceous crops, three possible rotation types are laid out. These types reflect proposals made in a nearby organic agricultural park (Safont & Artal, 2008) and recommendations on herbaceous crops native to the territory (Tuson, 2011, 2009).

Rotation 1. Montcada wheat – Chickpeas – Spelt – Lentils – Green Manure Rotation 2. Mustard – Triticale – Fenugreek – Green Manure – Barley

Rotation 3. Potatoes – Broad Beans (Favas)

The various agricultural uses viewed as possible within the model are indicated in Table 7.1. The yields from these uses are primarily estimations based on information from the agricultural census of 2009 (IDESCAT, 2009b) adapted to the conditions of organic cultivation to avoid overestimating their productive potential (De Ponti et al., 2012; Seufert et al., 2012).

Some other specific information comes from yields of the area’s own ecological farmland or from studies in comparable Mediterranean conditions (Consorci de Gallecs, 2010; Tuson, 2009). In the case of woodlands, the levels of productivity are estimated based on the annual growth in the aerial biomass of the dominant species (CREAF, 2007), while the pasturage potential is estimated using information provided by Robles (2008) and Taüll & Baiges (2007).

Table 7.1. Possible land-uses considered by the model SFRA for 2009. Source: Our own.

Irrigated Herbaceous crops Woody crops Pastures Forest Vegetables Rotation 1 Olive oil

Improved pastures

Holm oak

Rotation 2 Wine Pines

Fresh fruit

Rotation 3 Almonds Other forests

and brushland

As was the case with livestock, the proposed crops are also indicative and representative of the different categories. We reiterate that it would be necessary to consider other crops and rotations to ensure the system’s diversity and resilience. However, this would refer to the scale of plot, when the approximation in the present case is at the scale of landscape.

36 As in the first SFRA, boundary conditions and constraints regarding surface are associated to the soil fund because it is the only one fund with territorial expression (unlike society or livestock).

37 While it is true that this criterion did not generate consensus among the people consulted, the lack of time to discuss the results has impeded the possibility of undertaking other approximations. In any event, we consider that it is a conservative criterion for a first iteration of the model, whereas other approaches would probably result in greater farmland expandability.

Chapter 7. Possible horizons of agroecological landscapes, SFRA for 2009

4.1.2 Constraints considered

Once the boundary conditions are defined, it is necessary to indicate in the SFRA what constraints are set by the model’s socioecological limits. Starting from the flows in Figure 7.1, we proceed to indicate the constraints these flows entail for ensuring the reproduction of living funds.

Constraints for society

Diet is calculated differently according to the conditions set for desirability. Three scenarios are considered. The first one applies the current diet (MAGRAMA, 2009). The second one is expressed in the form of constraints to obtain a healthy diet. The main initial conditions of the healthy diet come from a study on the cardiovascular benefits of the Mediterranean diet (Estruch et al., 2013) and various nutritional criteria that strike a balance between sources of proteins, fats and carbohydrates (SENC, 2016). The third scenario is defined by a constraint that enables the maximization of total food output in the agroecosystem in terms of metabolizable energy. This would allow us to follow the objectives set in section 1.

Domestic residues that are returned to close nutrient cycles are estimated through the actual diet provided and a number of technical factors concerning the consumable fraction of these foodstuffs (Farran et al., 2004).

Constraints for livestock

The largest part of the defined constraints regards to livestock. We distinguish between input flows (feed, stall bedding) and output flows (manure, food). Some fundamental initial constraints are to link livestock in the different stages of their life cycle and to their rates of reproduction. In total, the three types of species are defined as having 31 stages in which their requirements and flows are different.

As for animal feed, this is one of the areas with the greatest uncertainty because of the products and by-products that animals can consume. Based on a review of recommended animal diets in organic production, therefore, we set the required energy consumption in terms of metabolizable energy (ME) and the minimum and maximum crude protein (CP) contained in the diet. Then the data on crop yields are transformed into ME or CP with technical coefficients taken primarily from Church (1984), and the general constraints on feed are defined. In addition, to ensure that the results are valid from a physiological standpoint, maximum thresholds are set for the incorporation of specific kinds of feed that could cause problems either because they contain antinutritional factors or because of their palatability (FEDNA, 2010).

For stall bedding, we use the criteria of Soroa (1953), and for the manure produced, we estimate the composition through an iteration process starting with data from ASAE (2000) and then checking results with a literature review.

Constraints for soil fund

With regards to the soil as a fund, constraints are established for crop rotations, for the total amount of land as a function of preceding use and crop adaptability to the soil, and for nutrient balances. The first constraints are simply equalities that must be fulfilled so that a specific use (e.g., Montcada wheat) has the same amount of land as the other uses that correspond to the rotation (e.g., the surface area of Montcada wheat must be equal to the surface area for chickpeas, spelt, lentils and green manure).

Chapter 7. Possible horizons of agroecological landscapes, SFRA for 2009

The second group of constraints relate to the possible uses that can be established as a function of the preceding use (use for 2009) and the uses that can follow. In this respect, based on the cartographic data on preceding and subsequent uses (CREAF, 2009), protected spaces (DTS, 2017), land-use maps for 1860 and 1956 (Tello et al., 2004) and crop suitability and slopes in the territory (Catalan Geological Institute, 2010; Rodríguez Valle, 2003), each point on the map is translated into a specific category that denotes the possibility of one use or multiple uses (Xae). From this data, we define the constraints that set the amount of land in each category.

As noted earlier, two distinct scenarios can be considered: one in which the amount of land of current woodlands (7,407 ha) must be maintained or increased and another in which it is only necessary to maintain the woodlands that have been given special conservation status or have been woodlands for the past 150 years (5,528 ha). As a result, the second scenario offers the possibility of increasing farmland by 1,879 ha above the current figure. Depending on the defined scenario, one set of constraints or another is activated. The two sets are labelled I and II, respectively. It must be also added that woodlands that are not dominated by Pinus halepensis or Quercus ilex subs. ilex are not considered suitable for a change of use either. This is to protect habitats that have little presence in the area (e.g., deciduous trees, oak groves, stone pine, European black pine or riparian forests).

Lastly, there are constraints concerning the maintenance of biogeochemical cycles, which can be closed in four main ways: incorporation of surplus biomass from crop residues, manure, domestic residues, and also the possibility of incorporating a certain amount of green forest biomass.

To generate these constraints, first the nutrient extractions of nitrogen, phosphorus and potassium (NPK) are determined for all crops. Then all other inputs and outputs of nutrient flows (volatilization, denitrification, weathering, leaching, etc.) are determined and the total fertilization requirements are set for the re-establishment of chemical fertility. Next follows the characterization of all the various nutrient sources (manures, crop residues, domestic residues and imported green biomass) in what is understood to be a process of joint composting. Once all the sources are defined, the constraint is established so that the incorporated material in NPK terms meets the NPK requirements resulting from the total needs of different land-uses. The construction of this nutrient balance takes into account the criteria defined by González de Molina et al. (2010) and the IPCC (2006a, 2006b) among other sources.

Constraints on farm-associated biodiversity

Lastly, we indicate the constraints in relation to the final fund, the farm-associated biodiversity. The focus here is on non-domesticated species. In this case, however, we are speaking of beta biodiversity (Gliessmann, 1998). The question, therefore, does not concern agroecological practices carried out at the scale of plot, but at the scale of landscape.

The debate on biodiversity is complex and the hypotheses are still open on the best ways to conserve it. The perspective adopted here is that of a land-sharing strategy (Fischer et al., 2014), according to which the best way to maintain or increase beta biodiversity is through the establishment of spaces in which productive human activity is combined with a degree of intermediate disturbance that does not impede the presence of diversity (Loreau et al., 2003;

Perfecto and Vandermeer, 2010). This approach is not incompatible, however, with the need to conserve certain spaces to prevent the degradation of priority habitats in which specialist species need low or null levels of disturbance. To this end, we also consider it necessary to maintain certain protected conservation areas as indicated earlier.

Therefore, there is a clear link between landscape patterns and biodiversity (Tscharntke et al., 2012). Agroforest mosaics provide a range of habitats that can sustain many species (Harper

Chapter 7. Possible horizons of agroecological landscapes, SFRA for 2009

et al., 2005; Pärt and Söderström, 1999). For this reason, we have selected a widely used indicator as an approximation of habitats for biodiversity through landscape patterns: the Shannon index, adapted for agrarian metabolism as in chapter 3 (Vranken et al., 2014). The Shannon index, when analysed together with human disturbances values, has shown that the portion of associated biodiversity in agroforest mosaics is highly significant in the region, while another important part accounts for much less disturbed areas (Marull et al., 2018).

However, the main problem with indicators and landscape patterns, is that they do not have threshold values by which to ensure a specific level of biodiversity. At present, they only permit comparative assessments. Thus, as the only constraint possible, we will consider that values of the Shannon index for the resulting landscape and, therefore, the equidiversity of land covers must be greater than the value of the index in the original landscape of 2009.

4.1.3 Defined scenarios

Once the mathematical structure of the model was defined, it was run for various scenarios. As can be seen in Table 7.2, six different scenaries are defined on two axes. The first dimension is the possibility of bringing woodlands under cultivation to increase cropland, while the second refers to diet, or food intake.

In the case of diet, we consider three different situations, which will determine the objective functions. First, the aim is to look at the potential for developing food provisioning

In the case of diet, we consider three different situations, which will determine the objective functions. First, the aim is to look at the potential for developing food provisioning