Advances in Water Resources 130 (2019) 258–269 Contents lists available at ScienceDirect Advances in Water Resources journal homepage: www.elsevier.com/locate/advwatres Relational analysis of the resource nexus in arid land crop production Violeta Cabello a,∗, Ansel Renner a, Mario Giampietro a,b a Institute of Environmental Science and Technology, Universitat Autònoma de Barcelona, 08193, Cerdanyola del Vallés (Barcelona), Spain b Isnstitució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Catalunya, Spain a r t i c l e i n f o a b s t r a c t Keywords: While a large number of descriptive studies have delineated the interlinkages between water, food and energy Nexus networks resources in the last decade, there is still need for systematic conceptualization of resource nexus interconnec- Metabolic processor tions. This paper proposes a theory of relational analysis of the nexus based on the analytical concept of nexus Relational analysis networks. A taxonomy of nexus interconnections, detailing sequential and hierarchical connections, is charac- Water-energy-food nexus terized between and amongst the technosphere and biosphere. We illustrate the use of a novel diagnostic tool Multi-functionality with regard to its ability to integrate macro-, meso- and microscale drivers of nexus problems. We apply this framework to problems generated by intensive crop production for exportation in an arid landscape (driven by external markets) and sustainable management of water resources (driven by public policies) in a southern Span- ish region. We elucidate interconnected causal mechanisms for groundwater overexploitation and profile different social-ecological patterns on a spatially-explicit basis. The proposed approach is capable of accounting for the water-energy-food resource nexus in an integrated and multi-level fashion, addressing the tensions generated by both multi-functionality and resource entanglement in complex social-ecological systems. 1. Introduction erned by different public policies and private markets with ‘silo’ dy- namics frequently expressing conflictive goals (Muranetto and Witmer, Mainstreamed by international calls for securing resources pressured 2017; Venghaus and Hake, 2018). Therefore, a need is claimed for ‘clos- by entangled global drivers, the water-energy-food nexus notion has ing the governance gap’ and improving policy coherence (Weitz et al., gained momentum at an increasing rate in sustainability research and 2017). Whereas the macroscale is epitomized by this interface between policy agendas. Interconnections, synergies and trade-offs are common policy and market constraints and economic sectors, the mesoscale is keywords within the nexus concept, where the diversity of definitions represented by the interface between those sectors and the water, food has gathered into at least four perspectives: (i) the nexus as governance and energy resource management domains they connect. Resource man- approaches that seek coordination and harmonization (Nilsson et al., agement domains are associated with regional spatial scales, since they 2012; Weitz et al., 2017); (ii) the nexus as the co-occurrence of re- are responsible for controlling the level of extraction of primary re- source use in economic sectors and supply chains, also referred to as sources. On the other hand, economic sectors are networks of enti- the resource-nexus (Font Vivanco et al., 2018); (iii) the nexus as the in- ties connecting different regions that metabolize extracted resources terconnection between different resource systems generated by specific along production chains (Franz et al., 2017). Nexus difficulties at this activities or technologies (D’Odorico et al., 2018); and, transversally, level are encountered in cross-sectoral management approaches when (iv) the nexus as transdisciplinary and co-production practices in sus- bringing together sectors and managing trade-offs between them (Pahl- tainability research (Howarth and Monasterolo, 2017; Scanlon et al., Wostl, 2017; Stein et al., 2018). Lastly, at the microscale and some- 2017). what paradoxically, many localized nexus problems have emerged as Nexus problems have been characterized as types of wicked or post- the result of techno-social innovations aimed at solving the conflict be- normal problems underpinned by uncertainty, ambiguity, contested tween policy goals (Cabello and Madrid, 2014). Classical examples of stakes and unpredictability (Harwood, 2018). Furthermore, these nexus nexus problems in this sense include desalination and irrigation effi- situations are argued to be the outcome of co-existing macro-, meso- and ciency. Innovations related to both of these problems add dependen- microscale processes (Cai et al., 2018), which can be related to the dif- cies on energy resources – such as electrical energy to drive water ferent conceptualizations of the nexus identified above. From a macro pumps – to the water-food nexus, thereby increasing the structural perspective, the nexus preoccupation arises from the acknowledgement complexity from a management perspective. Moreover, as solutions for that economic sectors responsible for resource co-occurrence are gov- reducing pressures over fresh water resources emerge, monitoring of ∗ Corresponding author. E-mail address: Violeta.Cabello@uab.cat (V. Cabello). https://doi.org/10.1016/j.advwatres.2019.06.014 Received 16 January 2019; Received in revised form 25 June 2019; Accepted 25 June 2019 Available online 26 June 2019 0309-1708/© 2019 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license. (http://creativecommons.org/licenses/by/4.0/) V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 social-ecological interactions such as water withdrawal or pollution are The focus of relational analysis thus centers on the organizational demanded. unity of the analyzed system, describing the various roles, or functional The nexus concept thus appears as an indicator of the need to im- behaviors, of system components. More specifically, relational analy- prove our capacity to account for the interconnections between complex sis views the fuzzy relations between function and structure observed social and ecological processes tensioned by multi-level and conflicting in complex systems as compelling subjects of analysis (Kampis, 1987a, drivers. Recent empirical research commences to delineate interconnec- 1987b; Rosen, 1970; Bechtel and Richardson, 2010). In this sense, de- tions as the different causal relations in processes involving more than scribed functions may be realized – in an impredicative and episte- one resource (Andrews-Speed et al., 2012; Bijl et al., 2018). Yet, whereas mological manner – by different structural elements. Within the field relations are the core research object of the nexus concept, little progress of hierarchy theory, Allen and Starr (2017, pp.43–67) also elaborate towards a categorization and formalization of interconnections has been on the impredicative duality of function and structure, to be ana- made. Some consensus has gathered around the differentiation between lyzed over a minimum of three analytical levels within a complex biophysical connections, coined as the resource nexus, and other social system. and governance relations (Cai et al., 2018; Galaitsi et al., 2018). Within Together with relational analysis, hierarchy theory helps define core the resource nexus, a first approximation to categorization is the Font Vi- characteristics of components in a complex metabolic network. Firstly, vanco et al. (2018) taxonomy of direct, dependent and interdependent system components are defined at the same time as a part (e.g. a liver relations. Their model shows that the co-occurrence of water and en- as organ of a human being) and a whole (e.g. a liver made up of cells). ergy along supply chains can be mostly explained by direct extraction or Secondly, as mentioned previously, system components handle the cou- one-way dependencies, while feedbacks appear minor. Their categories, pling between function and a structure (e.g. the function ‘professor’ must however, account only for interactions amongst economic sectors at the be realized by a particular material instance, for example, Prof. McGo- national and global scales. When moving to local and regional contexts, nagall). Thirdly, reflecting back on the Aristotelian building blocks of interdependencies are likely to gain relevance following the considera- relational analysis, system components may be arranged sequentially tion of social-ecological interactions (Albrecht et al., 2018). (material entailment) and hierarchically (functional entailment) in the This paper furthers previous efforts in categorizing nexus intercon- formation of a network. To clarify, a hierarchical disposition of system nections by proposing an operational concept of nexus network. For this components represents a chain of definitions of function. Each func- purpose, we draw on ideas from some branches of complexity theory, tional definition of a component provides the final cause or purpose namely relational analysis (Rosen, 1958, 2005, 2012), hierarchy theory to the realizing structure(s) – thereby its meaning. Each function may (Ahl and Allen, 1996; Allen and Starr, 2017) and societal metabolism be realized by one or more structural components (made of the mate- (Georgescu-Roegen, 1971, 1975; Giampietro and Mayumi, 2000). We rial cause), expressing their agency (efficient cause) when operating in add to the nexus pathways concept of Vivanco et al. (2018) with a an admissible environment and following certain patterns or codes (the multi-level conceptualization of nexus networks that includes interac- formal cause). tions within and between the biosphere and technosphere. This con- The final piece of this framework is the concept of societal ceptual framework is applied to nexus problems derived from intensive metabolism, a notion used to characterize the processes of energy and crop production in arid and semiarid landscapes with over-exploited wa- material transformation in a society that are necessary for its continued ter resources, which often find solutions in energy-intensive technolo- existence (for an overview of its application to ecological economics see gies. The region of Almeria in South-eastern Spain serves as illustrative Martínez Alier and Schlüpmann, 1987); for an overview of its applica- case study. The methodological framework Multi-Scale Integrated Anal- tion in sociology see Fischer-Kowalski, 1998). The metabolic approach ysis of Societal and Ecosystem Metabolism (MuSIASEM, Giampietro and enables the application of relational analysis and hierarchy theory to Mayumi, 2000) is applied to quantitatively formalize the network, ad- nexus relations in social-ecological systems. Giampietro (2018a) defines dressing the following question: why are groundwater resources over- such a system as a set of functional and structural components linked drafted in Almería? by a set of relations and operating within a given boundary to achieve a shared function (a given final cause). Building upon this definition, the MuSIASEM accounting framework (Mayumi and Giampietro, 2000) has 2. Relational analysis of nexus networks been proposed as one of the available integrative methods for quantita- tive analysis of the nexus (Giampietro et al., 2014; Keairns et al., 2016; 2.1. Theoretical ground Shannak et al., 2018). In the following section, we add a conceptualiza- tion of nexus networks to this framework. Relational analysis is a variety of complex system analysis that ex- plores the identity of living systems by representing them formally in metabolic networks characterized by the four Aristotelian causes – ma- 2.2. Multi-level nexus networks terial, formal, efficient and final (Rosen, 2005). This understanding of relational analysis finds roots in theoretical biology and the semi- A nexus network can be defined as the set of processes governing the nal work of Rashevsky (1954) and Arbib et al. (1973), furthered by interdependency between water, food and energy within and across a Rosen (2005, 2012) and more recently by Louie (2009, 2013, 2017) given boundary. As any network, nexus networks are represented with into a new conceptual framework to the modeling of living systems nodes and connecting edges. Nodes refer to specific processes consum- within the field of category theory. Following their discourse, rela- ing more than one resource and producing good(s) or service(s). In for- tional analysis addresses provocative questions of ‘why’ by expound- mal, operational terms, we represent these nodes using the concept of ing the pluralism of causal relations defined over the components of a the metabolic processor (Rosen, 2005, p.250). Most typically, metabolic system.1 processors (hereafter ‘processors’) perform work on a material substrate and transform it into something else performing a function in a complex 1 network. The elaboration of the processor concept into an analytical tool For illustrative example, assume there exists a constituent component A; within societal metabolism theory is a remarkable contribution of the (i) the material cause of A refers to the material out of which A is made; (ii) the formal cause of A refers to the pattern (form) to which the material cause MuSIASEM accounting framework (Ripa, 2017; Giampietro, 2018b). In is required to assume in order to form A; (iii) the efficient cause of A is, in this vein, a processor (Fig. 1) identifies an expected set of relations re- the majority of cases, the agent which transforms the material cause into A garding: (i) a structured process capable of producing a given output or (adhering to the formal cause, in abstract terms some function f); (iv) the final expressing a given function (n – ‘node’ – in Fig. 1); (ii) a profile of inputs cause reflects the for what? of A. and outputs in both the technosphere and the biosphere required for 259 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 directly higher. Hierarchical connections are formalized as many-to-one or many-to-many mappings between system components expressed at different levels of observation. For instance, one can represent a network of connected food pro- duction processes following various stages of production and processing (sequential relations within the technosphere). This sequential pathway may then be scaled up through the various agricultural subsectors, ulti- mately into the agricultural sector in a given region (hierarchical rela- tions within the technosphere). Alternatively, the same sequential path- way may be scaled up through the different scales of river basins sup- Fig. 1. A canonical processor. n denotes a node or singular processor and e plying water to the agricultural processors (sequential relations across denotes a category of edges describing either input or output relations with the the technosphere-biosphere boundary). These two example scaling op- technosphere or biosphere. erations are non-equivalent hierarchical mappings of the same set of se- quential relations and they enable the answering of different questions about the system under analysis. that purpose (e1-6 – ‘edge’ – in Fig. 1). In this sense, processors embody As in other system analyses, the specification of a nexus network the impredicative duality between structure and function of a system requires a research question or problem that enables the selection of component. analytical scales and the delimitation of system boundaries and system Processors such as the canonical processor in Fig. 1 can be connected components with associated functions within the system. From a societal in different ways in the formation of a nexus network. Fig. 2 presents this metabolism perspective, this is equivalent to identifying socioeconomic point using the same notation found in Fig. 1. In the analysis of social- sectors and their specific activities intertwined with ecological processes ecological systems, a minimum of three connection domains should be within a defined social-ecological boundary. In the logic of analyzing fi- identified: (i) connections that remain within the socioeconomic realm nal causes, the functionally defined components at a given level must (the ‘technosphere’); (ii) connections that remain in the ecological realm be further associated to patterns of structural components that are vi- (the ‘biosphere’); and (iii) connections bridging socioeconomic and eco- able within system constraints at a lower level, what is equivalent to logical processes (the ‘technosphere-biosphere boundary’). In any given specifying hierarchical relations. Following the previous example, the nexus problem, the co-occurrence and entanglement of resources im- agricultural sector would be split in subsectors (for instance animal vs plies causal connections within each of these three connection domains vegetal production) and subsectors into supply chains that are viable in simultaneously. a given region. In addition to these three connection domains and pursuing rela- We aim to design a diagnostic tool capable of analyzing the tensions tional analysis, a second dimension is added differentiating between se- generated by conflictive goals and multi-level drivers. Using a function- quential and hierarchical connections in Fig. 2. As already elaborated ally and structurally defined network, these tensions can be described by Vivanco et al. (2018), sequential connections are material flows be- as a problem of multi-functionality. The fact that different system com- tween specific activities or sectors therefore modelling expected outputs ponents coexist within a social-ecological system and that their multiple at a given level of observation. In formal terms, they are one-one or functions need to be sustained while they adapt to external drivers cre- one-many allocations with one-way or feedback direction determined ates many constraints on the type of structural elements that can fulfil by the material entailment between processors. In contrast to sequen- those functions. The spectrum of viable solutions to nexus problems is tial entailment, hierarchical connections are determined by functional limited in the context of competing drivers and any implemented local entailment. That is, the function to be expressed at one level has been de- solution will transform the network of relations by adding new struc- fined as ‘useful’ (as having a proper final cause) at the hierarchical level tural constraints. Fig. 2. A taxonomy of nexus interconnections resulting from the combination of the criterion of social-ecological relations and the criterion of sequential-hierarchical relations. n (node) denotes a processor, individuated by its index; e (edge) denotes a relation (either sequential or hierarchical) between two or more processors, where the first index identifies a source node and the second index identifies a target node(s). (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 260 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 2.3. Formalization of nexus networks groundwater (volume) requirement. In this vein, processors arranged in a network may describe in intensive and/or extensive terms the com- Once the network of metabolic processors is defined in semantic plete set of sequential and hierarchical relations within and between terms, the next analytical step is to instantiate them. In formal terms, a the technosphere and biosphere. These two different sets of numerical processor is a data array structure that contains information about the representation provide two different key types of information about the profile of inputs and outputs required in a given process. The computa- system of interest (King and Carbajales-Dale, 2016). Eq. (E4) formalizes tional form related to this profile shares a number of similarities with the relation between intensively defined processors, supply and demand. that of Life Cycle Assessment (LCA) (see Heijungs and Suh, 2002). The formalism works for diagnostic analyses as well as anticipatory anal- In relational terms, an j-indexed socioeconomic processor nj can be ysis. In an anticipatory analysis, changes in resource demand may be formalized as a column-vector, as shown in Equation E1. The elements estimated from anticipated changes in (i) technical coefficients (inten- of this vector, e1 through ei , represent for some processor n the set of sively defined processor data) e.g. because of technological change; and expected relations defined in intensive terms against a relation or set (ii) supply e.g. following demographic projections. of relations. Most typically, the relations in a processor are described ( ) 𝑛1 |⋯|𝑛𝑗 𝑄𝑠 = 𝑄𝑑 (E4) against just one element (e.g. per unit of a processor output or other- wise immutable quantity such as human activity or land use processor 3. Application to the nexus in arid land crop production inputs). ⎛𝑒1 ⎞ In this section we illustrate the process of building a nexus network 𝑛𝑗 = ⎜ ⋮ ⎟ (E1) for analysing the tension between intensive export-based crop produc- ⎜ ⎟ tion and sustainable management of water resources in the province of ⎝ 𝑒𝑖 ⎠ Almería in South-eastern Spain. Further, several rowwise partitions are erected in this column-vector. In the most general sense, and following the canonical processor pre- 3.1. Framing the problem sented in Fig. 1, the relations of a processor e1 through ei may be placed in one of four2 possible categories. The combination of two variables Almería is one of the driest regions in Europe (average annual each with two possible values creates these four categories: the rela- rain 200–600 mm). At the same time, it represents one of the largest tion’s domain (technosphere or biosphere) and its orientation (input or intra-European suppliers of fresh vegetables, exporting up to 64% output). The set S in Eq. (E2) lists these two variables and elaborates of their overall production. Main importing countries are Germany their possible values. (30%), France (16%), the Netherlands (12%), UK (11%) and Italy {( ) ( )} (7%) (Agencia Tributaria, 2013). Other major land uses for crops are 𝑡𝑒𝑐ℎ𝑛𝑜𝑠𝑝ℎ𝑒𝑟𝑒 𝑖𝑛𝑝𝑢𝑡 𝑆= , (E2) olive groves, almonds and varieties of citrus. Agriculture sustains 8% 𝑏𝑖𝑜𝑠𝑝ℎ𝑒𝑟𝑒 𝑜𝑢𝑡𝑝𝑢𝑡 of the employment in the region and 2.4% of its GDP (Junta de An- The last bit of information associated with each relation is its entail- dalucía, 2013). Contrary to other intensive exploitation systems in ment dependency or dependencies i.e. in the case of a singular depen- Spain, land property is highly distributed with an average farm size dency, whether a cause of one process is produced by or contributes to of 2 to 3 hectares and where most farmers are associated to coopera- another process(es). In the case of a singular dependency, we can say tives for marketing purposes (Varela-Martínez et al. 2016). The region that the validity of the first process both depends on and entails the is a renowned example of unsustainable exploitation of groundwater existence of another process(es). Fig. 2 depicted the various entailment resources in Spain, with many aquifers classified as severely overex- dependencies possible in our relational analysis. ploited and/or polluted according to the European Water Framework Processors may, therefore, be represented as column-vectors Directive (WFD) assessment. The European WFD also mandates that all (Eq. (E1)) partitioned along the elements of the set S (Eq. (E2)). The European water bodies must recover to a ‘good status’, a mandate ex- concatenation of all processors described as sparse column-vectors then pressed in terms of ‘environmental goals’ defined by the year in which provides the ‘dataset’ describing the system of interest for the purposes a good status shall be recovered following the implementation of new of the analysis, as shown in Eq. (E3). It should furthermore be noted management measures. Fig. 3 depicts these goals for the Almería region. that the description made by Eq. (E3) represents the most basic catego- Framing the nexus problem, a macroscale tension arises from the rization possible. An effective analysis necessarily includes additional goals of water policy, which require a reduction of the pressures over subledgers dependant on the goals of the analysis. For example, in anal- water resources, and European food markets eager to absorb the increas- yses of societal metabolism, an additional subledger differentiating be- ing local production. At the region level, the problem translates into the tween fund, flow and stock resource is typically regarded as an essential question of how to re-organize relations between agricultural and water relation refinement (Georgescu-Roegen, 1971). management regimes. Agriculture is responsible for 80% of the water 𝑒 ⋯ 𝑒1,𝑗 ⎞ usage, 62.5% of which is groundwater (Junta de Andalucía, 2015). The ( ) ⎛ 1,1 𝑛1 |⋯|𝑛𝑗 = ⎜ ⋮ ⋱ ⋮ ⎟ (E3) main solutions on the table so far have been the utilization of alterna- ⎜ ⎟ tive water resources, desalination and reclamation, or external transfers ⎝ 𝑒𝑖,1 ⋯ 𝑒𝑖𝑗 ⎠ as substitutes for groundwater and the improvement of irrigation effi- Generally, processors are stored in the dataset in intensive (unitary) ciency. A decade later, overdraft rates are not reported to decrease and terms. To maintain validity, care must be taken to relate intensive rep- the contribution of these innovations to the stated purpose is unclear. resentations with their original external reference. In an analysis of a Using the framework described above, we build a nexus network to pro- system of interest, the set of intensively described processors may be vide a relational analysis of the aquifer over-exploitation phenomena. proportioned by the size of supply (vector Qs ), or by any other variable As mentioned, relational analysis focuses on multiple causality in against which relations are defined, to fulfil demand (vector Qd ). In the complex problems by elucidating interconnected functions. Clearly, ad- context of this work’s case study, supply vector Qs would include mass dressing the question of why aquifers are overexploited in Almería can quantities of e.g. Almonds and Vegetables; demand vector Qd would be answered from different lenses. For instance, a governance and power include various resource demands e.g. land (area) requirement and relations reading would illuminate one sort of relevant responses regard- ing the key actors involved in the problem and their interconnections. 2 Relations within the system of interest are always within the technosphere, However, this paper focuses on the metabolic reading of the nexus. That an observation which explains how the six categories in Fig. 1 may be reduced is, on the type of answers provided by sequential (material) and hi- to four at this stage of the formalization. erarchical (functional) entailments between metabolic processors. We 261 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 Fig. 3. Map of aquifers in Almeria classified accord- ing to environmental goals in water districts man- agement plans 2015–21 (Junta de Andalucía, 2015; Gobierno de España, 2015). Environmental goals are defined by the year in which a good status shall be re- covered. Aquifers coloured in green were assessed as being in good status in 2015. Those in yellow (good status to be achieved in 2021), orange (in 2027) or red (in undetermined horizons later than 2027) have been assessed as being in less than good sta- tus, meaning they are quantitatively and/or qualita- tively over-exploited. Quantitative impacts are spec- ified in Figure 5. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) assume the view that the ultimate final cause for over-drafted ground- water resources in present-day Almería is the maintenance of an agri- cultural economic model based on valuable crop exports. Understand- ing the other causal mechanisms for the overexploitation phenomena requires a threefold analysis. First, it is necessary to understand what farming systems are driving water demands. Second, it is necessary to look at the role of groundwater within the overall pattern of water re- sources supplied to those farming systems by irrigation organizations. Lastly, it is necessary to contextualize withdrawals in the hydrological balance of the different aquifers. 3.2. Building the network Considering the whole region of Almería as a social-ecological sys- tem, a multi-level nexus network was defined connecting key compo- nents of food and water management domains identified above (Fig. 4). At the top analytical level (1), these domains are responsible for or- ganizing the implementation of public policies, programs or strategic goals for those sectors, posing top-down constraints over the behavior of lower level system components. As mentioned, bottom-up constraints Fig. 4. Definition of a multi-level nexus network linking crop supply and water can be uncovered by adding levels of functional-structural metabolic management in Almería. Black arrows represent sequential connections while grey arrows represent hierarchical connections. For a full description of cate- processors to the network. In this vein, the agricultural sector has been gories please refer to the appendix and the databases. Color key is shown in disaggregated into irrigation areas as the organizational units of water Fig. 2. (For interpretation of the references to color in this figure legend, the management in agriculture, and further into farming systems and even reader is referred to the web version of this article.) further into the different crops produced in each of those farming sys- tems. On the other hand, sequential connections relate functions within the same level adding more constraints to the network in terms of multi- functionality. In our network, sequential linkages are established within them. The most standard procedures are either data-driven (e.g. sample components of the water management system, namely water bodies, wa- a population and cluster it in typologies) or semantic-driven approaches ter suppliers and irrigation areas as water users. (e.g. pre-define typologies and sample representative instances). For the Almería case, we followed a mixed approach. Within the agricultural do- 3.3. Operationalizing the network main, we defined typologies of crop production processes and farming systems through a review of previous studies and techno-agronomic re- As explained in Section 2, the methodological approach proposed in ports (García-García et al. 2016; Varela-Martínez et al. 2016) and the this paper uses the numerical structure of processors to quantify nexus analyses of land use data for different crops petitioned from the Andalu- nodes and relations. For this purpose, one must identify physical in- sian Agricultural Administration. On the other hand, available georefer- stances matching the categories of system components, quantify pro- enced datasets for water system components in river basin management cessor variables and profile representative types of processors out of plans (Junta de Andalucía, 2015; Gobierno de España, 2015) enabled 262 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 the characterization of the full population of water bodies and irrigation open fields (IA2 and IA3). For those aquifers with extraction index over areas. Crop data was only provided for the productive year 2012–2013 1.1, Fig. 6(1) presents the annual volume of extraction split by type of whereas data on water extraction and use is available as an annual av- irrigation area (IA). erage for the 2010–2015 management period. Therefore, the analysis In absolute terms, most groundwater over-extraction in Almería is in- is bound to a single snapshot of the annual metabolism of the system curred from three aquifers (011, 012 and 013) by three irrigation areas without addressing its time evolution. clustered in IA1. These areas, populated by greenhouse farming systems Processors variables were populated with data gathered from dif- (F1 in Fig. 6(2)), grow a vast amount of fresh vegetables and elicit the ferent secondary sources and integrated in two databases describing highest rates of labor and gross value-added generation. As observed the crop production and the water management domains. Variables in Fig. 6(3), the IA1 cluster has introduced more alternative water re- were scaled from lower to upper levels in the network according to the sources than the other IAs, although they still represent a small share of land use patterns described across levels. Structured and documented the total water used. datasets and R code for scaling operations are published in a Zenodo3 In relative terms, the aquifers showing the highest extraction in- repository and the data management process from collection to visual- dices are associated with other types of farming systems, predominant ization is thoroughly described in the Appendix. Given the diversity of in clusters IA5 and IA7. IA5 is represented by a single irrigation re- data sources and their production methods, mostly based on estimations gion occupied by olive groves in hyper-intensive production mode (F6 in rather than accurate measurements, the technical and methodological Fig. 6(2)). This novel system has rapidly extended due to its larger mar- uncertainty associated to the datasets is notable. For the purposes of this gin of benefits as compared to intensive olive production and spurred paper, we restricted data analysis to the most trustful variables used by by public subsidies (F5). Whereas its contribution to annual overall crop public institutions in their management plans. Further efforts in reliable production in the region is minor (Fig. 6(4)), it consumes almost ex- data production for robust integrated assessments of nexus interconnec- clusively groundwater from aquifers with very low annual availability tions are needed in the future. (007–009). On the other hand, cluster IA7 gathers two large irrigation To address the proposed research question, we provide a diagnostic areas representing the second most relevant contribution to regional wa- analysis consisting of the interconnected visualization of some variables ter demand and crop production after IA1 (Fig. 6(3) and (4)). It com- in food and water processors at different levels in the network, namely, bines three main types of farming systems producing mixed vegetables withdrawal rates from aquifers by different irrigation areas (level 2), in open fields, with lettuce and citrus (F3, F4 and F8 in Fig. 6(2)). These patterns of water resources use in irrigation areas (level 2), patterns farming systems show middle intensities of labor and gross value-added of land uses in irrigation areas (levels 2/3/4) and crop production fac- generation. Water resources in these areas blend groundwater from tors in different farming systems (levels 3/4)). Groundwater availability over-drafted aquifers with desalinated water and an important share and the derived Extraction Index4 are used to check the sustainability of external transfers. Even so, current withdrawals contribute to surpass of different water-food nexus patterns against quantitative impacts on the very low availability rates in five aquifers (001–008). aquifers. For analytical purposes, the processors describing the 16 irriga- From the previous analyses we can conclude that the answer to the tion areas are clustered into 7 typologies based on their pattern of farm- question of why aquifers are over-drafted in Almería is at least twofold. ing systems (level 3). For this purpose, profiles associated with quanti- On one hand, the profitable greenhouse economy has very large de- tative criteria on land use were defined through a qualitative analysis mands of water so far fulfilled mostly with groundwater. On the other of most representative patterns and the 16 irrigation areas were coded hand, we find a large area of less water demanding farming systems re- accordingly (see Table A3 in the Appendix for further explanation and lying on aquifers with very limited water available. Implementing wa- Ouput/ClusterData.xlsx in the data folders for assignation of irrigation ter policy objectives requires ratcheting withdrawals down below avail- areas to each cluster). Processors variables, including withdrawals from ability (grey reference line in Fig. 6(1)). The volume of these reductions aquifers, were aggregated for the obtained categories and selected in- might not look dramatic in absolute terms, but it is an important fraction dicators calculated. This clustering enables profiling several typologies of water supplies for several irrigation areas as observed in Fig. 6(3)). of social-ecological patterns on a spatially explicit basis as presented in the next section. 4.2. Discussion 4. Results and discussion In the absence of an effective harmonization between water and agri- cultural policies in the EU (Söderberg, 2016), the tension between en- 4.1. Why are aquifers overdrafted in Almería? vironmental policy objectives and market demands in relation to crop production in Almería could be addressed by its: (i) final causes (e.g. We are now in the position of providing a quantitative relational an- what if the agricultural model in Almería is transformed to meet sus- alyzes of our nexus problem. Fig. 5 presents the geographical display of tainable aquifer yields?); (ii) material causes (e.g. what if the available the clustered irrigation areas overlapping aquifers in the region classi- groundwater is increased?); (iii) formal causes (e.g. what if the pattern fied by their Extraction Index. Fig. 6 visualizes the above described vari- of supplied water resources is transformed?); or (iv) efficient causes (e.g. ables of several connected processors within the food and water regimes. what if farming systems adapt to reduce their water demand?). These According to the water districts management plans, 12 out of 27 aquifers four categories of causal relations span both structural and functional in Almería show an extraction index over 0.8, meaning their exploitation considerations, and may be combined in the search for robust solutions. rate is considered unsustainable. The spatial distribution of this index With coastal irrigation areas showing high irrigation efficiency rates reveals a certain predominance of high extraction rates along the coast- and with the increment of external transfers blocked by social conflicts,5 line as compared to inner rural areas. These rural irrigation areas are the main strategy implemented over the past decade in the analyzed ir- to a large extent dominated by low intensity almond and olive produc- rigation areas lay in increasing the production of alternative water re- tion (irrigation area type IA4, farming systems F5 and F7 in Fig. 6(2)). sources in order to replace groundwater (i.e. a formal cause strategy). Connected aquifers show either low or moderate extractions rates (0.8– However, as observed in our analyses, the penetration of these water 1.1) in those areas with presence of vegetables or lettuce production in supplies is still minor in most irrigation areas. Discussed reasons for this 3 5 https://zenodo.org/record/2539219#.XDx9PRBRcVs. The two external transfers in the province, Tajo-Segura and Negratin- 4 Under the Water Framework Directive, the Extraction Index is calculated as Almanzora, are associated with noteworthy socio-environmental conflicts in the the ratio between annual withdrawals and available groundwater. origin basins (Hernández-Mora et al., 2014). 263 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 Fig. 5. Map of clustered irrigation areas overlapping relative with- drawal for agricultural purposes in Almería. An Extraction Index over 0.8 is deemed unsustainable under the Water Framework Directive. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Fig. 6. Quantitative relational analysis of aquifers overdraft driven by crop production in Almería. Color codes for sub-figures 1, 3 & 4 indicate the cluster of irrigation areas as in Fig. 5. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 264 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 outcome include the combination of the high energy cost of related tech- Funding nologies and the context of increasing electricity prices while lowering crop prices6 (March et al., 2014). This burden is more significant for This research has received funding from the European Union’s farming systems other than greenhouses given their more limited mar- Horizon 2020 Research and Innovation Programme under grant agree- gin of benefits. As a result, and looking into the future, either new strate- ment No 689669 corresponding to the project Moving Towards Adap- gies should be considered or the crop production function of Almería is tive Governance In Complexity: Informing Nexus Security (www.magic- likely to be forcibly altered. nexus.eu); the Spanish Ministry of Education, Culture and Sport Our diagnostic analysis shows three social-ecological patterns in- (FPU15/03376); the Spanish Ministry of Science, Innovation and Uni- volved in the overdraft phenomena, namely: (i) intensive greenhouses – versities, through the “María de Maeztu” program for Units of Excellence withdrawing very large quantities of water; (ii) hyper-intensive olives – (MDM-2015-0552)]. drawing from aquifers with extremely low availability; and (iii) mix of open field fruits and vegetables – drawing from aquifers with extremely Supplementary materials low availability. These clusters of WEF interconnections face different economic and biophysical constraints that may require diverse adaptive Supplementary material associated with this article can be found, in pathways. The multiple ways in which these changes may unfold open a the online version, at doi:10.1016/j.advwatres.2019.06.014. space for simulation and scenario analysis in future research. Our nexus network can be expanded to appraise the viability of alternative solu- tions capable of maintaining socially desired system functions. Appendix: Data management A key contribution of this paper is the methodological development of a data modeling structure for relational analysis of a WEF nexus net- 5. Conclusions work with entangled food and water supply. A database per resource domain, one for agriculture and another for water, has been populated Nexus research has mobilized a myriad of existing methodological with processor data describing network components and their relations approaches, and emphasized interconnections as its core research ob- for the province of Almería (Spain). In fulfilment of the open access ject. This paper furthers these efforts by proposing a theory of relational requirements of our funding body European Union’s Horizon 2020 Re- analysis of nexus networks and a taxonomy of interconnections. We ar- search and Innovation Programme, data has been made available under gue for the need to attend to competing drivers of nexus problems at the license Creative Commons Attribution 4.0 International in Zenodo different scales and to social-ecological interactions. Nexus networks repository: https://zenodo.org/record/2539219#.XDx9PRBRcVs are conceptualized in a multi-level fashion that addresses cross-scale As an integrated accounting method, the proposed numerical struc- feedbacks in terms of bottom-up and top-down constraints to multi- ture of processors requires collecting and integrating data from multi- functionality. This is illustrated in an application to nexus problems ple data sources. This Appendix describes the data management process generated by intensive export-based crop production in conflict with from data collection to visualization. sustainable management of water resources. The scale where this ten- sion is generated is European while regional solutions do not attend A.1. Sampling, data collection and validation the diversity of water-food management patterns and increase nexus interdependencies. The analysis reveals several interconnected drivers Data was collected between March and June 2018 from various sec- of the aquifer overdraft phenomena, at different analytical levels but ondary data sources and direct petition to public administrations. In also in different geographical locations, proving the usefulness of the general, we found water datasets better documented and accessible than proposed diagnostic tool. The research nonetheless is limited to a sin- agricultural datasets. In what follows, we explain the processes of sam- gle snapshot and improved time-series are needed for more conclusive pling, definition of variables, data collection and validation used to pro- results. As signaled in many other nexus papers, existing data sources duce the two databases serving as inputs to the quantitative analysis. face problems of methodological and technical uncertainty and further coordinated efforts are required if the nexus is to become a useful sci- (a) The agriculture database (Input/CropDatabase.xlsx) ence for policy approach. Future advances shall be directed towards (i) embedding the tool in larger extended peer research processes where Survey sampling to implement quality assurance and uncertainty management protocols; This database contains data on processors describing system com- and (ii) anticipatory analysis in the assessment of the option space for ponents within the agricultural domain. These components are hierar- accommodation of new top-down functional constraints or bottom-up chically organized in our network (Fig. 4 of the paper), from level 4 structural transformations of WEF interconnections. (L4) describing types of crop production processes, to level 3 describ- ing farming systems (L3) and level 2 (L2) describing irrigation areas. Level 1 (L1), referring to the whole irrigated crop production system in Almería, is derived from previous levels through scaling operations. Acknowledgments In the absence of available data on the full population of farming sys- tems, we worked with representative typologies for characterizing pro- The authors are grateful to project partners for their useful reviews cessors at L4 and L3. Besides the analysis of land use data, data inform- and support to this research, especially to Dr. Carmela Iorio for pro- ing the decisions for which typologies should be included was obtained viding guidance on the clustering process. This article reflects only the from a review of grey literature and techno-agronomic reports describ- author’s view. None of the funding agencies are responsible for any use ing local systems of production (Tolon-Becerra et al. 2013; García-García that may be made of the information it contains. et al. 2016; Varela-Martínez et al. 2016). Starting with L4 on crop production processes, data was petitioned from the Andalusian Agricultural Administration by phone to its office 6 The average energy costs in the region range from 0.88 kwh/m3 for ground- in Almería and followed up by email.7 Land use data on irrigated crops water pumping, to 4.5 kwh/m3 for desalination and to 1.66 kwh/m3 for wastew- ater reclamation. Over the last years, farmers in the region strongly championed 7 for subsidies to desalinated water, so far begetting little echo in public authori- For the province of Almería, data petitions contact is: ties. . 265 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 Table A1 Categories for crop production processors (L4 in the network). Crop category Greenhouses MC1 Greenhouses MC2 Open fields intensive Open fields hyper-intensive Fresh Pepper Aubergine Broccoli vegetables Tomato Cucumber Lettuce Watermelon Green beans Green beans Zucchini Melon Melon Pepper Pepper Tomato Tomato Watermelon Watermelon Zucchini Zucchini Oil crops Olives Olives Fruits Almonds Citrus in the different agricultural management units (to a large extent over- Agriculture) and water footprints (calculated in previous studies using lapping irrigation areas in the region) was requested for the time se- standardized modeling procedures). Other data are statistical estima- ries 2010–2015. Out of those years, only data for the production year tions developed by public institutions based on a number of assump- 2012–2013 was provided. For this reason and considering the absence tions, such as production ratios, water uses or labor requirements. They of available time series for other variables in the crop production pro- were accessible in online data portals, management plans and reports. cessors, the analyzes was restricted to a single annual snapshot. From Lastly, measurements of real instances were only found in land use data all irrigated crops grown in the season 2012–2013, we selected the most obtained through GIS methods. representative ones based on two criteria: The methodological and technical uncertainty associated to most of the above described data production methods is noteworthy. The lack of • Those crops producing more than 10,000 tonne or occupying more a systematic accounting of environmental and social variables in agri- than 5000 ha in the selected year; cultural statistics, especially of water-related variables, hinders the po- • In significantly different production systems in terms of yielded tential of the proposed analytical tool and the capacity to produce mean- tonne/ha∗ year, namely: greenhouses with two cropping periods ingful assessments of the sustainability of nexus interconnections. This (MC2), greenhouses without crop rotation (MC1), open fields with is an open discussion within the nexus research community that needs intensive production and open fields with hyper-intensive produc- to further involve public administrations and their statistical offices at tion. all scales of resource governance. The resulting list of 24 crop production processors is shown in (a) The water database (Input/WaterDatabase.xlsx) Table A1. Crop categories correspond to those in the provided dataset and represent averages for several species. For instance, citrus includes Survey Sampling different types of lemons, oranges and mandarins. This database contains data on processors describing system com- Defined with a similar mixed procedure of land use and literature ponents within the water management domain, namely irrigation areas analyses, the sample for L3 includes 8 typologies of farming systems as water users, water providers and water bodies. Water suppliers are showing significantly different patterns of crop production processes. On desalination and reclamation plants, plus two external water transfers. the other hand, spatial data on irrigation areas (L2) is publicly available These components are sequentially connected in our network. in the management plans of the two river basins over which the region is Contrary to the agricultural domain, water management districts hydrologically divided (Junta de Andalucía, 2015; Gobierno de España, have georeferenced datasets locating system components. Therefore, we 2015). Therefore, the full population of 16 irrigation areas was sampled. were able to include the whole population in the database with the ex- Definition of variables for characterizing processors and their ception of water suppliers that were characterised as typologies based relations on different technologies. Processors in this database contain variables characterizing the Definition of variables for characterizing processors and their input-output of agricultural production at the different analytical lev- relations els. These variables, and therefore the column-vector structure of pro- Processors for irrigation areas within the water domain split their cessors, may vary with the level of observation. For instance, crop pro- water requirements in the pattern of water resources fulfilling those duction processes are characterized through variables such as water and requirements. On the other hand, processors describing water bodies different nutrient requirements or crop yield, while farming systems pro- include their inputs and outputs of water as expressed in their annual cessors include other input variables such as the total use of fertilizers, hydrological balance. Lastly, water suppliers include the total supply pesticides or plastics. capacity, the annual water supply and the ratios of electricity consump- Because of the lack of available data on yields at different analytical tion. levels, all variables are intensively defined per hectare of land used. Hi- Sequential connections in this database refer to the different water erarchical connections between components are therefore operational- flows from different suppliers and/or water bodies to the irrigation ar- ized in terms of land use patterns described across levels. No sequential eas. They are expressed in extensive terms (Hm3 /year). connections are considered for the crop production system. Data collection and validation Data collection and validation Data was gathered from the websites of the two water districts in Input data for processor variables were collected from secondary which the region of Almería is divided: the Andalusian Mediterranean data sources, previous studies or reports. Data sources vary depending Basins Water District (Junta de Andalucía, 2015) and the Segura Water on the variable and are all indicated in the database (see ‘data_source’ District (Gobierno de España, 2015). Georeferenced datasets were pub- column in processors sheets and ‘dataset_key’ sheet for references). Most licly available in their Spatial Data Infrastructures. However, most data gathered data are estimations obtained through different methods as ex- for characterizing processors variables had to be extracted from tables plained in the ‘Comments’ column in each dataset. Some variables were in ∗ .pdf documents of their current management plans. produced through simple modeling techniques as in the case of fertil- River basin districts in Spain do not perform an annual accounting izer use (calculated using the guidelines set by the Spanish Ministry of of actual water uses. Available data refer to estimated average annual 266 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 Fig. A1. Calculated vs. statistical values for the total production of different crops in Almería for the year 2012/2013. Fig. A2. Calculated vs. river basin management plans (rbmp) val- ues for gross water use in irrigation areas in Almería, average for the years 2010–2015. water balances for the period 2010–2015. Despite the obvious uncer- may be found in the data folder. Output datasets (collected in Calibra- tainty associated to an estimated five-year average, we consider this tion.xlsx) contain full processors describing farming systems and irriga- data source more reliable than those of agricultural databases because tion areas with both extensive and intensive variables. it is the baseline information used by water districts for the management Calibration of scaling operations was performed on only two vari- period 2015–2020. ables. First, the overall production of different crops (total annual tonnes), contrasting model outputs at L1 (the whole region of Alme- A.2. Data organization and processing ria) with the annual statistics of the Andalusian administration for the region (Junta de Andalucía, 2013). Second, the total water used by irri- Input database organization gation areas according to management plans was compared against the The two databases are organized in a similar structure that consist aggregated net water requirements from the model, divided by a coeffi- of the following datasets: cient of irrigation efficiency to obtain gross water use. Comparison for these variables is shown in the following Figs. A1 and A2. • Metadata, including the description of workbook sheets. Fig. A1 shows that the model underestimates production of fresh • Acronyms: list of acronyms used in all sheets. vegetables. This means that either land use or yield data in the model • Datasource_key: references and links to data sources. is lower than real values. On the other hand, calculated citrus and olive • Parameters: parameters used for estimations or up-scaling of vari- productions were 20% and 37% larger than statistical estimations for ables. that year, respectively. Regarding water use estimations, the mismatch • Instances: list of research objects with external referent in the case is especially relevant in DM21, the largest irrigation area with intense study. greenhouse production (in cluster IA1 in Figs. 5 and 6 of the paper). This • Relations: input data describing either sequential or hierarchical means that either total greenhouse land and/or water requirements per connections. hectare for the different crops are larger than reported in data sources, • processors_X: series of sheets containing input data for processors de- or irrigation efficiency is significantly smaller. scribing network components. Each dataset includes interface data Sequential connections between instances of water bodies and users classified along the categories defined in Section 2 of this paper (four are not considered in water management plans. Therefore, they were or more categorical variables plus the numerical continuity), geo- calculated through spatial analyses. Layers of irrigation areas, water graphic and temporal scales, data sources and comments document- bodies and suppliers were overlapped and the most probable water ing relevant information on estimation procedures or data sources. sources for the different users located. This information was contrasted Data processing with total withdrawals from those sources for agricultural purposes and Processor datasets in input databases contain only those variables adjusted when necessary in order to obtain an estimation of withdrawal that are newly defined for each component (properties that cannot be by each irrigation area from each aquifer in Almería. derived from a hierarchical relation). The rest of variables were aggre- Clustering process gated from lower to upper levels, through linear or non-linear transfor- The 16 irrigation areas (L2) in the region were clustered into 7 mations. The scaling was perform using R scripts, for which the code typologies based on their land use pattern of farming systems (L3). 267 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 Table A2 Profiles of irrigation areas based on the land used for different farming systems Description Quantitative criteria Code Mostly mix of vegetables in greenhouses >60% of land used for greenhouses, remaining mix of 1 other crops More than half of land used for mix of vegetables in open fields >50% of land used for open field vegetables mix, 2 remaining citrus or olive groves in intensive production More than half of land used for lettuce production in open fields >50% of land used for open field lettuces, remaining 3 citrus or olive groves in intensive production More than half of land used for olives production complemented by almonds >30–50% of land used for olive groves and almonds in 4 intensive production Mostly olives production in hyper intensive production >30–50% of land used for olive groves in hyper intensive 5 production, 20–40% olive and almond mix, remaining open field or citrus Mostly citrus supplemented by lettuce production in open fields >40% of land used for citrus production, remaining 6 lettuces in open fields Mix of lettuce, citrus and vegetables in open fields 30% lettuce, 30% citrus, 30% mix vegetables in open fields 7 Table A3 List of indicators used in visualizations. Indicator Reference Unit Calculation Water Extraction Index Fig. 5 % Divides annual withdrawals from an aquifer by groundwater availability, which is calculated as inflows minus environmental water requirements. Aquifer withdrawal rate Fig. 6-1 Hm3 /year Sum of annual aquifer withdrawal performed by clustered irrigation areas. Water use pattern Fig. 6-2 Hm3 /year Annual water used from different sources by clustered irrigation areas. Land use pattern Fig. 6-4 ha/year Hectares of land used for different crop production by clustered irrigation areas. Labor Fig. 6-3 h/ha∗ year Hours of labor employed per hectare by farming systems. Gross Value Added Fig. 6-3 €/ha∗ year Gross added value generated according to price for farmers per hectare by farming systems. Yields Fig. 6-3 tonne/ha∗ year Yields per hectare by farming systems. For this purpose, profiles associated with quantitative criteria were Albrecht, T.R., Crootof, A., Scott, C.A., 2018. The water-energy-food nexus: a sys- defined through a qualitative analysis of most representative patterns tematic review of methods for nexus assessment. Environ. Res. Lett. 13, 043002. https://doi.org/10.1088/1748-9326/aaa9c6. (Table A2) and the 16 irrigation areas were coded accordingly (see Allen, T.F.H., Starr, T.B., 2017. Hierarchy: Perspectives for Ecological Complexity, second Ouput/ClusterData.xlsx in the data folder for assignation of irrigation ed. The University of Chicago Press, Chicago. areas to each cluster). Processors variables, including withdrawals from Ahl, V., Allen, T.F.H., 1996. Hierarchy Theory: A Vision, Vocabulary, and Epistemology. Columbia University Press, New York. aquifers, were aggregated for the obtained categories and selected indi- Andrews-Speed, P., Bleischwitz, R., Boersma, T., Johnson, C., Kemp, G., vanDeveer, S.D., cators calculated as explained in the following section. 2012. The Global Resource Nexus: The Struggles for Land, Energy, Food, Water, and Selection of indicators Minerals. Transatlantic Academy, Washington, DC. Arbib, M.A., Milsum, J.H., Rashevsky, N., Rescigno, A., Richardson, I.W., Roberge, F.A., The following indicators shown in Table A3 were extracted or cal- Rosen, R., 1973. In: Rosen, R. (Ed.). Foundations of Mathematical Biology: Supercel- culated from the previously described datasets. Structured datasets of lular Systems, III. Academic Press, New York. indicators can be found in Visualization/VizData.xlsx. Bechtel, W., Richardson, R.C., 2010. Discovering Complexity: Decomposition and Local- ization as Strategies in Scientific Research. MIT Press. Visualization Bijl, D.L., Bogaart, P.W., Dekker, S.C., van Vuuren, D.P., 2018. Unpacking the nexus: dif- The above described indicators were visualized in a canvas with four ferent spatial scales for water, food and energy. Glob. Environ. Change 48, 22–31. connected graphs Fig. 6 of the paper), generated using the Python pro- https://doi.org/10.1016/j.gloenvcha.2017.11.005. gramming language and organized in Tableau. The first graph (F6-1) Cabello, V., Madrid, C.M., 2014. Water use in arid rural systems and the integration of water and agricultural policies in Europe: the case of Andarax river basin. Environ. portrays the rate of annual withdrawal from over-drafted aquifers in Dev. Sustain. 16, 957–975. https://doi.org/10.1007/s10668-014-9535-8. Almería, split by clustered irrigation areas. The second graph (F6-2) Cai, X., Wallington, K., Shafiee-Jood, M., Marston, L., 2018. Understanding and managing shows a Voronoi tessellation of the pattern of water resources in the the food-energy-water nexus – opportunities for water resources research. Adv. Water Resour. 111, 259–273. https://doi.org/10.1016/j.advwatres.2017.11.014. seven clustered irrigation areas in terms of relative annual water sup- D’Odorico, P., Davis, K.F., Rosa, L., Carr, J.A., Chiarelli, D., Dell’Angelo, J., Gephart, J., ply. The third graph (F6-3) displays three relevant farm production fac- MacDonald, G.K., Seekell, D.A., Suweis, S., Rulli, M.C., 2018. The global food-energy- tors in relation to their crop production processes. Finally, the fourth water nexus. Rev. Geophys.. https://doi.org/10.1029/2017RG000591. Fischer-Kowalski, M., 1998. Society’s metabolism: the intellectual history graph (F6-4) shows the Voronoi tessellation of the relative contribution of materials flow analysis, part I, 1860–1970. J. Ind. Ecol. 2, 61–78. of different crop production processes to the land use pattern of irriga- https://doi.org/10.1162/jiec.1998.2.1.61. tion areas. Altogether, the visualization enables a relational analysis of Font Vivanco, D., Wang, R., Deetman, S., Hertwich, E., 2018. Unraveling the nexus: ex- ploring the pathways to combined resource use: unravelling the nexus. J. Ind. Ecol.. several causes for the aquifer overexploitation phenomenon, intercon- https://doi.org/10.1111/jiec.12733. nected at different levels in the described nexus network. Franz, M., Schlitz, N., Schumacher, K.P., 2017. Globalization and the water-energy-food nexus – Using the global production networks ap- proach to analyze society-environment relations. Environ. Sci. Policy. https://doi.org/10.1016/j.envsci.2017.12.004. References Galaitsi, S., Veysey, J., Huber-Lee, A., 2018. Where is the added value? A Re- view of the Water-Energy-Food Nexus Literature. Stockholm Environment Insti- Agencia Tributaria. 2013. Annual statistics for external trade. Cámara Oficial de Comercio, tute, Stockholm SEI working paperAvailable at. https://www.sei.org/publications/ Industria, Servicios y Navegación de España. Available at (Last access Sep 2018): added-value-review-water-energy-food-nexus-literature/. http://aduanas.camaras.org/. 268 V. Cabello, A. Renner and M. Giampietro Advances in Water Resources 130 (2019) 258–269 García-García, M.C., Céspedes López, A.J., Pérez Parra y, J.J., Lorenzo Míngue, P. Martínez Alier, J., Schlüpmann, K., 1987. Ecological economics: energy, environment, and (Eds.), 2016. El sistema de producción hortícola protegido de la provincia de Society. Basil Blackwell, Oxford [Oxfordshire] ; New York, NY, USA. Almería. IFAPA Available at (last access Sep 2018): https://www.researchgate.net/ March, H., Saurí, D., Rico-Amorós, A.M., 2014. The end of scarcity? Water desalina- profile/Jeronimo_Perez-Parra/publication/308118507_El_sistema_de_produccion_ tion as the new cornucopia for mediterranean Spain. J. Hydrol. 519 (Part C), 2642– horticola_de_la_Provincia_de_Almeria/links/57da9cf108ae5292a3767c52/El-sistema- 2651. Water governance across competing scales: Coupling land and water man- de-produccion-horticola-de-la-Provincia-de-Almeria.pdf. agementIncorporating water resources in integrated urban and regional planning Georgescu-Roegen, N., 1971. The Entropy Law and the Economic Process. Harvard Uni- https://doi.org/10.1016/j.jhydrol.2014.04.023 . versity Press. Mayumi, K., Giampietro, M., 2000. Multiple-scale integrated assessment of societal Georgescu-Roegen, N., 1975. Energy and economic myths: institutional and analytical metabolism: introducing the approach. Popul. Environ. 22, 109–153. economic essays. South. Econ. J. XLI (3), 347–381. Muranetto, S., Witmer, M. 2017. Water-land-energy-food-climate nexus: poli- Giampietro, M., 2018a. Anticipation in agriculture. In: Poli, R. (Ed.), Handbook of Antic- cies and policy coherence at european European and international scale. ipation. Springer, Cham. https://doi.org/10.1007/978-3-319-31737-3_23-1. SIM4NEXUS Project. Deliverable 2.1. Available at: http://www.pbl.nl/sites/ Giampietro, M., 2018b. Perception and representation of the resource nexus at the inter- default/files/cms/publicaties/WP2_Deliverable%202.1_15nov17_FINAL.pdf face between society and the natural environment. Sustainability 10. article no. 2545 Nilsson, M., Zamparutti, T., Petersen, J.E., Nykvist, B., Rudberg, P., McGuinn, J., 2012. https://doi.org/10.3390/su10072545. Understanding policy coherence: analytical framework and examples of sector– Giampietro, M., Mayumi, K., 2000. Multiple-scale integrated assessments of societal environment policy interactions in the EU. Environ. Policy Gov. 22, 395–423. metabolism: integrating biophysical and economic representations across scales. https://doi.org/10.1002/eet.1589. Popul. Environ. 22 (2). Pahl-Wostl, C., 2017. Governance of the water-energy-food security nexus: a Giampietro, M., Aspinall, R.J., Ramos-Martin, J., Bukkens, S.G.F., 2014. Resource ac- multi-level coordination challenge. Environ. Sci. Policy. https://doi.org/10.1016/ counting for sustainability assessment: the nexus between energy, food, water and j.envsci.2017.07.017. land use. Routledge Explorations in Sustainability and Governance. Routledge, New Rashevsky, N., 1954. Topology and life: in search of general mathematical principles in York. biology and sociology. Bull. Math. Biophys. 16, 317–348. Gobierno de España. 2015. Segura water district management plan 2015–2021. Avail- Ripa, M., (Ed). 2017. Report on nexus security using quantitative story-telling. MAGIC able at (last access Sep 2018): https://www.chsegura.es/chs/planificacionydma/ project Deliverable 4.1. Available at: http://www.magic-nexus.eu/documents/ planificacion15-21/. d41-report-nexus-security-using-quantitative-story-telling. Harwood, S.A., 2018. In search of a (WEF) nexus approach. Environ. Sci. Policy 83, 79–85. Rosen, R., 1958. A relational theory of biological systems. Bull. Math. Biophys. 20, https://doi.org/10.1016/j.envsci.2018.01.020. 245–260. by Heijungs, R., Suh, S., 2002. The computational structure of life cycle assessment. In: Rosen, R., 1970. Structural and Functional Considerations in the Modelling of Biological Tukker, A. (Ed.), Eco-Efficiency in Industry and Science Springer Science + Business Organization. The Center. Media Dordrecht, Dordrecht. Rosen, R., 2005. Life Itself: A Comprehensive Inquiry Into the Nature, Origin, and Fabri- Hernández-Mora, N., Ituarte, L., del, M., La-Roca, F., Calle, A.L., Schmidt, G., 2014. Inter- cation of Life. Columbia University Press, New York. basin water transfers in Spain: interregional conflicts and governance responses. In: Rosen, R., 2012. Anticipatory Systems: Philosophical, Mathematical, and Methodological Schneier-Madanes, G. (Ed.), Globalized Water. Springer, Netherlands, pp. 175–194. Foundations, second ed. Springer-Verlag, New York. Howarth, C., Monasterolo, I., 2017. Opportunities for knowledge co-production Scanlon, B.R., Ruddell, B.L., Reed, P.M., Hook, R.I., Zheng, C., Tidwell, V.C., Siebert, S., across the energy-food-water nexus: making interdisciplinary approaches 2017. The food-energy-water nexus: transforming science for society. Water Resour. work for better climate decision making. Environ. Sci. Policy 75, 103–110. Res. n/a-n/a https://doi.org/10.1002/2017WR020889. https://doi.org/10.1016/j.envsci.2017.05.019. Shannak, S., Mabrey, D., Vittorio, M., 2018. Moving from theory to practice in the water- Junta de Andalucía. 2013. Annual agricultural and fisheries statistics. Avail- energy-food nexus: an evaluation of existing models and frameworks. Water-Energy able at (last access Sep 2018): https://www.juntadeandalucia.es/organismos/ Nexus. https://doi.org/10.1016/j.wen.2018.04.001. agriculturapescaydesarrollorural/consejeria/sobre-consejeria/estadisticas/paginas/ Söderberg, C., 2016. Complex governance structures and incoherent policies: implement- agrarias-anuario.html. ing the EU water framework directive in Sweden. J. Environ. Manage 183 (Part 1), Junta de Andalucía. 2015. Andalusian mediterranean basins water district. Management 90–97. https://doi.org/10.1016/j.jenvman.2016.08.040. Plan 2015-2021. Available at (last access Sep 2018): https://www.juntadeandalucia. Stein, C., Pahl-Wostl, C., Barron, J., 2018. Towards a relational understand- es/medioambiente/site/portalweb/menuitem.7e1cf46ddf59bb227a9ebe205510e1 ing of the water-energy-food nexus: an analysis of embeddedness and gov- ca/?vgnextoid=4d01df1837fa1510VgnVCM2000000624e50aRCRD&vgnextchannel= ernance in the Upper Blue Nile region of Ethiopia. Environ. Sci. Policy. 953d4ae7a9aa1510VgnVCM2000000624e50aRCRD. https://doi.org/10.1016/j.envsci.2018.01.018. Kampis, G., 1987a. Some problems of system descriptions I: function. Int. J. Gen. Syst. 13 Tolon-Becerra, A., Lastra-Bravo, X.B., Fernández-Membrive, V.J., 2013. Huella hídrica (2), 143–156. https://doi.org/10.1080/03081078708934964. y sostenibilidad del uso de los recursos hídricos. Aplicación al Poniente Alme- Kampis, G., 1987b. Some problems of system descriptions II: information. Int. J. Gen. Syst. riense. M + A. Revista Electrónic@ de Medio Ambiente 14 (1), 56–86. Available at: 13 (2), 157–171. https://doi.org/10.1080/03081078708934965. https://www.ucm.es/data/cont/media/www/pag-41205/61articulo.pdf (lastly ac- Keairns, D.L., Darton, R.C., Irabien, A., 2016. The energy-water-food nexus. cessed Sep 18). Annu. Rev. Chem. Biomol. Eng. 7, 239–262. https://doi.org/10.1146/annurev- Valera Martínez, D.L., Belmonte Ureña, L.J., Molina Aiz, F.D., López Martínez, A., chembioeng-080615-033539. 2016. Greenhouse agriculture in Almería. A Comprehensive Techno-Economic King, C.W., Carbajales-Dale, M., 2016. Food–energy–water metrics across scales: Analysis. Cajamar Caja Rural Available at (lastly accessed Sep 18): http://www. project to system level. J. Environ. Stud. Sci. 6, 39–49. https://doi.org/10.1007/ publicacionescajamar.es/pdf/series-tematicas/economia/greenhouse-agriculture-in- s13412-016-0390-9. almeria.pdf . Louie, A.H., 2009. More Than Life Itself: A Synthetic Continuation in Relational Biology, Venghaus, S., Hake, J.-F., 2018. Nexus thinking in current EU policies – the in- 1. Ontos Verlag. terdependencies among food, energy and water resources. Environ. Sci. Policy. Louie, A.H., 2013. The Reflection of Life: Functional Entailment and Imminence in Rela- https://doi.org/10.1016/j.envsci.2017.12.014. tional Biology, 2. Springer. Weitz, N., Strambo, C., Kemp-Benedict, E., Nilsson, M., 2017. Closing the governance gaps Louie, A.H., 2017. Intangible Life: Functorial Connections in Relational Biology. Cham in the water-energy-food nexus: insights from integrative governance. Glob. Environ. Springer International. Change 45, 165–173. https://doi.org/10.1016/j.gloenvcha.2017.06.006. 269