A framework for designing multi-functional agricultural landscapes: Application to Guadeloupe Island

To improve agriculture faced with regional sustainability issues, agricultural landscapes providing a diversity and high level of ecosystem services are necessary. We have developed and tested the MOSAICA-f framework to build innovative multi-functional agricultural landscapes that can consider explicitly: 1) the performance of cropping systems at the field scale, 2) farmers' decision processes on the adoption of cropping systems, and 3) possible scenarios for innovations and policy changes at the regional scale. This framework is based on a scenario approach that encompasses normative, exploratory and optimized scenarios to assess the relevance of combinations of new agricultural policies, changes to the external context (market and regulations) and innovations in cropping systems. The impacts of these changes on sustainability issues are simulated using the regional bioeconomic model MOSAICA for farmers' decision processes regarding the adoption of cropping systems at the field scale throughout a region. Applied in Guadeloupe (French West Indies), the MOSAICA-f framework enabled the design of a scenario increasing agricultural added value, food and energy self-sufficiency, employment and the quality of water bodies and reducing greenhouse gas emissions. This sustainable scenario combines new cropping systems tuned to farm types with a reorientation of subsidies, an increased workforce and banning food crop production on polluted soils. It can be used to understand the potential contribution of agriculture to sustainability issues and to help local decision makers define policies that will account for the spatial diversities of farms and fields in a landscape. Beyond the design of such a win-win scenario, MOSAICA-f has revealed trade-offs in the provision of services by agriculture.


Introduction
Agricultural landscapes account for one third of the land used by humans worldwide 45 (FAOSTAT 2008). While agriculture has constantly increased food production, it is 46 responsible for other positive and negative environmental, economic and social impacts at the 47 global and local scales (Tilman et al., 2002). Although agriculture can ensure the production 48 of food, energy, materials and services for society (including the alleviation of poverty), 49 agriculture faces several sustainability problems, such as climate change and water and soil 50 pollution. The ability of agriculture to provide multiple services in a sustainable manner is 51 therefore being questioned (Klapwijk et al., 2014).

53
Agronomists have been designing new agricultural systems at the field and farm scales in 54 order to improve sustainability. However, the design of innovative agricultural systems at 55 these scales has certain limitations when addressing regional and global issues. For instance, 56 at the field scale, some cropping systems may fail to respond to sustainability issues defined 57 at the regional scale because of the low scaling integration and spatial heterogeneity at the 58 regional scale (Dale et al., 2013). Agronomists must therefore integrate a landscape 59 perspective when designing new agricultural systems adapted to local regions, and when 60 addressing sustainability challenges at the regional scale (Dale et   To determine whether a particular combination of factors such as agricultural policies (e.g.   Moreover, some of these studies do no account for interactions between scales when trying to 89 identify the factors driving spatial dynamics (Houet et al., 2014). Several modelling 90 frameworks do not integrate the regional scale when assessing the services provided by  The framework presented in the paper aims to use the MOSAICA bioeconomic model in an 120 iterative manner in order to aid the building of multi-functional agricultural landscapes. The 121 model is applied in several steps involving different types of scenarios in order to understand 122 the potential for improvements to the landscape in terms of their contribution to regional 123 issues and to identify relevant drivers for change that will optimise their contribution.

No Yes
Optimization of the sum of the farmer's utilities (U*) Step 1: Reference mosaic The reference values of the indicator of interest, Yref, is obtained => Step 2 Step 3: Exploratory scenario Drivers tested to obtain the value of Yexpl Step 4 • If Yexpl > Y * => Use the Go sustainable scenario Step 5: "Go sustainable" scenario Drivers from step 4 are combined here Indicator providing information of the response to the sustainability issue of interest (Y) Step 2: Optimized scenario      The scenarios defined during the pre-modelling component were run using the MOSAICA    (Table 1).

303
Normative scenarios are parameterized in step 4 using the same drivers as those used in step 3 304 of the exploratory scenario and by implementing a constraint equation at the regional scale in 305 order to reach the "target value" obtained from the optimized scenario. identified at the regional scale, several indicators were used to assess the contribution of The MOSAICA framework was tested in Guadeloupe, an island located in the Caribbean.

341
This territory presents suitable conditions for implementing the framework for several 342 reasons. First, due to its insularity, flows of agricultural products are recorded at both entrance 343 to and exit from the territory (Agreste, 2011;INSEE, 2012). Second, Guadeloupe has to deal 344 with many local issues that limit the economic, environmental and social sustainability of the sufficiency, a high level of unemployment and a risk of pollution of water resources by 347 pesticides (rivers and drinking-water abstractions) used for local consumption (PDRG, 2011).

348
Another challenge is to "decrease food contamination due to chlordecone in soils".

349
Chlordecone is a remnant pesticide that was used between 1965 and 1993 on 15% of Increasing energy self-sufficiency   The sustainability goals for agriculture in Guadeloupe are to: i) increase crop production for 399 local markets, ii) increase biomass production for electricity production, iii) decrease the risks

405
The contribution of agricultural systems to greenhouse gas emissions was also evaluated 406 because it is a key component in efforts to mitigate climate change. Indicators were first of all 407 calculated for the reference cropping system mosaics obtained from the calibration (Table 2).    However, the decrease in this value from 4.5 to 3.5 was significant. A normative 481 scenario also needs to be drawn in step 4 to reach the target value.  The drivers used for each scenario are described in Table 2. We considered these drivers of change as being effective in reaching the set of target values 514 when using the optimized scenario in step 2 because the average contribution to other issues 515 increased by 8% for the "increase energy self-sufficiency" issue and only decreased by 3% for 516 the "decrease of the risk of pollution of water resources" issue, which was below the 20% 517 threshold set in the framework (Table 1).

518
All of the drivers tested under the exploratory scenarios helped to reach or exceed the target 519 values fixed by the optimized scenarios. When the drivers did not reach these values, we 520 noticed that reaching them under the normative scenarios had no significant negative side effects. Next, these drivers were combined in step 5 under a "Go sustainable" scenario, which 522 reflects optimization of the overall farmers' utilities for the selected political, agronomic or 523 external drivers of change. The contributions of the different sustainability issues are presented in Figure 3 and can be 546 used to analyse the relationships between the different sustainability issues. between the current mosaic (on the left) and the "Go sustainable" scenario (on the right).

559
The contributions of the different cropping systems to sustainability issues can also be 560 analysed spatially. We illustrate this spatial analysis in Figure 4, which shows the changes in 561 the spatial variations of the contributions of sub-regions to food and energy self-sufficiency 562 and the increases in overall agricultural added value and agricultural added value from local 563 foodstuffs. Using the same method, the spatial variability of the contributions of cropping 564 system mosaics to local issues is displayed at the sub-regional and field scales in order to

A framework to guide the scenario-based integrated analysis of agricultural systems 643
The MOSAICA-f framework can help to parameterize a multi-functional scenario to improve 644 the contributions of agricultural systems at a regional level to several sustainability issues. To can or cannot achieve these regional goals, and exploratory scenarios are helpful when 661 selecting a set of drivers to meet these goals. For each goal, the targets defined with the 662 optimized scenarios can provide information on the structural gap between the reference 663 cropping system mosaic and the optimal cropping system mosaic for a given sustainability been implemented at the field, farm, sub-regional and regional scales, and specifically 697 targeted certain fields, farms or sub-regions. Thus, in our pathway for scenario building, we This framework may be of particular use to inform regional planning because it generates 711 optimal outcomes at the regional scale and provides information on the spatial organization of that are likely to be more efficient than regional policies. We therefore hypothesize that the scale. In addition, the results of this study show that it is important to account for spatial 789 heterogeneity in regional studies, and also to consider multiple drivers when the aim is to 790 achieve multi-functional agriculture. This proposed framework could help decision makers, farmers and society understand the pathways needed to achieve transition towards a more 792 sustainable future in regions where significant investments are made in data acquisition at the 793 field, farm and regional scales.