Crop-livestock integration, from single practice to global functioning in the tropics: Case studies in Guadeloupe

Europ. J. Agronomy 80 (2016) 9–20

Contents lists available at ScienceDirect

European Journal of Agronomy journal homepage: www.elsevier.com/locate/eja

Crop-livestock integration, from single practice to global functioning in the tropics: Case studies in Guadeloupe Fabien Stark a,b,c,∗ , Audrey Fanchone c , Ivan Semjen d , Charles-Henri Moulin e , Harry Archimède c a

CIRAD, UMR SELMET, Montpellier SupAgro, 2 place Pierre Viala, 34060 Montpellier Cedex 1, France AgroParisTech, Centre de Montpellier, 648 Rue Jean Franc¸ois Breton, 34090 Montpellier, France c INRA, URZ, Centre Antilles-Guyane, Domaine Duclos, Prise d’Eau, 97170 Petit Bourg, Guadeloupe, France d ISTOM, 32 Boulevard du Port, 95000 Cergy, France e Montpellier SupAgro, UMR SELMET, Montpellier SupAgro, 2 place Pierre Viala, 34060 Montpellier Cedex 1, France b

a r t i c l e

i n f o

Article history: Received 18 December 2015 Received in revised form 9 June 2016 Accepted 13 June 2016 Keywords: Mixed crop-livestock systems Diversity Complexity Network analysis Agroecology Caribbean

a b s t r a c t Agricultural systems will have to produce more and better in a changing world. Mixed crop livestock systems (MCLS) are sound alternative ways to progressively achieve these goals through crop-livestock integration (CLI). CLI exploits the synergies between cropping and livestock systems, for example, through organic fertilization and the use of crop residues to feed livestock, and offers many opportunities to improve productivity, as well as to increase resource use efficiency and improve the resilience of the whole farming system. In the scientific literature, authors advocate the interest of MLCS and CLI, based on theoretical considerations, modelling and empirical evidence from local case studies. But these studies do not clearly identify the respective roles of the diversity of activities and CLI management practices in improving performances at the level of the whole farming system. The aim of this study was thus to assess CLI at farm scale in a range of MCLS and to explain farm performances by analyzing the combination of activities and the level of integration. This study was conducted in Guadeloupe, (French West Indies), where MCLS and CLI are complex but important challenges for local agricultural. Ecological network analysis was used to study the structure, functioning and performance of agrosystems. To this end, a range of eight farms was selected to characterize CLI as practices, and as a network of nitrogen flows at farm level. The land and labor productivity were then assessed along with the resilience, efficiency, productivity and self-sufficiency of the network of flows. Results show that CLI only applies to certain types of production, including feeding pigs with a wide range of crop residues (crop residues provide from 16 to 45% of the N supply to pigs) or organic fertilization of small market gardens and plots used to grow tubers (manure provides 24–100% of the N supply to plots). But at whole system level, CLI remains low: in seven cases, the N circulating within the system – ICR- represent only between 0.7 to 3.5% of the total N circulating through the system; only one farm presents a higher intensity of CLI, with an ICR of 18.9%. Consequently, performances and especially efficiency and productivity, depend more on the nature of the activity than on CLI management practices. © 2016 Elsevier B.V. All rights reserved.

1. Introduction

∗ Corresponding author at: CIRAD, UMR SELMET, Montpellier SupAgro, 2 place Pierre Viala, 34060 Montpellier Cedex 1, France. E-mail addresses: [email protected], [email protected] (F. Stark). http://dx.doi.org/10.1016/j.eja.2016.06.004 1161-0301/© 2016 Elsevier B.V. All rights reserved.

Agricultural systems will need to produce more and better in a changing world (Griffon, 2006). This will require adaptation to the increasing scarcity of natural resources while responding to the demands of an increasing global population. Designing local sustainable agricultural systems is a necessarily converging task (Dumont et al., 2012; Herrero et al., 2010). Such agricultural systems need to be simultaneously efficient, productive, resilient and

10

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

self-sufficient, as proposed by agro-ecology (Altieri et al., 2012; Bonaudo et al., 2014; Gliessman, 2004). Mixed crop-livestock systems (MCLS) are sound alternative ways to progressively achieve these goals (Altieri, 2008; GonzalezGarcia et al., 2012). As stressed by Herrero et al. (2010), the synergies between cropping and livestock husbandry offer many opportunities to achieve sustainable agriculture, by increasing productivity in parallel with resource use efficiency. These synergies refer to the use of biomass from crops to feed animals, of manure to maintain soil fertility, of animal traction for tillage and transport (Lawrence and Pearson, 2002). In developing countries in intertropical areas, agriculture is frequently supported by MCLS, often in a context of smallholder agriculture (Steinfeld et al., 2006; Udo et al., 2011), with low external inputs (Schiere et al., 2002). In developed countries in temperate areas, although modernization has resulted in specialized farming systems (crops versus livestock), in a context of high external inputs, MCLS are still on the agenda (Ryschawy et al., 2014). The complex resource and energy exchanges between crop and livestock production (Sumberg, 2003) correspond to crop-livestock integration (CLI). Many analytical studies are reported in the literature. Their aim is to identify technical levers to improve CLI, such as evaluation of the feed value of unconventional feed resources (Renaudeau et al., 2014; Xandé et al., 2007) or ways of managing manure to enhance the organic matter content of soils and increase crop yields (Sierra et al., 2013). But none of these studies consider CLI practices at the level of the whole farming system. At farm or regional levels, many authors advocate the interest of MLCS and CLI, based on theoretical considerations or modelling and empirical evidence from local case studies (Devendra and Thomas, 2002; Moraine et al., 2014; Thornton and Herrero, 2001). But the respective roles of diversity (mixing activities) and CLI management have not yet been clearly explained, although recent studies are progressing in this direction. For instance, using a modelling approach to a crop and sheep farming system, Sneessens et al. (2014) demonstrated the importance of the respective weight of the two activities in the efficiency of the system. To assess the impact of CLI and diversity on food self-sufficiency at farm scale, Rufino et al. (2009a) proposed a method called ‘ecological network analysis’ (ENA), which measures the level of integration. However, these approaches do not treat the configuration and intensity of the diversity of flows in CLI at farming system level. Moreover, the agroecological dimension of CLI is not apprehended by current methods. The aim of the present study was to apply farming system framework to ENA to assess CLI at farm scale in a variety of MCLS. This framework was applied to case studies in Guadeloupe (French West Indies), to analyze CLI at whole system level, with the goal of providing a new perspective on agroecological features of agriculture. First, the variety of CLI practices implemented in Guadeloupe was characterized to assess the level of CLI at farm scale and to evaluate the agroecological performances of whole MCLS. ‘Agroecological performances’ refers to the resilience, efficiency, productivity and self-sufficiency of farming systems (Altieri et al., 2012; Bonaudo et al., 2014). To this end, a range of MCLS was selected based on access to production factors and combinations of activities. ENA was then used to characterize CLI as practices and to describe the network of nitrogen flows at the level of the system. Nitrogen has been identified as the main limiting nutrient both for livestock and crop production and also as a source of pollution for agrosystems (Giller et al., 1997; Rufino et al., 2009a). The nitrogen concentration in animal feed along with the nitrogen used for crop fertilization determine feed supplementation with concentrates and the cost of mineral fertilizer, both of which are imported into Guadeloupe and are becoming increasingly expensive. The land and labor productivity of MCLS was assessed along with the resilience,

efficiency, productivity and self-sufficiency of the network of flows. The results are discussed according to their implication for future Guadeloupian MCLS, compared with the results of CLI in the literature, and the prospects of using ENA to analyze CLI. 2. Study site The analysis was conducted in Guadeloupe, a French Overseas Department, located in the Caribbean (latitude 16◦ 13 N, longitude 61◦ 34 W). The Guadeloupian archipelago (1434 km2 ) comprises two main islands (Basse-Terre and Grande-Terre) with different soil and climatic conditions. Basse-Terre (848 km2 ) is volcanic and is the wettest part of the archipelago whereas Grande-Terre (586 km2 ), the driest part, is limestone. Sierra et al. (2015) divided the archipelago of Guadeloupe into five agro-ecological regions. Soils are vertisols in Grande-Terre and in the eastern part of BasseTerre and mean rainfall is 1100 and 900 mm/yr, respectively. Soils are ferralsols in the northern part of Basse-Terre with mean rainfall of 2300 mm/yr, andosols in the central part of Basse-Terre with mean rainfall of 3800 mm/yr and nitisols in the southern part of Basse-Terre with mean rainfall of 2200 mm/yr. For more details on the soil and climatic conditions in Guadeloupe, see Sierra et al. (2015). Guadeloupian agriculture is mainly based on small MCLS, which represent 80% of the farms in the territory with an average size of 4.1 ha, and at the same time is oriented toward exports of highly subsidized agricultural products that require large quantities of external inputs. The main crops are sugarcane and banana, which account for respectively 14,000 ha and 2500 of the 31,400 ha of agricultural land on the island (Agreste, 2015). Rearing small and large ruminants is also a widespread traditional practice in this area, and currently accounts for 10,000 ha of pasture (Mahieu et al., 2008). Food crop and small livestock systems are also associated with the two main crops at farm scale, they are less subsidized and target the local market. These systems include market gardening, orchards, tuber and fruit production, pig, poultry and rabbit breeding. Products destined for the local market do not meet local demand, and the agricultural trade balance shows a large deficit (Agreste, 2015). Moreover, both livestock and crop production depend to a great extent on imported and increasingly expensive feed concentrate and mineral fertilizer. Guadeloupe is thus an interesting case because it combines a variety of farming systems, in a tropical context, but in the institutional context of the European Union (Archimède et al., 2012). CLI involves complex issues but is an important stake for local agronomic research and development institutions in the new agroecological paradigm (Gonzalez-Garcia et al., 2012; Ozier-Lafontaine et al., 2011; Stark et al., 2010). 3. Method 3.1. Sampling: typology of MCLS in Guadeloupe and choice of case studies An exploratory study was conducted in 2010 to have an overview of the forms of MCLS in Guadeloupe. Despite the fact that MCLS are the main form of farming, information concerning them is limited, due to a strong chain organization of agriculture. A sample of 111 farmers was selected, based on the non-probabilistic ‘snowball’ sampling method (Harper et al., 2013). In order to cover the territory homogenously, the sample was readjusted to take into account four of the five agro-ecological areas of the territory (Sierra et al., 2015). The central part of Guadeloupe was not included, as it is a preservation area where agricultural activities are limited. Qualitative field surveys enabled the collection of two types of data on each farm: socio-economic data (land tenure, labor, household,

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

11

Table 1 Three sets of structural, functioning and performance analysis indicators for agrosystems adapted from ENA (Latham, 2006).

Structural analysis Diversity of flows

Organization of flows

Indicators

Mathematical formula

Number of links within the network Number of compartments Density of internal links

Fi n Fi/n

Average mutual information

AMI = k

n n+2   T

ij

T.. i=1 n

T

Statistical uncertainty

Hr = −

Realized uncertainty

AMI/Hr

Total system throughflows

TST =

.j

T..

T T..

log2 Tij T

i. .j

j=0 T

log2 T...j

j=0

Functioning analysis Intensity of flows

n 

Ti i=1

Circulation of flows

j=n 

TT =

Total internal throughflows

fij i=1

Cycling of flows

Internal circulation rate

ICR = TT/TST

Finn’s cycling index

FCI =

Development capacity

C=−



Performance analysis Resilience

TSTc TST

A=

Ascendancy

Tij log

i.j Tij log

Ti. T.j





=−

T..

 Tij T.. 

i.j

Overhead

 Tij 

Tij log

T2



ij

Ti. T.j

i.j

Productivity (P)

Outflows/Total system Throughflows

P=

1 TST

n 

YOi

 Self-sufficiency (SS)

(TST − Inflows)/Total system Throughflows

SS =

i=1

TST −



n 

ZOi

/TST

i=1

Efficiency (Eff)

Outflows/Inflows

Eff =

P SS

k Constant scalar in the AMI equation. Ti. Total inflow for compartment i. T.j Total outflow for compartment j. fij .Tij Flow from compartment j to compartment i. T.. Total system throughput (sum of the network links). Ti =

n 

fij + Zi0 − (x˙ i )− Compartmental throughflow.

j=1

(x˙ i )− Negative state derivative for compartment i. fij .Tij Flow from compartment j to compartment i. Zi0 .Zj0 Flow into compartment i or j from outside the network. TSTc Total system cycled throughflow. Y0i .Y0j Outflows (usable flows) from the network for compartment i or j.

equipment, sources of income), and crop and livestock activities (cropping pattern, number of animals, management practices). A typology of production systems was built, in which a production system is considered to be a combination of production factors and productions (Brossier, 1987). Due to the wide range and varying proportions of crop and livestock activities per farm, the choice was made to characterize the combination of products using the intensity of production per hectare of the whole combination of activities (Stark et al., 2012). Two discriminant variables were used in a graphic analysis (Bertin, 1977) to distinguish types of production systems: access to land (a limiting factor for production on this small island) and the intensity of the combination of products. Next, access to other production factors were used as explanatory variables for each type of production system: workforce, with the

number of hectares cultivated per family labor unit to include labor intensity, and the equipment and buildings (fixed capital), in a qualitative way with three modalities, from low to high access to fixed capital. Among the 111 farms in the sample, eight case studies were identified because 1/they were considered as representative of the different types in the typology (products and productions factors), 2/they cover the diversity of agro-ecological zones in Guadeloupe, and 3/no substantial changes in farm structure and functioning had taken place since the exploratory survey in 2010. Farmers on the eight sample farms were interviewed in 2014 or at the beginning of 2015, considering the same previous annual agricultural campaign (from March 2013 to February 2014).

12

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

Fig. 1. Conceptual model proposed for the ENA analysis of case studies in Guadeloupe.

3.2. Data collection: comprehensive approach to farming systems To characterize the functioning of MCLS and CLI practices on the eight farms, information was collected using a comprehensive approach to farming systems (Bonneviale et al., 1989). Information at whole system level is needed to describe how a farming system is run, i.e. the coherent management of productive resources that are combined through activities (crop and livestock systems; Devienne and Wybrecht, 2002). This approach consists in iterative semi-structured interviews during which qualitative and quantitative data are collected on how the farming systems function. This information concerns the components of the farming system and any interactions between the components, i.e., the socio-ecological environment, the decision-making system, allocation of the production factors, and the structure and functioning of the crop and livestock systems. For the characterization of CLI, particular attention was paid to interactions between crop and livestock activities, so as to qualify these practices in terms of logic, objectives and constraints as well as to quantify them in terms of the intensity and distribution of flows. Information concerning crop and livestock management practices including organic fertilization, crop products and by-products used as animal feed was recorded. 3.3. Data processing: socio-economic analysis of MCLS The socio-economic performances of farming systems were assessed through added value (AV) which measures the richness generated by the production system. AV is equal to the difference between the value of the goods that are produced (gross product) and the value of goods and services that are consumed during the production process (intermediate consumption). The AV of the farming system is the sum of the AV of each cropping and livestock

activity. Land productivity (AV/agricultural area) and labor productivity (AV/working day) were obtained by dividing the AV by the units of production factor (land or work) used in the production process. Land productivity assesses the result of the intensification of the production process, while labor productivity is related to the economic efficacy of labor incorporated in the production process (Cochet, 2012). Because the majority of subsidies are linked to production, the sum of the AV and the subsidies related to specific activities at farm scale were also considered. 3.4. Biotechnical analysis of MCLS Ecological network analysis (ENA) was used to assess the agroecological performances of MCLS, previously adapted to farming system analysis by Rufino et al. (2009a). ENA comprises a systemic and holistic analysis of interrelated species in an ecosystem (Ulanowicz, 2004). The characterization of the structure, functioning and performance of farming systems is based on a set of indicators derived from ENA. Structure and function analysis makes it possible to characterize CLI practices at system level while performance analysis makes it possible to assess the agroecological performances of MCLS (Table 1). The first step of ENA consists in implementing a network model of flows to represent all compartments and their interactions (Fath et al., 2007), plus the interactions between the compartments and their environment. When all the ecological compartments and interactions have been identified, all the flows in the network have to be quantified in a single appropriate unit. The second step of ENA is calculating the structure, functioning and performance indicators. Structural analysis was performed to characterize the organization of the system, which depends on the diversity of flows that make up the network, and on the size of the flows. Func-

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

13

Table 2 Characteristics of the three types of mixed farms identified in Guadeloupe.

N

Small labor intensive (SIL) 51

Medium capital intensive (MCI) 23

Medium extensive (ME) 37

Intensity Standard gross margin/ha (D )

7501 (6466)

7840 (2331)

3070 (983)

Land Area (ha)

4.4 (2.0)

14.1 (8.3)

14.0 (5.5)

Capital Farms with limited access to capital (%) Farms with medium access to capital (%) Farms with high access to capital (%)

41 53 6

9 57 35

24 41 35

Labor Family labor (n) Farms with outside employees (%) Family labor/ha (n)

1.1 (0.5) 43 0.25

1.4 (0.5) 43 0.10

1.4 (0.5) 43 0.10

Livestock rate Percentage of standard gross margin accounted for livestock

46 (27)

31 (24)

39 (26)

xx: mean or proportion; (xx): standard deviation. In italics: explanatory variables.

tional analysis was performed to evaluate the activity of the system, which is related to all the agricultural practices used in the management of the productions concerned. In the conceptual framework of agro-ecology, the proposed performance analysis is linked to four dimensions: resilience, productivity, self-sufficiency and efficiency (Bonaudo et al., 2014; Ulanowicz et al., 2009).

4. Choices for the implementation of ecological network analysis (ENA) 4.1. System conceptualization The farms were depicted using a single conceptual model, corresponding to an aggregation at community level in ecosystem analysis, in order to compare systems with different crop and livestock production systems (Fig. 1). The homogenization of a system in a common conceptual model is a necessary task not only in ENA but generally in modelling, because the results of analysis are a function of the level of aggregation chosen to represent reality. The objective is to characterize the diversity of CLI practices precisely at system level. To achieve this objective, it is important to consider not only one livestock and one crop compartment but several compartments, to represent the diversity of agricultural practices as a network. Consequently, this conceptual model does not count all the cropping and livestock productions that exist in Guadeloupe but brings together different species. It corresponds to the main farming activities according to their agronomic features (crop cycle, species, storage, etc.) and their aim (animal feed, human food, fertilization, export, etc.). Sugar cane and banana are represented as specific compartments, due to their central role in Guadeloupian agriculture (Agreste, 2015). The four other cropping compartments include crops traditionally grown in Guadeloupe, as a function of their production cycle and their management practices: tubers (cultivated on mounds, with a medium term cycle), market gardens (short cycle, associated crops), fruit (medium term, specialized) and agroforestry (perennial crops). The livestock compartments correspond to animal species traditionally raised in Guadeloupe: cattle, pigs, poultry and rabbits. The cattle compartment includes the herd and the natural pasture (spontaneous vegetation) used as the basis of their diet with no specific management practices (e.g. fertilization, weeding). However, some indicators were calculated considering natural pasture as a separate compartment in order to assess the importance of grazing pasture in the functioning of MCLS. Note that this conceptual model does

not include a specific fodder compartment, because of the farmers interviewed do not dedicate specific parts of their farm to feeding animals but use non-farming land to graze their cattle and crop residues as animal feed. The storage compartments considered are manure storage and forage storage. The interest of representing storage compartments is to distinguish between “direct” CLI practices such as the application of fresh manure or the distribution of whole sugarcane and “processed” CLI practices, such as compost, or using crushed sugarcane to feed animals. Since this is a static study based on an annual averages, multiannual cropping systems (perennial crops, crop rotations) were taken into account through the normalization of all flows during the production cycle on an annual basis. The demographic dynamics of livestock systems were also normalized on an annual basis. The flows selected were those that correspond to flows of material between compartments, and between compartments and the socio-ecological environment. Integration flows correspond to (1) flows of fertilizer from the livestock and manure storage compartments to the crop compartments and (2) flows of animal feed from the crop and forage storage compartments to the livestock compartments. Inflows corresponds to flows of fertilizer (mineral fertilizer or manure from another farm, for example), flows of animal feed (concentrates, supplements, forage originating outside the farm, crop residues, cereals, for example), and living plant material, live animals for fattening, or animals to be used for reproduction. Outflows correspond to usable exported products (live animals, meat, crop yields, forage, crop residues or manure used by other actors). Losses correspond to sources of pollution such as unused manure, emissions from livestock manure (building, storage and grazing) and from crops (manure and mineral fertilizer applications). Crop residues are not considered as losses, as they are returned to the soil and are thus a potential source of nutrients for the same compartment (self-flows are not take into account in ENA calculations).

4.2. System modelling In this study, the network of flows resulting from the conceptualization of a system is expressed as nitrogen (N), due to the important role of N for both crop and livestock development (Rufino et al., 2009a). Several methods were used to quantify flows. First, available data were collected from farmers in iterative interviews. These data concerned the quantity of animal feed and of fertilizer, the composition of the feed concentrates, the composition of

14

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

Table 3 Area, labor and size of the herd and crops on the eight mixed farms in Guadeloupe. SIL1

SIL2

MCI1

MCI2

ME1

ME2

ME3

ME4

Production factors Area (ha) Herds (TLU* ) Family labor (labor unit) Labor (man-day)

2.7 6.0 1 288

2.5 27.5 1 156

16.6 26.8 2 301

13.0 23.5 1 284

10.0 9.2 1 145

12.9 17.6 1.5 137

14.6 32.8 1 382

13.0 2.4 2 1335

Forage crops Forage area (% of total area)

0.0

59.5

24.1

0.0

30.0

46.5

24.0

0.0

Food crops (ha) Banana (for export) Sugarcane (for export) Fruit crops Forestry-Arboriculture Market garden crops Tuber crops

– – 0.5 0.2 2.0 –

– – – 1.0 – 0.2

– 9.0 0.1 – 1.0 2.5

– 6.0 5.0 – 0.5 1.5

– 6.0 – – 1.0 –

– 6.8 – – – 0.1

– 10.0 0.1 – 0.5 0.5

9.0 – – – – –

Livestock (TLU) Polygastrics (cattle) Monogastrics (pigs, poultry, rabbits)

– 6.0

2.4 24.4

20.0 6.8

– 23.5

4.8 –

17.6 –

32.8 –

– 2.4

*

TLU: Tropical Livestock Unit.

Table 4 Rates of nitrogen provided by the crop-livestock integration practices on the eight mixed farms in Guadeloupe.

SIL1 SIL2 MCI1 MCI2 ME1 ME2 ME3 ME4

Animal feeding practices

Organic fertilization practices

Market garden crops and arboriculture residues used to feed pigs (16%) and poultry (1%) Fruit and agroforestry residues used to feed pigs (13%) and rabbits (3%) Crop residues (banana, tubers, and market garden crops) and sugarcane used to feed pigs (13%) Crop residues (banana, tubers, and market garden crops) and sugarcane used to feed pigs (18%) and poultry (10%) Sugarcane and market garden crop residues used to feed pigs (45%) Sugarcane used to feed cattle (1%) Fruit residues used to feed cattle (0.1%) Banana residues used to feed pigs (29%)

Poultry manure used to fertilize market gardens (24%) Rabbit manure used to fertilize tubers (70%) Pig manure used to fertilize market gardens (1.2%) Pig and poultry manure used to fertilize market gardens (55%) Pig manure used to fertilize market gardens (100%) – Cattle manure used to fertilize market gardens (7%) –

Rate of Nitrogen: Nitrogen provided by the CLI practice/Total N supplied to the compartment.

mineral fertilizer, crop and livestock production, organic fertilization, the distribution of crop residues, etc. From these data, annual flows of materials were calculated. The flows of N were then calculated from the material flows and the N content of the material was estimated using data in the literature (see Appendix A of Supplementary material). The flows that could not be calculated this way, i.e. N emissions from manure, the amount of animal dejections and grass intake were estimated from data proposed by Peyraud et al. (2012) according to the management practices and the species concerned (Appendix A of Supplementary material). Based on collected and estimated data, a matrix was drawn for each case study. The columns show the origin of the flows (inflows from outside and from other compartments) and the rows show the destination of the flows (compartments, outflows and losses). Based on the flow matrix, the indicators in Table 1 were calculated using algorithms developed by Rufino et al. (2009a) and by the authors of this paper using a spreadsheet. 5. Results 5.1. Diversity of mixed farming systems 5.1.1. Typology of MCLS Using the sample of 111 farms (Table 2), three types of MCLS were distinguished based on the intensity of production factors and the combination of production: small labor intensive farms (SIL); farms with moderate capital investment (MCI); and medium size extensive farms (ME), which represented respectively 46%, 21% and 33% of the sample. SIL and MCI can be distinguished from ME

by their more intense combination of production (standard gross margin of 7501 and 7840 D /ha versus 3070 D /ha). ME and MCI can be distinguished from SIL by their larger agricultural area (14.0 and 14.1 ha versus 4.4 ha), their higher access to capital (35% versus 6% of farms with high access to capital), their less intensive family labor (0.1 versus 0.25 family labor/ha) and to some extent, to a lower standard gross margin from livestock (39% and 31% versus 46%). 5.1.2. Characteristics of the case studies Table 3 lists the characteristics of the eight farms selected. The names of case studies are indexed according to the type of MCLS to which they correspond. SIL1 and SIL2 correspond to small labor intensive farms (average area of 2.6 ha and average intensity of family labor of 0.4 labor unit/ha); MCI1 and MCI2 correspond to medium capital intensive farms (average area of 14.8 ha and average salaried labor of 293 man-days); and ME1, ME2, ME3 and ME4 correspond to medium size extensive farms (average area of 12.6 ha and average herds of 15.5 TLU). SIL1 and SIL2 differ in labor intensity, SIL1 grows high-value crops (market garden crops) using twice as much labor as SIL2 which breeds small livestock in an intensive system (rabbits, pigs, and poultry). MCI1 and MCI2 produce both sugarcane and high-value crops (market garden and tuber crops) but MCI1 owns grazing cattle with a high livestock density (5 TLU per ha), while MCI2 practices intensive pig and poultry production (23.5 TLU) and high-value production of pineapple (5 ha). ME1, ME2, and ME3 farms correspond to traditional Guadeloupian systems, with large areas under sugarcane and grazing cows, associated with a small area of high-value products (market garden or tuber crops).

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

15

Table 5 Values of structural and functional indicators based on ENA to characterize crop-livestock integration on the eight mixed farms in Guadeloupe.

Set 1. Structural indicators Diversity

Organization

Set 2. Functional indicators Intensity Internal circulation Cycling Indicators including pasture grazing*

SIL1

SIL2

MCI1

MCI2

ME1

ME2

ME3

ME4

n Fi Fi/n

5 5 1.0

7 4 0.6

7 6 0.9

7 7 1.0

5 4 0.8

3 1 0.3

6 3 0.5

2 1 0.5

AMI Hr 1-AMI/Hr

1.28 1.96 0.35

1.04 1.89 0.45

1.16 1.66 0.30

0.81 1.24 0.35

0.89 1.18 0.25

1.12 1.32 0.15

1.2 1.53 0.22

0.67 0.74 0.09

TST (Kg N/year) TST/ha TT (Kg N/year) ICR (%) FCI (%) TST/ha ICR (%) FCI (%)

346.3 128.3 65.5 18.9% 0.55% 128.3 18.9% 0.5%

447.4 177.5 15.5 3.5% 0.00% 211.3 17.5% 13.3%

2,684.0 161.7 19.2 0.7% 0.01% 280.8 41.0% 23.9%

3,419.0 263.0 69.0 2.0% 0.47% 263.0 2.0% 0.5%

1,376.7 137.7 11.1 0.8% 0.11% 151.5 9.9% 7.6%

2,242.9 173.9 16.8 0.7% 0.00% 325.2 39.0% 23.2%

3,274.1 224.3 43.1 1.3% 0.00% 428.2 44.5% 30.7%

3,802.1 292.5 44.8 1.2% 0.00% 292.5 1.2% 0.0%

See Table 1 for definitions of the indicators. In bold: indicators used for the analysis. * Cattle grazing considered as two compartments, pasture and cattle, with corresponding grazing and dejection flows between them.

The stocking rate of ME1 and ME2 lies between the two (1.6 and 2.9 TLU per ha of forage area), compared to ME3 with a higher stocking rate (9.4 TLU/ha). The last farm, ME4, is specialized in producing bananas for export, which is very labor intensive (1335 man-days per year), associated with a small-scale pig production (2.4 TLU).

5.2. Crop-livestock integration practices 5.2.1. Description of CLI practices Crop-livestock integration practices were first analyzed practice by practice, to get an idea of the relative importance of CLI according to the type of main agricultural practices (Table 4). Feeding animals with crop residues mainly concerned pig production. Whatever the nature of the residues (sugarcane, banana, tubers, arboriculture, market garden crops), the N provided by crop residues on the six farms that bred pigs ranged from 13% to 45% of the total N supplied. To a lesser extent, farms that raised poultry and rabbits also fed them with crop residues, but this only represented between 1% and 10% of the total N supplied. Crop residues were used as complementary cattle feed much less frequently, i.e. on only two out of the five farms that raised cattle, representing less than 1% of total N intake. Organic fertilization, which is based on the use of livestock manure, was only used for market garden crops and tubers. Pig, poultry, rabbit or cattle manure was used by six out of the seven farms that cultivated market garden crops or tubers to fertilize them. The rate of organic fertilization ranged between 2 kg N/ha (MCI1) and 73 kg N/ha (MCI2). The proportion of N supplied by manure varied from 1% to the total amount of N applied to fertilize these crops. This variability depended on the fertilization practices, as no mineral fertilizer was supplied in some cases (ME1) but reached 580 kg N/ha in others (ME3). When cattle were present (five case studies), N flows linked to grazing predominated, in both directions, i.e. from cattle to pasture (dejection) and from pasture to cattle (grass intake). Indeed, these flows represented from 88% to the total quantity of N distributed to the cattle and a large proportion of pasture fertilization (from 29% to the total amount of N applied).

5.2.2. Flow analysis of CLI Second, CLI was analyzed in terms of a network of flows to account for it as a whole (Table 5).

The number of compartments ranged from two to seven, corresponding to the range of crop and livestock activities observed in the exploratory survey, with an average of 4.3 activities. The number of throughflows, i.e. CLI flows, ranged from one flow for ME2 and ME4 to seven flows for MCI2. CLI practices only concerned certain activities in MCLS. The organization of flows also varied with the type of MCLS. Flows were more homogenous on SIL farms (highest values of 1AMI/Hr , between 0.35 and 0.45), and more heterogeneous on ME farms (lowest values of 1-AMI/Hr , between 0.09 and 0.25). The values on MIC farms were intermediate (1-AMI/Hr of 0.3 and 0.35). The intensity of the whole network of flows was much lower on SIL farms (TST < 450 kg N) than on medium size farms (TST > 1300 kg N). However, when expressed per area, the relative intensity values were comparable. Overall, TST/ha values ranged from 128.3 kg N/ha to 292.5 kg N/ha, depending on the combination of production and on agricultural management practices. The internal circulation rate, which summarizes the quantity of N circulating in throughflows as a function of the total circulation of flows (TST) was very low, except in SIL1 (19.9%). The other farms had an ICR of only 0.7% to 3.5% of N originating from throughflows. Cycling, measured by FCI, was nil or quasi nil (less than 1%) on all farms, even on SIL1, which had the highest intensity of throughflows. Four farms had a zero value, which is explained by the fact that, in these cases, throughflows connected different compartments, with no possibility for nutrients to return to their compartment of origin. Indicators were also calculated considering cattle grazing as two separate compartments, pasture and cattle. Results showed that grazing pasture plays an important role in flows of resources. TST/ha increased from 10% on farms with a low proportion of cattle grazing (SIL2 and ME1) to 91% on farms more specialized in cattle grazing (MCI1, ME2 and ME3). Similarly, ICR increased on these farms, making them more integrated systems in terms of the intensity of their internal flows. The farms that were most specialized in cattle grazing became the most integrated, with respectively 41%, 39% and 44.5% of flows originating from internal circulation. Concerning cycling, FCI, which was zero or close to zero on these farms, increased considerably, up to 30.7% on ME3. In the same way as flows from cattle to pasture (dejection), and complementary flows from pasture to cattle (grass), cycling is made possible by the strong intrinsic relationships between these two complementary compartments.

16

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

Table 6 Value of the agroecological performances of the eight mixed farms in Guadeloupe.

Resilience

Productivity Self-Sufficiency Efficiency

A (Kg N/year) ␾ (Kg N/year) C (Kg N/year) ␾/C Outflows (Kg N/year) P (%) Inflows (Kg N/year) SS (%) Eff (%)

SIL1

SIL2

MCI1

MCI2

ME1

ME2

ME3

ME4

775.3 1,079.9 1,855.2 0.58 132.7 38.3% 274.7 20.7% 48.3%

890.0 1,693.8 2,583.7 0.66 96.3 21.5% 414.7 7.3% 23.2%

4,842.7 7,524.0 12,366.7 0.61 932.6 34.7% 2,477.7 7.7% 37.6%

3,513.7 8,258.1 11,771.8 0.70 672.5 19.7% 3,350.0 2.0% 20.1%

1,589.9 1,490.5 3,080.4 0.48 429.8 31.2% 1,292.4 6.1% 33.3%

3,325.5 1,506.0 4,831.5 0.31 488.5 21.8% 1,881.5 16.1% 26.0%

4,713.9 6,808.6 12,522.5 0.54 1,011.9 30.9% 2,665.7 18.6% 38.0%

2,962.4 1,386.2 4,348.6 0.32 512.6 13.5% 3,757.3 1.2% 13.6%

See Table 1 for definitions of the indicators. In bold: Indicators used for the analysis.

5.3. Performances of MCLS 5.3.1. Agroecological performances Agroecological performances were analyzed according to the network of flows (Table 6). The resilience of the network of flows was higher on SIL and MIC farms, with /C between 0.58 and 0.70. These results correspond to the most diversified and the most organized flow network, the more connected the flow network, the greater the ability of the network to adapt. N productivity ranged from 13.5% to 38.3% across MCLS types. These results are mainly a function of the relative N efficiency of each activity combined in a given production system (Godinot et al., 2015). The low productivity of ME4 is linked to the low N productivity of banana production (13% of N exported). The low productivity of SIL2, MCI2 and ME2 is due to the lower productivity of livestock production. The higher productivity of SIL1, MCI1, ME1 and ME3 (38.2, 34.8, 31.2, and 31.0% respectively) is due to the higher N productivity of certain crops (market garden, tubers, pineapple and sugarcane), which are the main crops produced in these systems. N self-sufficiency was low in all cases, representing less than 20% of total activity. The N self-sufficiency of SIL2, MCI1, ME1 and ME4 (between 1.2 and 7.7%) was very low due to the high dependence on inputs, feed supplements in the case of SIL2, and mineral fertilization of crops in the other cases (pineapple, banana, and sugarcane). N self-sufficiency of SIL1, ME2 and ME3 was higher (between 16.1 and 20.7%), because fewer feed supplements were needed for cattle thanks to grazing (ME3 and ME2), or less mineral fertilization for market garden crops (SIL1). Considering the efficiency of the use of N, logically, the less productive and self-sufficient systems were the least efficient (SIL2, MCI2 and ME4 with an efficiency of 23.2, 20.1 and 13.5%, respectively). Equally, the most productive and self-sufficient systems were the most efficient (SIL1 and ME3 with efficiency of 48.3 and 38.0%, respectively). Intermediate systems in terms of efficiency were due to either low productivity through higher self-sufficiency (ME2), or low productivity through higher self-sufficiency (MCI1 and ME1).

5.3.2. Socio-economic performances Fig. 2 shows economic results according to land and labor productivity, and also the relative importance of subsidies in counterbalancing economic added value. Small labor intensive farms had the highest land productivity, >10,000 D /ha. Due to the small amount of land used for production (limiting factor), and consequently a proportionally higher level of labor/ha, labor productivity was low, under 200 D /work day. Land productivity of SIL1 was twice that of SIL2, which can be explained by the higher added value of the market garden crops cultivated by SIL1 than that of small intensive livestock breeding by SIL2.

The land productivity of medium capital intensive farms (MCI1 and MCI2) was between 7000 D and 10,000 D per hectare while labor productivity was higher, over 400 D per work day. Thanks to considerable fixed capital investment (the right kind of barn, tractors), and high added value crops (pineapple for MCI2, tubers and market garden crops for MCI1), MCI was able to cultivate an intermediate area and produce a good yield without too much labor, land productivity was consequently good and labor productivity particularly high. The land productivity of medium extensive farms (ME1, ME2, ME3 and ME4) was low (between 1300 D and 3800 D /ha), and labor productivity was also low (between 13 D and 228 D per work day), due to the low added value of the main agricultural activities (sugarcane, banana and cattle raising) in these systems. The work load was similar, because the medium amount of land to cultivate, without much mechanization, implies a high work load. The case of ME4 is different, with a very high level of labor (1335 man/day per year), relative to area, due to banana production. Moreover, the added value of banana is low, mainly due to the high cost of intermediaries. Consequently, land and labor productivity were very low (AV/ha of 1306 D and AV/work day of 13 D ). On the other hand, subsidies paid for some products reduced the differences between farms (Fig. 2). SIL farms do not receive subsidies and because they are small, created wealth was sometimes low (less than 20,000 euros for one family worker on the SIL2 farm, due to the importance of livestock activities, which account for 72% of added value). The SIL1 farm created more wealth thanks to the cultivation of high added value crops. In all the other cases, sugarcane, banana or cattle entitle the farm to subsidies, thereby increasing perceived wealth. For instance, the low land and work productivity of ME4 are largely compensated for and total perceived wealth was the same as on medium capital intensive farms.

6. Discussion 6.1. CLI in practice in Guadeloupe As demonstrated by these results, CLI is of real interest, as it leads to complementary livestock feeding and fertilization, e.g. for pig feeding and fertilization of market garden crops and tubers. These practices correspond to farming traditions and to local knowledge in Guadeloupe (Zébus et al., 2004). Small pig systems are very widespread, and are multi-functional, as they produce manure to fertilize the home garden, make use of crop residues and fruit during the production season, and provide meat for traditional events. The traditional creole garden is a complex cropping system combining market garden crops, tubers, fruit, and trees, with organic manure used on small areas. Moreover, these crops are considered to be robust, and respond well to organic fertilization. Previous analytical studies that focused on a specific CLI practice confirmed their interest. For instance, concerning the use of crop-residues as

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

17

Fig. 2. Added value (per hectare and per work day), and amount of subsidies (subsidies per hectare) on the eight mixed farms in Guadeloupe.

animal feed, Archimède et al. (2012) simulated ruminants fed on banana by-products in Guadeloupe; Renaudeau et al. (2014) evaluated the digestive efficiency, growth performance and feeding behavior of pigs fed banana meal. In other tropical contexts, where access to mineral fertilization is limited for smallholders, Tittonell et al. (2007), analyzed nutrient use efficiencies and crop responses to the application of manure on maize and soybean in Zimbabwe; Blanchard et al. (2013) analyzed the efficiency of organic fertilizer in southern Mali. However, even if CLI practices between two specific activities may be frequent and substantial, CLI is still not that important at the level of the whole system in Guadeloupe. When they combine export crops, such as sugarcane and banana, with livestock, farmers are not encouraged to integrate the activities. The system of subsidies based on provisions to buy inputs, such as mineral fertilizers, and standardized technical operations, does not encourage them to use organic fertilization. Using crop residues to feed animals is time consuming. Consequently, when CLI is considered at the level of the whole farming system, its intensity is not very high. Crops are mainly managed using mineral fertilization, whereas livestock systems are managed by purchasing feed concentrate, or through pasture grazing for ruminants. In addition, the agricultural economy is organized in sectors, which does not encourage transversal management practices like CLI. There are also technical limitations to CLI. The high cost of labor, equipment that is not optimal for the size of the farms concerned, and the lack of specialized equipment to enable the use of crop residues and manure, does not encourage the adoption of CLI despite references concerning the nutritive value of a wide range of tropical crop products and by-products, for example in the Feedipedia database (Sauvant et al., 2013). Consequently, even if the most integrated MCLS, (SIL1), appears to be the best in terms of agroecological performances, it is important to recall that its performances are also linked to other attributes. The contribution of CLI to the performance of MCLS is not clear. It is difficult to dissociate the effect of a set of agricultural practices from the overall functioning of a combination of activities. Agroecological performance depends primarily on the nature and the management practices used for crops and livestock combined at system level, as well as socio-economic performances, which

are correlated with farm type, and access to production factors, as Sneessens et al. (2014) already showed for cereal-sheep systems. Comparing MCLS with similar combinations of activities could be one way to analyze the contribution of CLI to the performance of MCLS. Even though MCLS are very diverse in Guadeloupe, certain combinations of activities prevail. This is the case of sugarcanecattle grazing systems, which are diversified to varying degrees, or small labor intensive systems, which appear to be the most integrated. It would be interesting to investigate CLI practices within specific combinations of activities to assess their contribution to improving MCLS as a whole. According to the characteristics of the type of SIL, which appear to be the most integrated systems, CLI could be increased by intensifying current flows. It could concern feeding animals from cropping systems, through better use of available crop residues, or by associating existing productions with the production of forage. Likewise, better storage and processing of manure (Tittonell et al., 2009), combined with crop associations with legumes, could increase the proportion of organic fertilization. According to the characteristics of the type of MCI, which are integrated to a certain extent in terms of diversity of flows but hardly in terms of intensity of flows, CLI could be increased, but the main possibility concerns reducing inputs. It could concern both mineral fertilization of sugar cane and feed concentrate of cattle and pigs. Potential solutions are feeding animals with sugar cane, a crop that occupies an average of 50% of available land, and the valorization of manure produced by intensive livestock systems which is currently not really exploited (Ryschawy et al., 2012). According to the characteristics of the type of ME, which are hardly integrated at all both in terms of intensity and diversity of flows, CLI could be developed by increasing the use of available resources on farm but also by diversifying products, to increase complementarities between productions. Export crops (sugar cane and banana) are hardly integrated and highly dependent on inputs. Potential solutions could be viewed in terms of reallocation of production factors, by reducing the share of these productions and attributing them to forage crops. Like in other types, manure valorization remains scare and organic fertilization remains an open option.

18

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

6.2. CLI in other contexts Concerning the low level of the intensity of CLI at farm scale, the results of this study are consistent with those reported by authors who used ENA to analyze MCLS in East Africa and Madagascar (Alvarez et al., 2013; Rufino et al., 2009b). Similar values were obtained for FCI, ranging from 0 to 0.55% in this study, from 2.5 to 4.4% by Alvarez et al. (2013), and from 0.9 to 11% by Rufino et al. (2009b), which, despite some variations, are low in all cases. Nevertheless, the farms studied in Guadeloupe were less diversified, and less organized than those analyzed by Alvarez et al. (2013) (1AMI/Hr = 0.42 to 0.59) and Rufino et al. (2009a) (1- AMI/Hr = 0.44 to 0.55). According to the schematic characterization of different modes of mixed crop-livestock farming proposed by van Keulen and Schiere (2004), the farms studied in Guadeloupe correspond to cases of high external input agriculture (HEIA), with throughput linear nutrient flows. The studies of Rufino et al. (2009a,b) and Alvarez et al. (2013) concern low external input agriculture (LEIA), with web-like nutrient flows. In Guadeloupe, the intensity of CLI is low because of the high level of use of external inputs (fertilizers and concentrates). Even if recycling exists in mixed systems, it does not weigh much compared to input flow. The high throughput flows, compared to small recycling flows, determine the low organization of flows, quantified by the 1-AMI/Hr. In LEIA, recycling is a little more important, but above all, the organization of the flows is more web-like, as the farms are more diversified, with a lot of small flows between activities. However, comparison with other studies is difficult. The conceptual models differ across studies that use ENA to appraise agrosystems. Alvarez et al. (2013) and Rufino et al. (2009b) analyze situations of subsistence farming, and consequently consider the family as a component of the system. The household is considered as a compartment in the analysis of smallholder food self-sufficiency, which consequently increases throughflows between compartments, which partially explains the higher rate of recycling observed in those LEIA situations. In the present study, farms in Guadeloupe produced with the aim of selling their products on local and export markets so the family was not considered as a compartment. As a result, potential flows of self-consumption of human food, if they exist, are not take into account. In the context of high external input agriculture (HEIA), farm specialization is a general trend, but mixed systems still exist in Australia (Bell and Moore, 2012) or France (Perrot et al., 2012). In those mixed systems in HEIA, the intensity of the CLI must be rather low, as observed in Guadeloupe, and the global performances of the farm, such as efficiency, mainly depend on the partial performance of each activity combined in the mixed system. However, in systems with cultivated grasslands, in rotation with cereal crops for instance, like those encountered in the mixed farming systems in south-western France (Ryschawy et al., 2014), on beef cattle and cereal farms, or in Australia, in cereal-sheep systems (Bell and Moore, 2012), recycling due to grazing and direct restitution of animal excreta short-term flows, in an annual campaign, is also a way to integrate livestock (through forage crops) and crops, within multi-annual cropping systems. In that context, the conceptual model used in this study is no longer appropriate. Such forage crops, even if they are grazed, could not be combined with ruminants in a single compartment. In those mixed systems in HEIA, the intensity and organization of CLI could be higher than those observed in this study.

6.3. ENA implications The use of ENA made it possible to analyze CLI at the level of the whole farming system, independently of crop and livestock

production, which has undoubted advantages when characterizing different CLI practices and MCLS performances. The complex nature of MCLS requires suitable tools to analyze its holistic and systemic properties (Gonzalez-Garcia et al., 2012). By representing the structure and functioning of the farming system in the same dimension, like the N flow network in this study, provides a new view on properties that are not visible through direct observation (Fath et al., 2007). The notion of the organization of flows makes it possible to investigate the connectivity of farming activities, which result in varying degrees of restriction on system development, and, as a consequence, opportunities to improve CLI (Kones et al., 2009; Latham and Scully, 2002). In the same way, the internal circulation rate proposed in this study makes it possible to investigate the contribution of the flow network to the total activity of the system. This in turn, makes it possible to investigate to what extent CLI contributes to system activity and to what extent CLI could be improved, as a function of the flows (feed, fertilization). Each agricultural activity involves different biological processes, resulting in different levels of ecological efficiency (van Ittersum et al., 2013). Consequently, performances and especially efficiency and productivity, depend more on the nature of the activity than on management practices. However, this does not mean that the functioning of the whole system cannot be improved through CLI, by enhancing performances, as a function of a given set of activities. Another important research challenge is the multidimensional assessment of farming systems (Bockstaller et al., 2009; Lichtfouse et al., 2009; Loiseau et al., 2012). Applying ENA to agrosystems using a biotechnical approach makes it possible to assess and compare the agroecological properties of farming systems in the same dimension (Bonaudo et al., 2014). Existing trade-offs between efficiency, productivity, resilience and self-sufficiency should thus be analyzed to rank expected performances according to scale and context (Darnhofer et al., 2010; Dumont et al., 2014; Homann-Kee Tui et al., 2015; Sabatier et al., 2015). Acknowledgments This study was funded by the CIRAD and the FEADER of Guadeloupe (EU, the French Ministry of Agriculture, the Regional Council of Guadeloupe), as part of a the “polyculture-élevage” researcheducation-development project implemented by the Centre INRA Antilles Guyane, the agricultural high school of Guadeloupe (EPLEFPA) and the Chamber of Agriculture. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.eja.2016.06.004. References Agreste, Ministère de l’agriculture, de l’agroalimentaire et de la forêt, 2015. Statistique Agricole Annuelle, Edition 2015, Données En Lignes, from http:// agreste.agriculture.gouv.fr/IMG/pdf/D97115A11.pdf. Altieri, M.A., Funes-Monzote, F.R., Petersen, P., 2012. Agroecologically efficient agricultural systems for smallholder farmers: contributions to food sovereignty. Agron. Sustain. Dev. 32, 1–13, http://dx.doi.org/10.1007/s13593011-0065-6. Altieri, M.A., 2008. Small Farms as a Planetary Ecological Asset: Five Key Reasons Why We Should Support the Revitalisation of Small Farms in the Global South. Third World network (TWN). Alvarez, S., Rufino, M.C., Vayssières, J., Salgado, P., Tittonell, P., Tillard, E., Bocquier, F., 2013. Whole-farm nitrogen cycling and intensification of crop-livestock systems in the highlands of Madagascar: an application of network analysis. Agric. Syst., http://dx.doi.org/10.1016/j.agsy.2013.03.005. Archimède, H., Gourdine, J.L., Fanchone, A., Tournebize, R., Bassien-Capsa, M., Gonzalez-Garcia, E., 2012. Integrating banana and ruminant production in the French West Indies. Trop. Anim. Health Prod. 44, 1289–1296, http://dx.doi.org/ 10.1007/s11250-011-0070-4.

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20 Bell, L.W., Moore, A.D., 2012. Integrated crop–livestock systems in Australian agriculture: trends, drivers and implications. Agric. Syst. 111, 1–12, http://dx. doi.org/10.1016/j.agsy.2012.04.003. Bertin, J., 1977. La graphique et le traitement graphique de l’information. Ed. Flammarion Paris. Blanchard, M., Vayssieres, J., Dugue, P., Vall, E., 2013. Local technical knowledge and efficiency of organic fertilizer production in south Mali: diversity of practices. Agroecol. Sustain. Food Syst. 37, 672–699, http://dx.doi.org/10.1080/ 21683565.2013.775687. Bockstaller, C., Guichard, L., Keichinger, O., Girardin, P., Galan, M.-B., Gaillard, G., 2009. Comparison of methods to assess the sustainability of agricultural systems. A review. Agron. Sustain. Dev. 29, 223–235, http://dx.doi.org/10. 1051/agro:2008058. Bonaudo, T., Bendahan, A.B., Sabatier, R., Ryschawy, J., Bellon, S., Leger, F., Magda, D., Tichit, M., 2014. Agroecological principles for the redesign of integrated crop–livestock systems. Eur. J. Agron., http://dx.doi.org/10.1016/j.eja.2013.09. 010. Bonneviale, J.-R., Jussiau, R., Marshall, É., 1989. Approche globale de l’exploitation agricole: comprendre le fonctionnement de l’exploitation agricole: une méthode pour la formation et le développement. Institut national de recherches pédagogiques. Brossier, J., 1987. Système et système de production. Cah. Sci. Hum. 23, 377–390. Cochet, H., 2012. The systeme agraire concept in francophone peasant studies. Geoforum 43, 128–136, http://dx.doi.org/10.1016/j.geoforum.2011.04.002. Darnhofer, I., Bellon, S., Dedieu, B., Milestad, R., 2010. Adaptiveness to enhance the sustainability of farming systems. A review. Agron. Sustain. Dev. 30, http://dx. doi.org/10.1051/agro/2009053. Devendra, C., Thomas, D., 2002. Crop–animal interactions in mixed farming systems in Asia. Agric. Syst. 71, 27–40. Devienne, S., Wybrecht, B., 2002. Analyser le fonctionnement d’une exploitation. Mémento de l’agronome, 345–372. Dumont, B., Fortun-Lamothe, L., Jouven, M., Thomas, M., Tichit, M., 2012. Prospects from agroecology and industrial ecology for animal production in the 21st century. Animal, 1–16, http://dx.doi.org/10.1017/S1751731112002418. Dumont, B., González-García, E., Thomas, M., Fortun-Lamothe, L., Ducrot, C., Dourmad, J.Y., Tichit, M., 2014. Forty research issues for the redesign of animal production systems in the 21st century. Animal 8, 1382–1393, http://dx.doi. org/10.1017/S1751731114001281. Fath, B.D., Scharler, U.M., Ulanowicz, R.E., Hannon, B., 2007. Ecological network analysis: network construction. Ecol. Modell. 208, 49–55, http://dx.doi.org/10. 1016/j.ecolmodel.2007.04.029. Giller, K.E., Beare, M.H., Lavelle, P., Izac, A.-M.N., Swift, M.J., 1997. Agricultural intensification, soil biodiversity and agroecosystem function. applied soil ecology, soil biodiversity. Agric. Intensif. Agroecosyst. Funct. 6, 3–16, http://dx. doi.org/10.1016/S0929-1393(96)00149-7. Gliessman, S.R., 2004. 2 Agroecology and Agroecosystems. Godinot, O., Leterme, P., Vertes, F., Faverdin, P., Carof, M., 2015. Relative nitrogen efficiency, a new indicator to assess crop livestock farming systems. Agron. Sustain. Dev. 35, 857–868, http://dx.doi.org/10.1007/s13593-015-0281-6. Gonzalez-Garcia, E., Gourdine, J.L., Alexandre, G., Archimede, H., Vaarst, M., 2012. The complex nature of mixed farming systems requires multidimensional actions supported by integrative research and development efforts. Animal 6, 763–777, http://dx.doi.org/10.1017/S1751731111001923. Griffon, M., 2006. Nourrir la planète. Odile Jacob. Harper, C., Jones, N., Marcus, R., 2013. Research for Development: A Practical Guide. Sage. Herrero, M., Thornton, P.K., Notenbaert, A.M., Wood, S., Msangi, S., Freeman, H.A., Bossio, D., Dixon, J., Peters, M., van de Steeg, J., Lynam, J., Rao, P.P., Macmillan, S., Gerard, B., McDermott, J., Sere, C., Rosegrant, M., 2010. Smart investments in sustainable food production: revisiting mixed crop-livestock systems. Science 327, 822–825, http://dx.doi.org/10.1126/science.1183725. Homann-Kee Tui, S., Valbuena, D., Masikati, P., Descheemaeker, K., Nyamangara, J., Claessens, L., Erenstein, O., van Rooyen, A., Nkomboni, D., 2015. Economic trade-offs of biomass use in crop-livestock systems: exploring more sustainable options in semi-arid Zimbabwe. Agric. Syst. Biomass Cereal Cropping Syst.: Lessons Implic. Dev. World 134, 48–60, http://dx.doi.org/10. 1016/j.agsy.2014.06.009. Kones, J.K., Soetaert, K., van Oevelen, D., Owino, J.O., 2009. Are network indices robust indicators of food web functioning? A Monte Carlo approach. Ecol. Modell. 220, 370–382, http://dx.doi.org/10.1016/j.ecolmodel.2008.10.012. Latham II, L.G., Scully, E.P., 2002. Quantifying constraint to assess development in ecological networks. Ecol. Modell. 154, 25–44, http://dx.doi.org/10.1016/ S0304-3800(02)00032-7. Latham, L.G., 2006. Network flow analysis algorithms. Ecol. Modell. 192, 586–600, http://dx.doi.org/10.1016/j.ecolmodel.2005.07.029. Lawrence, P.R., Pearson, R.A., 2002. Use of draught animal power on small mixed farms in Asia. Agric. Syst. 71, 99–110. Lichtfouse, E., Navarrete, M., Debaeke, P., Souchère, V., Alberola, C., Ménassieu, J., 2009. Agronomy for sustainable agriculture. A review. Agron. Sustain. Dev. 29, 1–6, http://dx.doi.org/10.1051/agro:2008054. Loiseau, E., Junqua, G., Roux, P., Bellon-Maurel, V., 2012. Environmental assessment of a territory: an overview of existing tools and methods. J. Environ. Manage. 112, 213–225, http://dx.doi.org/10.1016/j.jenvman.2012.07.024. Mahieu, M., Archimède, H., Fleury, J., Mandonnet, N., Alexandre, G., 2008. Intensive grazing system for small ruminants in the Tropics: The French West Indies experience and perspectives. Small Rum. Res. 77, 195–207.

19

Moraine, M., Duru, M., Nicholas, P., Leterme, P., Therond, O., 2014. Farming system design for innovative crop-livestock integration in Europe. Animal 8, 1204–1217, http://dx.doi.org/10.1017/S1751731114001189. Ozier-Lafontaine, H., Boval, M., Alexandre, G., Chave, M., Grandisson, M., 2011. Vers l’émergence de nouveaux systèmes agricoles durables pour la satisfaction des besoins alimentaires aux Antilles-Guyane. http://www6.inra.fr/ciag/Revue. Perrot, C., Caillaud, D., Chambaut, H., 2012. Economies d’échelle et économie de gamme en production laitière. Analyse technico-économique et environnementale des exploitations de polyculture-élevage. Rencontres Rech. Rumin. 19, 33–36. Peyraud, J.-L., Cellier, P., Dupraz, P., 2012. Les flux d’azote liés aux élevages, réduire les pertes, rétablir les équilibres. Renaudeau, D., Brochain, J., Giorgi, M., Bocage, B., Hery, M., Crantor, E., Marie-Magdeleine, C., Archimède, H., 2014. Banana meal for feeding pigs: digestive utilization, growth performance and feeding behavior. Animal 8, 565–571. Rufino, M.C., Hengsdijk, H., Verhagen, A., 2009a. Analysing integration and diversity in agro-ecosystems by using indicators of network analysis. Nutr. Cycl. Agroecosyst. 84, 229–247, http://dx.doi.org/10.1007/s10705-008-9239-2. Rufino, M.C., Tittonell, P., Reidsma, P., Lopez-Ridaura, S., Hengsdijk, H., Giller, K.E., Verhagen, A., 2009b. Network analysis of N flows and food self-sufficiency-a comparative study of crop-livestock systems of the highlands of East and southern Africa. Nutr. Cycl. Agroecosyst. 85, 169–186, http://dx.doi.org/10. 1007/s10705-009-9256-9. Ryschawy, J., Choisis, N., Choisis, J.P., Joannon, A., Gibon, A., 2012. Mixed crop-livestock systems: an economic and environmental-friendly way of farming? Animal 6, 1722–1730, http://dx.doi.org/10.1017/ S1751731112000675. Ryschawy, J., Joannaon, A., Choisis, J.P., Gibon, A., Le Gal, P.Y., 2014. Participative assessment of innovative technical scenarios for enhancing sustainability of French mixed crop-liverstock farms. Agric. Syst. 129, 1–8. Sabatier, R., Oates, L.G., Brink, G.E., Bleier, J., Jackson, R.D., 2015. Grazing in an uncertain environment: modeling the trade-off between production and robustness. Agron. J. 107, 257–264. Sauvant, D., Tran, G., Heuzé, V., Bastianelli, D., Archimède, H., 2013. Feedipedia: une encyclopédie mondiale des ressources alimentaires pour les animaux d’élevage, in: 5. Journées d’Animation Scientifique Du Département Phase (JAS Phase 2013). Paris, France. Schiere, J.B., Ibrahim, M.N.M., van Keulen, H., 2002. The role of livestock for sustainability in mixed farming: criteria and scenario studies under varying resource allocation. Agric. Ecosyst. Environ. 90, 139–153, http://dx.doi.org/10. 1016/S0167-8809(01)00176-1. Sierra, J., Desfontaines, L., Faverial, J., Loranger-Merciris, G., Boval, M., 2013. Composting and vermicomposting of cattle manure and green wastes under tropical conditions: carbon and nutrient balances and end-product quality. Soil Res. 51, 142–151. Sierra, J., Causeret, F., Diman, J.L., Publicol, M., Desfontaines, L., Cavalier, A., Chopin, P., 2015. Observed and predicted changes in soil carbon stocks under export and diversified agriculture in the Caribbean: the case study of Guadeloupe. Agrc. Ecosyst. Environ. 213, 252–264, http://dx.doi.org/10.1016/j.agee.2015.08. 015. Sneessens, I., Benoit, M., Brunschwig, G., 2014. Un cadre d’analyse pour évaluer les gains d’efficience permis par les interactions culture-élevage: une typologie des systèmes de polyculture-élevage couplée à une quantification de l’intégration. Innov. Agronom. 39, 125–137. Stark F., Alexandre R., Diman J.L., Alexandre G., 2010. A participatory approach in agricultural development: A case study of a Research-Education-Development project to optimize mixed farming systems in Guadeloupe (FWI), Congrès SAPT, Gosier, Guadeloupe, November 2010. Stark, F., Diman, J.L., Fanchone, A., Alexandre, R., Alexandre, G., Archimède, H., 2012. Characterization of mixed farming systems and crop-livestock integration in Guadeloupe (French West Indies). In: II International Symposium on Integrated Crop-Livestock Systems, Porto Allegre, Brazil, November 2012. Steinfeld, H., Gerber, P., Wassenaar, T., Castel, V., Rosales, M., De Haan, C., 2006. Livestock’s Long Shadow. FAO Rome. Sumberg, J., 2003. Toward a dis-aggregated view of crop–livestock integration in Western Africa. Land Use Policy 20, 253–264, http://dx.doi.org/10.1016/S02648377(03)00021-8. Thornton, P.K., Herrero, M., 2001. Integrated crop-livestock simulation models for scenario analysis and impact assessment. Agric. Syst. 70, 581–602. Tittonell, P., Zingore, S., van Wijk, M.T., Corbeels, M., Giller, K.E., 2007. Nutrient use efficiencies and crop responses to N, P and manure applications in Zimbabwean soils: exploring management strategies across soil fertility gradients. Field Crops Res. 100, 348–368, http://dx.doi.org/10.1016/j.fcr.2006.09.003. Tittonell, P., Rufino, M.C., Janssen, B.H., Giller, K.E., 2009. Carbon and nutrient losses during manure storage under traditional and improved practices in smallholder crop-livestock systems—evidence from Kenya. Plant Soil 328, 253–269, http://dx.doi.org/10.1007/s11104-009-0107-x. Udo, H.M.J., Aklilu, H.A., Phong, L.T., Bosma, R.H., Budisatria, I.G.S., Patil, B.R., Samdup, T., Bebe, B.O., 2011. Impact of intensification of different types of livestock production in smallholder crop-livestock systems. Livest. Sci. 139, 22–29, http://dx.doi.org/10.1016/j.livsci.2011.03.020. Ulanowicz, R.E., Goerner, S.J., Lietaer, B., Gomez, R., 2009. Quantifying sustainability: resilience, efficiency and the return of information theory. Ecol. Complex. 6, 27–36, http://dx.doi.org/10.1016/j.ecocom.2008.10.005.

20

F. Stark et al. / Europ. J. Agronomy 80 (2016) 9–20

Ulanowicz, R.E., 2004. Quantitative methods for ecological network analysis. Comput. Biol. Chem. 28, 321–339, http://dx.doi.org/10.1016/j.compbiolchem. 2004.09.001. van Ittersum, M.K., Cassman, K.G., Grassini, P., Wolf, J., Tittonell, P., Hochman, Z., 2013. Yield gap analysis with local to global relevance—a review. Field Crop Res. 143, 4–17, http://dx.doi.org/10.1016/j.fcr.2012.09.009. van Keulen, H., Schiere, H., 2004. Crop-livestock systems: old wine in new bottles. In: New Directions for a Diverse Planet. Proceedings of the IV International Crop Science Congress, Australia.

Xandé, X., Despois, E., Reneaudeau, D., Gourdine, J.L., Archimède, H., 2007. The influence of sugar cane diet on growth performances and carcass traits in Creole pigs. J. Rech. Porcine 39, 231–238 http://www.journees-rechercheporcine.com/texte/2007/qual/q04.pdf. Zébus, M.-F., Alexandre, G., Diman, J.-L., Despois, É., Xandé, A., 2004. Activités informelles, normalisation et développement L’élevage porcin en Guadeloupe. Cah. Agric. 13, 263–270.