Preventive Veterinary Medicine 39 (1999) 111±128

From `ecopathology' to `agroecosystem health'$

B. Fayea,*, D. Waltner-Toewsb, J. McDermotta,b
a
CIRAD-EMVT, BP 5035, 34032 Montpellier, Cedex 1, France
b
Department of Population Medicine, University of Guelph, Guelph, Ont., Canada, N1G 2W1

Accepted 6 December 1997

Abstract

The `epidemiologic revolution' of the 1960s arose in response to the inability of reductionist
methods to provide practical solutions to the complex problems of health and production in
livestock systems. In a farm, there are not only interactions between animal factors and herd
husbandry factors such as feeding, housing, and microbiological environment, but also with a
number of other `non-animal' factors. For this reason, a `global' or `holistic' approach, aimed at
explaining animal health status within the overall dynamic of a livestock production system, was
developed in France under the title of `ecopathology'. In ecopathology, the discipline of
epidemiology is integrated into a systemic approach, including: the development of a preliminary
conceptual model, sampling based on the structure of the livestock production system, the
establishment of a field study by a multidisciplinary team, the organization and management of the
animal health and production information, data analysis, the distribution of results to all participants
and the development of a preventive medicine programme.
The farm is also influenced by the social, economic and environmental setting to which it
belongs. To account for this, a change of scale is necessary. The three elements of the livestock
production system considered in ecopathology (farmer, herd and resources), at the level of the
agroecosystem become a human community (farmers, consumers, decision-makers), an animal
population, and the complex of human, social and economic conditions within the system. The
concept of agroecosystem health is closely linked to the overall principle of improving the
sustainability of the system. This and other measures of the health status of an agroecosystem can
be assessed with methods developed by epidemiologists and other disciplines within a system's
perspective. In this systems view, ecopathology provides a basis for assessing herd health whereas
agroecosystem health develops the broader context into which ecopathology contributes. # 1999
Elsevier Science B.V. All rights reserved.

Keywords: Ecopathology; Agroecosystem health; Epidemiology; Agricultural systems; Sustainability

* Corresponding author. Tel.: +33-73-62-42-63; fax: +33-73-62-45-48.

$
Presented originally as a Plenary paper at the VIII International Symposium on Veterinary Epidemiology
and Economics, Paris, France, July 8±11, 1997.

0167-5877/99/$ ± see front matter # 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 7 - 5 8 7 7 ( 9 8 ) 0 0 1 4 9 - 4
112 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128

1. The epidemiological revolution

In an overview paper published in the first issue of Preventive Veterinary Medicine,

Schwabe (1982) presents a perspective on the developments of epidemiology and
economics in veterinary practice in the latter half of this century. In Schwabe's view, four
crises became apparent in the 1950s: (i) the persistence of health problems in some herds
even after many named infectious diseases had been substantially controlled; (ii) the
increasing demands of governments to estimate the economic costs and benefits of animal
health; (iii) the absence of appropriate research methods to understand and control
etiologically complex diseases affecting production; and (iv) the inability of private
veterinarians and producers to develop programs to control health and production
constraints associated with intensive farming practices. Because the `germ-theory'
approach failed in these contexts, there was an impetus for an ``holistic and
epidemiologic approach of the causality of the diseases'', fostering the initiation and
progression of what Schwabe called an `epidemiological revolution' in the 1960s to better
identify, quantify and analyze the multiple and interacting causes of many animal health
and production problems.

2. System theory

These changes mirrored a similar revolution among agronomists who were developing
the concepts of farming systems theory (Von Bertalanffy, 1970). Faced with numerous
failures of agricultural development projects in both industrialized and developing
countries, many agronomists questioned the efficacy of their traditional reductionist
research methods (SeÂbillotte, 1978) and searched for more holistic paradigms. Central to
farming systems theory was the socio-economic principal that ``farmers' practices
respond to certain aims and constraints, and the ignorance of those is the principal
source of the failure of technical solutions developed from research'' (Tourte, 1965). This
expansion of the traditional technical domains of animal health and production practice to
include consideration of the economic, social, and technical objectives of farmers has
been applied in practice by many veterinary epidemiologists (e.g. Bigras-Poulin et al.,
1985), who have wisely noted that without this integration, the objectives of the
epidemiological revolution would remain unfulfilled.
Formally, farming systems research considers the interactions between three elements:
the farmer, the herd and available resources. The farmer makes decisions in accord with
both technical and/or behavioral criteria. These decisions are made operational through
farming practices which by direct or indirect pathways and feed-back loops influence
both the performance of the herd and the quantity and quality of animal products.
Farmers also influence and are influenced by the resources available in their particular
farming system which of course greatly influence herd productivity. The performance of
the farming systems have been assessed by various measurable indicators which can be
used in whole-farm decision-making.
Landais (1994) described a conceptual model of a livestock production system from his
perspective as an animal scientist (see Fig. 1), but its lack of explicit accounting for
 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128 113

Fig. 1. Working diagram for the farming system adapted from Landais (1994).

animal health as a component of the system is not acceptable to many veterinary

epidemiologists. In fact, the basis of the epidemiologic approach to studying animal
health and production problems developed in France, called ecopathology (Tuffery, 1977;
Barnouin and Brochard, 1981; Tillon, 1982), aims precisely to explain health within the
farming system.

3. Ecopathology ± epidemiologic assessment of livestock systems

Just as farmers make decisions based on technical criteria within the farming systems
paradigm, ecopathologists seek to define herd health `referentials' as criteria for
supporting preventive medicine and herd management decisions to improve health
performance. Both animal-level (e.g. relative risk of diseases related to physiological
factors) or herd-level (e.g. the relative incidence of animal diseases in comparison to
other farms in the same area) referentials can be used depending on the type of decision
to be made. These decisions will also depend, as in the farming systems paradigm, on the
farmer's behavioral attitudes with respect to disease occurrence and risk and they will be
implemented through various practices (e.g. management, culling, feeding practices,
vaccination) which, separately or in combination, will influence the risk of `disease'. In
114 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128

ecopathology, disease or health performance is interpreted as just another output of the

farming system, like production performance, which can vary depending on environ-
mental conditions (animal housing, climatic factors) and available resources (feeding,
inputs). For routine decision-making, it can be beneficial to define easily measurable and
important health indicators to reflect both health performance (e.g. somatic cell count as
a subclinical mastitis indicator) and risk status (e.g. cleanliness score as a hygiene
indicator). However, as in any complex system, the development of useful indicators can
be both very rewarding as well as difficult. In some circumstances, health indicators have
been particularly useful in identifying, at an early stage, imbalances and malfunctions in
livestock production systems (Barnouin et al., 1994).
Ecopathologists integrate epidemiologic principles into a systematic method for
studying health status within farming systems using an ecopathological survey. Such a
survey consists of several distinct but complementary stages:

 The sampling of farms by type of farming system. Since we consider the disease as an
output of the system controlled by inputs and farmer decisions, the farming system is
the operational basis for future actions.
 The use of rigorous observational study methods by multidisciplinary teams
(veterinarians, epidemiologists, zootechnicians, farm technicians, professional organi-
zations, statisticians, computer scientists and others according to the needs of the
survey). This allows for the collection of a diversity of parameters, enlarging the frame
of epidemiologic information with concepts and methods from non-medical disciplines
(animal husbandry, economics, sociology). In terms of herd health delivery, this also
allows veterinary inputs to be better integrated into herd management (feeding,
reproduction, and production). Thus, the ecopathological approach is both multi-
disciplinary and multiprofessional.
 The management and organization of an information system in a specific data base.
The collection of data in this context is characterised by a high degree of complexity
and linkages and with a great diversity of studied factors. This requires that careful
attention be paid to developing preliminary conceptual models based on prior
biological knowledge and subsequently revised by analysis of the assembled database.
This approach has been advocated for the analysis of complex biological systems with
a priori knowledge gaps (Mallows and Tukey, 1982). From this, empirical models are
developed using prior concepts refined in an iterative approach. Thus, unlike
experiments, data processing and analysis cannot be precisely defined at the beginning
of the study, since the analysis at any stage will be guided by the results of the previous
step. As a consequence, a fixed framework for data processing will be impossible to
construct and thus the data organization needs to be sufficiently flexible to support
unexpected applications (Lescourret et al., 1992).
 A staged approach to data analysis. This will emphasize exploratory (especially
graphical methods), followed by methods to develop `structured' disease models
reflecting both previous theory and exploratory results and finally modelling to
estimate the relative importance of different risk factors and health indices within the
`structure' developed. However, these analyses may be complicated by a number of
problems including multicollinearity, confounding and interaction (Dohoo et al., 1996).
 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128 115

The general strategy for dealing with multicollinear data revolve around reducing and/
or re-categorizing the usually large number of independent variables prior to
investigating associations with the disease. This screening of associations between
independent and dependant variables can be done by various multivariable techniques,
such as correlation analysis, linear or logistic regression, correspondence analysis and
others.
 The dissemination of results to all stakeholders and key players in the system. A
comprehensive programme to transfer the results of the survey includes: (i)
information transfer through brochures, technical documents, newspaper and magazine
articles, and workshop papers; (ii) skills transfer through training, development of
intervention guides and prevention manuals; (iii) raising awareness through discussion
groups, meetings and other publicity; and (iv) follow-up actions for assessing opinion,
refining materials and evaluating and measuring the impact of the proposed
modifications (Rosner, 1993).

The main objective of ecopathology is very practical, to identify the risk factors and
their interactions which most influence health performance in specific priority farming
systems. This approach has been used in tropical countries (Faye et al., 1994a) but the
main studies have been conducted in France (Barnouin, 1992; Faye, 1995; Faye and
Barnouin, 1996). The ecopathological study conducted in Brittany, France is a good
example of the latter. The study was conducted over four years (1986±1990) among
intensive dairy farms. This particular farming system was chosen since these farms were
supposed to represent the trend of French dairy farming towards increased farm size and
intensification (Faye and Barnouin, 1987). The study farms consisted almost exclusively
of Holstein±Friesian cattle with mean annual milk yields of at least 6500 kg, practiced
winter feeding based on silage, and had cubicles or free-stalls as the main animal housing.
Data were collected by a working group using a protocol developed by a multi-
disciplinary team. Data were collected at the cow (e.g. clinical diseases, production and
reproductive performance, cleanliness and body score, individual feed supplementation,
behaviour, and test results (e.g. milk bacteriology and blood metabolic profile)) and farm
(e.g. farming practices and conditions, baseline diet, eco-climatic parameters, inputs
assessment) levels.
A total of 4129 dairy cows (Holstein breed) contributing 8945 lactations and belonging
to 47 intensive dairy farms were involved. Farmers and veterinarians volunteered to
participate in the study. The participating farmers were purposively sampled as being
representative of the main dairy farming system in Brittany and all were members of the
Dairy Herd Improvement Association (DHIA). They were selected for their ability to
detect and record diseases efficiently, as judged by their veterinarians and confirmed
through a pre-study period. The State Veterinary Services from three French departments
(FinisteÁre, Ille-et-Vilaine and Morbihan) and the Veterinary Technical Association
participated in this survey. Pre-study training and discussions about data collection and
study design methods were used to help standardize the survey process among the
collaborators.
Veterinarians and farmers were given a glossary that included definitions of the main
clinical diseases of the dairy cow. Every six weeks, nine specialized surveyors from the
116 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128

Fig. 2. Integration of the conceptual data models (thin framed boxes) with specific submodels (thick framed
boxes) into the global model of an epidemiological survey on dairy cows (adapted from Lescourret et al., 1993).

State Veterinary Services visited the farms to collect observational and management data,
to measure individual body condition and cleanliness scores for cows around calving, and
to collect milk samples for bacteriological analysis. Every month, individual milk
samples were collected from all lactating cows by DHIA technicians to estimate
parameters for milk yield and individual SCC. These data were accessible from DHIA
computer records (Fig. 2).
A database named Gwen-Ha-Du (`black-and-white') was developed. This database
included five basic data structures (Lescourret et al., 1993). The first comprised
identification and unique life histories for each cow, including identification, genealogy,
and culling. The second and third contained data collected throughout the cow's life,
including individual morbidity events and production and reproductive performance
measures (i.e. milk production, milk composition, bacteriological quality of milk, drying
off, calving and services). The last two data structures contained herd-level data on herd
management and feeding practices.
Results concerning all main elements of this dairy farming system were disseminated
and published. At the farmer level, a typology of farmer behaviour was developed using
the temperament of their animals as an indicator (Faye, 1996). At the animal level, the
factors affecting the distribution of clinical mastitis in space and time (Lancelot et al.,
1997), the importance of intramammary infections (Faye et al., 1994b), placental
retention (Chassagne et al., 1996), mortality (Faye and PeÂrochon, 1995), and the
association between diseases and culling (Beaudeau et al., 1994a, b) were studied. At the
herd environment and resources level, the dietary factors associated with somatic cell
count (SCC) (Barnouin et al., 1995) and the housing factors associated with udder health
(Faye et al., 1994c) were the main research components.
 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128 117

After reflection and discussion of the results, it was considered that periparturient
diseases were the most important health factor. Thus, health management programmes
targeted at changes in feeding practices, which most influenced periparturient disease
incidence, were developed and extended (Barnouin et al., 1994).
In ecopathology, the diversity of disciplines involved both strengthens the contribution
of epidemiologists (Sabatier et al., 1994) and enlarges the frame of the epidemiological
survey (Calavas et al., 1996). This is done both by including non-medical disciplines (e.g.
animal production, economics, sociology) and by mobilizing the practical knowledge of
extensionists and farmers in management, reproduction, and feeding. These experiences
in expanding the scope of studies and mobilizing stakeholders can provide a useful
starting point for evaluating livestock health and production in even broader contexts.

4. From farming systems to agroecosystems

Agroecosystem health is a professional application of systems approaches which have

been successful at animal, herd and farm level. Just as we can talk meaningfully about
animal, herd, farm, community and public health, it is possible to talk about the health of
an agroecosystem.
Although ecopathology enlarges the scope of veterinary investigations to consider the
complexities of farming systems, agroecosystem health both enlarges the window of
observation and changes its focus. Like ecopathology, agroecosystem health is an
application of systems theories. Thus, research and its applications are part of the same
iterative process, where management interventions become part of the research process
itself. Unlike ecopathology, the unit of interest is the entire complex socio-ecological
system within which agricultural activities take place. Like both herd health and
ecopathology, and unlike epidemiology, the ultimate aim of this process is more synthetic
than analytic.
It is useful to think about the science and practice of agroecosystem health in terms of
the processes of reasoning which are applied in herd health management programs. One
might think of agroecosystem health management as comprising at least five (not
necessarily linear) steps: (1) a description of the agroecosystem in systemic terms; (2)
identification of decision-makers and/or stakeholders; (3) establishment of goals, that is,
perceived attributes of a healthy system, operational objectives related to those goals at
various scales and time horizons, and determination of measurable indicators which will
give information about whether or not those objectives are being achieved; (4)
identification and implementation of desirable and feasible changes, which often involves
resolving conflicts among goals set by different decision-makers at different levels; and
(5) monitoring of the selected indicators, and adapting to changing circumstances or
unexpected or undesirable outcomes. This may mean revisiting the description of the
system, the objectives selected, and the management options chosen.

4.1. Describing the system

An agroecosystem can be defined as ``a socio-ecological system managed primarily for

the purpose of producing food, fiber and other agricultural products comprising
118 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128

domesticated plants and animals, biotic and abiotic elements of the underlying soils,
drainage networks, and natural vegetation and wildlife'' (Gallopin, 1995; Waltner-
Toews, 1996). Such systems are self-organizing, that is, through internal autocatalytic
feedback loops, they re-create themselves in particular forms. Furthermore, agroecosys-
tems exist in nested hierarchies from farms and their sub-units up to regional and global
levels. Such hierarchies, in which larger units are comprised of, but with characteristics
which are greater than, the units within them, have been called holarchies, with each unit
being referred to as a holon (Checkland, 1981). These holons are both bounded by, and
necessarily interdependent with, other holons. Farms, subwatersheds and regions are
inextricably linked in a socio-ecological web (Giampietro, 1994).
Since considerable work has already been done in studying and assessing farming
systems, most recent work on agroecosystem health has focussed on scales larger than the
farm (Waltner-Toews, 1996). For example, at the sub-watershed level, the three
constitutive elements of the livestock farming system from an ecopathology view (farmer/
herd/resources) become: (i) a human community including different actors of livestock
activity (farmers and their organizations, political, economic and environmental
managers), (ii) an animal population (domestic herds, wild animals) and (iii) the
biophysical environment (water, nutrients, energy).
Land-time cross budgets have been proposed to link operational objectives of farmers
(for instance, minimization of risk, maximization of income) and activities to achieve
those objectives (such as rotations, livestock systems) with both ecological and economic
variables (Giampietro and Pastore, 1997). This also links the internal workings of farm
activities to the larger energy and matter flows across watersheds.
Another way to understand agroecosystems at a particular level is to create attractor
models in several dimensions. This is based on the observation that many socio-
ecological systems fall into a few patterns of behaviour and resist pressures for change. In
some cases, these appear to represent what Gilberto Gallopin (1997) has called `perverse
resilience', leaving communities in states of economic stagnation and environmental
degradation despite our best efforts to reverse their decline. In other cases,
agroecosystems appear to resist such degradation. Often, such complex systems will
experience sudden, discontinuous changes which can be described in terms of the
catastrophe theory first developed by Rene Thom (see Kay, 1991). If conceptual models
of such attractors and potential `flips' can be created, we can begin to identify their
determinants. These are similar to what have been called determinants of health in the
population health literature.

4.2. Identification of responsible decision-makers and stakeholders

In herd health programs and ecopathology, management and governance are integrated.
Once one moves beyond the farm gate however, not only are conflicting goals and
perspectives associated with different people, but governance functions and management
functions are separated. This is further complicated by the occurrence of multiple
perspectives across multiple scales.
Institutional and stakeholder analysis can be carried out to determine who should be at
the table, and how to involve them while recognizing political and economic power
 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128 119

differentials. In some cases, new organizations can be created to bridge economic,

environmental, public health or political jurisdictions. The International Joint Commis-
sion for the Great Lakes is one such example, but there are many other economy-
environment round tables and non-governmental multi-stakeholder groups that have
emerged throughout the world in the past few years, at levels ranging from local
communities to the biosphere, in what sociologist George Francis (1996) has called
`virtual governance'.
In order to better understand complex problems, some researchers have found
influence diagrams, or signed digraphs (Caley and Sawada, 1995) to be helpful.
These identify feedback loops which reinforce or weaken particular system states, and
clarify who has the power to influence those connections. In these diagrams, which
some have termed `spaghetti diagrams', various components of the system are connected
by arrows which are labelled as positive if they co-vary in the same direction, or
negative if they co-vary in opposite directions. For instance, a line between income and
meat-eating might be positive, while that connecting water quality with human disease
would be negative. By examining these diagrams one can judge whether particular
relationships are likely to be amplified or dampened when changes are made; when time-
lags are considered, one can predict whether to expect stabilization or oscillation of
patterns. In the context of such diagrams one can begin to explore why, for instance, the
education and empowerment of women might have a greater long-term effect on the
health of agroecosystems (both directly through increasing available knowledge to the
system and indirectly by decreasing demands on the system through self-directed
population control) than technological interventions which may appear to be more
attractive in the short run.
Even as one branch of agroecosystem health management is based on a scientific
description of the system, this cannot proceed without reference to the other branch, the
articulation of what kind of system is considered to be desirable by decision-makers and
stakeholders (Boyle et al., 1996). Waltner-Toews and Nielsen (1995) proposed a
classification strategy for agroecosystem health concerns, which includes hierarchical
(holarchical) level and dimension (economic, social, biophysical) as two classification
axes of a three-dimensional matrix, with the third axis used to describe the desired
attributes (e.g. integrity and effectiveness) of health (Fig. 3).

4.3. Setting goals and operational objectives, and selecting indicators

Having set out a holarchy and several dimensions, one must derive some criteria by
which to judge health. Most definitions of health combine a notion of a harmonious
balance (which might be termed integrity, measured in terms of the current functioning of
the system) and the notion of capacity to achieve some goals (sometimes measured in
terms of various forms of social, economic or natural capital). In complex system terms,
one would say that the goal of health management is to foster the ability of the system to
self-organize and evolve over time to adapt to environmental change (Kay and Schneider,
1994). More specifically, one might ask: are the quality and quantity of internal and
external resources sufficient, and is their organization appropriate, for the agroecosystem
to meet its goals (Waltner-Toews and Wall, 1997)?
120 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128

Fig. 3. A classification framework for agroecosystem health initiatives (adapted from Waltner-Toews and
Nielsen, 1995).

In order for holons to self-organize and adapt, the boundaries of each holon should
remain intact while at the same time fostering interchange among them. Thus, it is
important for farmers' livelihoods to be able to trade, but not to become absorbed into
larger and larger farms; similarly, rural communities cannot remain isolated; however, to
give global free traders an absolute carte blanche to remove materials and energy from a
local community is tantamount to peeling the skin off a live cow and letting its life
exchange freely with the environment.
Since agroecosystems are human re-arrangements of ecological systems, these goals
are fundamentally human ones reflecting a desire to enhance the quality of life for
ourselves and our descendants. Even the desire to maintain biodiversity can be seen as a
way of protecting future options for human community development. In general, one
moves from articulating general goals, to setting operational objectives, to selecting
measurable indicators which can be used to assess progress. Furthermore, just as
variables of interest to those studying the health of farming systems incorporate measures
of longevity, disease, productivity and capacity to respond in relation to goals, so
indicators of agroecosystem health must reflect social, economic and biophysical
dimensions of the system at various levels (Fig. 3).
What emerges from this combination of holarchic system description and goal-setting
by decision-makers is a process which incorporates the participation of farmers,
communities and regional governing organizations in defining and monitoring the health
status of the system of which they are a part. Because the capacity to adapt to
unpredictable future changes is an essential part of health, and local empowerment is an
element in this capacity, selection of indicators a priori by outside experts is counter-
productive. The operational objectives which reflect agroecosystem goals, and indicators
which measure their achievement, can only be identified through a process of negotiation
among those who live in and/or have a stake in the health of those agroecosystems. The
role of researchers and agroecosystem health practitioners in this context shifts from the
 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128 121

conventional view of experts giving advice, to that of facilitators helping to explore the
possible consequences of alternative choices.

4.4. Implementing desirable and feasible changes

In herd health, ecopathology and agroecosystem health, every management interven-

tion must be viewed as a kind of experiment. Furthermore, while epidemiological
techniques inherited from human medicine are designed to isolate the effects of
individual components, systemic approaches are designed to assess the success of the
overall, interacting system.
In ecopathology, the passage from an object `herd' to the object `farm' has
required a conceptual change to integrate the multiple purposes of farming and the
multiples objectives of the farmer (Bawden, 1995; Faye, 1996). The further general-
izations made in agroecosystem health framework better allow the assessment of the
interrelationships between farming systems and their environment which were recognized
but largely ignored in farming systems studies. First, the complexity of relationships
between farms and other components of the agroecosystem (related in part to spatial and
temporal feedback loops) is added by enlarging the context of study. In a full
agroecosystem situation, it has become clear that there are not only conflicts
between different dimensions at any given scale (e.g. trade-offs between the richness
of social life with many small farms and the increased income associated with economies
of scale) but also across scales. Thus, the goal of a farmer to retain as much income as
possible is in conflict with the national goal of paying off foreign debts; a country may
want to increase milk production for food security reasons even as the farmer decides to
reduce production for personal reasons. Furthermore, whereas a farm family may
negotiate goals and implement better management practices for a farm, it is not always
clear who should, or who can, do this at larger scales. Niels Roling of Wageningen
Agricultural University has identified a meaningful integration of the multiple
perspectives of stakeholders as the central problem facing all agroecosystems approaches
(Roling, 1996).
In view of the fact that hypotheses about agroecosystem health relate to the system as a
whole, they are necessarily complex and multi-faceted. Patterns of self-organization to
which the system is drawn are strongly determined by system goals and their
implementation. For example, globalized agricultural markets, facilitated through unified
road-building and transportation networks, apply increasingly steep external gradients to
local agroecosystems, sometimes with undesired and even tragic consequences. In
seeking to maximize achievement of a single goal, such as productivity or economic
efficiency, many unstated but highly cherished goals like equity, democracy, public health
and sustainability may be critically undermined. The occurrence of Bovine Spongiform
Encephalopathy or vero-toxin producing E. coli in beef and Salmonella in poultry is a
natural consequence of systemic changes based on the single-goal attractor of increasing
supplies of cheap meat. In fact, many emerging diseases in humans and animals are a
consequence of policy decisions based on achieving single (albeit desirable) goals.
It is at this stage of deciding on appropriate management programs that it becomes
necessary to facilitate negotiations between various decision-makers to accommodate
122 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128

goals that may be conflicting. The creation of platforms for negotiation or other fora is
thus an essential part of agroecosystem health management.

4.5. Monitoring and adaptation

The monitoring of indicators, reflecting goals and objectives set by decision-makers

and stakeholders through a process of negotiation informed by science, is the most
feasible way to test agroecosystem health hypotheses. Management, in this context is
research; management programs should be considered as `natural' experiments which
should be designed and monitored as rigorously as laboratory experiments but with
important differences. Agroecosystem health management demands that decision-makers
become part of the research process, and that new programs at various scales be formally
implemented in a way that allows for monitoring and adjustment. By incorporating
human communities into the research process, by negotiating trade-offs between multiple
goals, and by making policy and management changes part of the research process,
agroecosystem health management is simultaneously research and management. Since

Fig. 4. A research process for agroecosystem health (adapted from Rowley et al., 1997).
 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128 123

the stakeholders are intimately involved in this process, one is not left, in the end, trying
to sell experimental results to recalcitrant farmers or policy-makers.
Agroecosystem health management, being both participatory and structured in a
looping fashion which mimics the system itself (Fig. 4), does not result in a single
definable outcome. In this, agroecosystem health management is quite unlike a simple
experimental approach which asks a simple question under controlled conditions and
receives a simple answer, or industrial management, which may focus on a simple
outcome such as producing more pigs more cheaply. This may be frustrating to those of
us used to well-defined experiments and business managers accustomed to the stable
world of the 1960s and 1970s. Many executives of multi-enterprise corporations,
however, have already recognized the significance of similar approaches in devising
adaptive strategies to a rapidly changing world (Checkland, 1981). It is none too soon for
public institutions to follow suit. The outcome of this approach is a sustainable web of
learning organizations, from farms to communities to global institutions, which are
capable of acting, monitoring and adapting in a world characterized by perpetual change.

5. Epidemiology and agroecosystem health

Although ecopathology and agroecosystem health seem to have a natural comple-

mentarity, it is not entirely clear how these ideas relate to what we have come to think of
as epidemiology. One could make an argument that agroecosystem health merely
introduces epidemiologists to a new, and larger, unit of concern ± agroecosystems as large
cows, as it were.
For those who are working in herd health programs, the shift to ecopathology and then
to agroecosystem health seems to represent merely a broadening of vision and an increase
in complexity. A complex system such as an agroecosystem may be legitimately
described in different ways (e.g. economically, by energy flows, by the health of the rural
community health, by the happiness of the agricultural workers) which may be rooted in
different paradigms and result in different (sometimes contradictory) goal-priorities.
While a farmer may balance income losses from disease against preventive program
costs, in agroecosystems, we may find ourselves balancing drinkable water against cheap
food, or sustainable rural communities against high commodity prices or overproduction.
The problem is more intractable than this, however. By thinking of reality in terms of a
holarchy, and by putting people inside the system, we simultaneously reduce the sample
size, and the reference population, and blur the separation between conventional science
and management. One role for epidemiology makes use of conventional tools to select,
compare and study the dynamics among smaller holons (animals, herds, farms). This is
certainly a necessary component of systems approaches.
If we consider that epidemiology is the study of the determinants of health and disease
in populations, we might also say, however, that ecopathology and agroecosystem health
represent simply different ways of understanding these determinants, which are not based
on statistical, quantitative methods. In this role, conventional analytical tools, while
useful for studying short-time linear components of the system, are of little use in
studying long-term, multi-level feed-back loops with multiple outcomes. Some
124 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128

researchers in this field have stated flatly that in this arena there is no global truth, only
local truth; that is, every solution is unique, and the only thing that can be generalized is
the process of assessment and adaptation (Kay and Schneider, 1994). Unlike
ecopathology, agroecosystem health has a very short history, and it is still not entirely
clear what is context-specific and what can be generalized.
Currently, researchers from the University of Guelph are involved in agroecosystem
studies in Canada (The Agroecosystem Health Project, University of Guelph), Peru
(Rowley et al., 1997), Honduras and Kenya, with plans on the table for Ethiopia and
Nepal. Over the next several years, these studies should identify many of the problems,
and, hopefully, many of the solutions, necessary to apply agroecosystem health concepts.

6. Final words

Ecopathology may be considered as an enrichment of epidemiology by the concepts of

systemic ecology. In that ecologic context, the herd may be considered as an anthropo-
biocenose, i.e. a specific community where micro-organisms (which could be pathogens)
and animals live together and in which their reciprocal relationships often exhibit a high
level of organization. The farm, or physical space occupied by the herd could then be
considered as an anthropo-biotope. Agroecosystem health studies further this process by
examining the larger milieu in which these ecologic units exist, not only with broader
ecologic discipline studies but also with ecologic studies in a much broader sense
including concepts from the agricultural, health, epidemiologic and social sciences. In
this context, agroecosystem health studies are to ecopathology what ecopathology is to
herd medicine ± a broader context in which diagnostic medicine can be understood and
evaluated. Ecopathology provides the context for herd medicine and agroecosystem
health provides the context for ecopathology.
Systemic views of complex reality are open to many different legitimate perspectives,
which is why at least one of the authors (Waltner-Toews) sees poetry as being
complementary to scientific understanding in the search for resolutions to some of our
most intractable health and environmental problems.
Mad Cows and a Bill from the Power Company
It begins in New Guinea
with the fear of your ancestors.
It begins with the love of wisdom and intelligence.
It begins with respect for the dead,
with women and children first.
It begins with our cleverness
and our envy.
It begins with eating the brains
of those whom you admire.
It ends with Kuru, a spongy Jacuzzi
of laid back prions, engulfing the brain.
It ends with a young woman
throwing herself into the fire.
 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128 125

It begins in America with a dust bowl,

with a world war's devastation.
It begins with hungry children
in Europe and Africa.
It begins with our cleverness
and our lust for power and our tractors.
It begins with the cows coming in from the green
wilderness in droves.
It begins with the praise of hamburgers
on every tongue.
It begins with some spare change
after shopping for food,
and the thrill of a new car.
It begins with recycling, with efficiency,
with cows consuming cows.
It ends with old men
hungry for power.
It ends with a mob of mad cows
fed to power stations,
shimmering up in smoke
from the incinerators.

It begins with the love of life.

It begins in a white coat.
It begins with cutting and sewing.
It begins with new drugs.
It begins with our cleverness
and our fear of death.
It begins with ingesting those
who have what we want.
It begins with blood transfusions,
with hearts, kidneys, and corneal transplants,
with bone marrow and dura mater.
It ends with rabies, with AIDs,
with a slow toppling of prions
across the brain.
It ends with a young man,
unable to walk, unable to speak
his own name.

It begins with flowers and wine.

It begins with clever conversation.
It begins with the love of children,
of making them, and, surprisingly, caring for them.
It begins with a house and a car
126 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128

and a school and a computer.

It ends with paying
the power bills.
It comes round
to huddling by a campfire
singing plaintive melodies
with old guitars,
and, in the ash-black darkness at our backs,
the sound of cows or bears foraging,
and the slow sigh of a rising moon.

References

Barnouin, J., Brochard, M., 1981. L'enqueÃte eÂcopathologique continue: 1. Une nouvelle meÂthode de travail en
pathologie chez les ruminants. Bull. Techn. CRZV Theix, INRA 46, 11±14.
Barnouin J., Approche eÂcopathologique de la composante nutritionnelle des troubles de sante chez la vache
laitieÁre. TheÂse Univ. Montpellier II, 1992, p. 175.
Barnouin, J., Chacornac, J.P., Aissaoui, C., Le Idilbi, N., Mazur, A., 1994. Comment deÂpister les deÂseÂquilibres
biologiques et les troubles de sante chez la vache laitieÁre dans le cadre d'enqueÃte d'eÂcopathologie. Vet. Res.
25, 104±109.
Barnouin, J., Chassagne, M., Aimo, I., 1995. Dietary factors associated with milk somatic cell counts in dairy
cows in Britanny, France. Prev. Vet. Med. 21, 299±311.
Bawden, R., 1995. On the systems dimension in FSR. J. Farming Syst. Res. and Ext. 5, 1±18.
Beaudeau, F., Frankena, K., Fourichon, C., Seegers, H., Faye, B., Noordhuizen J., 1994a. Associations between
health disorders during two consecutive lactations and culling in dairy cows. Livest. Prod. Sci. 38, 207±
216.
Beaudeau, F., Frankena, K., Fourichon, C., Seegers, H., Faye, B., Noordhuizen, J., 1994b. Associations between
health disorders of french dairy cows and early and late culling within the lactation. Prev. Vet. Med. 19,
213±231.
Bigras-Poulin, M., Meek, A.H., Blackburn, D.J., Martin, S.W., 1985. Attitudes, management practices and herd
performances. A study of Ontario farm managers. II. Associations. Prev. Vet. Med. 3, 241±250.
Boyle, M., Kay, J.J., Pond, B., 1996. State of the Landscape Reporting: the development of indicators for
the provincial policy statement under the planning act. Ontario Ministry of Natural Resources, Maple,
Ontario.
Calavas, D., Ducrot, C., Sabatier, P., 1996. Interaction entre observateur et observe en eÂcopathologie. Reflexions
meÂthodologiques et eÂpisteÂmiologiques. Nature, Sciences et SocieÂte 4, 341±350.
Caley, M.T., Sawada, D., 1995. Mindscapes: the epistemology of Magoroh Maruyama. Gordon and Breach
Publishing, Newark, NJ.
Chassagne, M., Barnouin, J., Faye, B., 1996. EpideÂmiologie descriptive de la retention placentaire en systeÁme
intensif laitier en Bretagne. Vet. Res. 27, 491±501.
Checkland, P., 1981. Systems Thinking, Systems Practice. John Wiley and Sons, Chichester, UK.
Dohoo, I., Ducrot, C., Fourichon, C., Donald, A., Hurnik, D., 1996. An overview of techniques for dealing with
large numbers of independent variables in epidemiologic studies. Prev. Vet. Med. 29, 221±239.
Faye, B., Barnouin, J., 1987. Objectifs et motivations de l'enqueÃte eÂcopathologique Bretagne. Bull. Techn.
INRA-CRZV, Theix, France.
Faye, B., LefeÁvre, P.C., Lancelot, R., Quirin, R., 1994a. Ecopathologie animale: meÂthodologie et applications en
milieu tropical. INRA-CIRAD, Versailles, pp. 118.
Faye, B., Dorr, N., Lescourret, F., Barnouin, J., Chassagne, M., 1994b. Les infections intra-mammaires chez la
vache laitieÂre dans l'enqueÃte eÂcopathologique Bretagne. INRA Prod. Anim. 7, 55±65.
 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128 127

Faye, B., Dorr, N., Lescourret, F., Barnouin, J., Chassagne, M., 1994c. Farming practices associated with the
udder infection complex. Vet. Res. 25, 213±218.
Faye, B., 1995. L'eÂleveur, son environnement, ses pratiques, et la sante de son troupeau. TheÁse. Univ. Paris XII,
pp. 225.
Faye, B., PeÂrochon, L., 1995. La mortalite des vaches laitieÁres dans l'enqueÃte eÂcopathologique Bretagne. Vet.
Res. 26, 124±131.
Faye, B., 1996. Relations homme-animal dans les eÂlevages laitiers intensifs de Bretagne (France).
Anthropozoologica 22, 17±22.
Faye, B., Barnouin, J., 1996. L'eÂcopathologie ou comment aborder la pathologie multifactorielle. INRA Prod.
Anim., Hors seÂrie 50 eÂme anniversaire, pp. 127±134.
Francis, G., 1996. Implications for institutional arrangements. Presentation to the New Science: a workshop on
the ecosystem approach and its application to agriculture and governance, Februrary 21, 1996. University
of Guelph, Guelph, Canada.
Gallopin, G., 1997. Development and application of an integrated conceptual framework for tropical
agroecosystem research based on complex systems theories. First international workshop of the CIAT/UG
project, May 26±28, 1997. CIAT, Cali, Columbia.
Gallopin, G., 1995. The potential of agroecosystem health as a guiding concept for agricultural research.
Ecosystem Health 1, 129±140.
Giampietro, M., 1994. Sustainability and technological development in agriculture. Bioscience 44, 677±689.
Giampietro, M., Pastore, G., 1997. A model of analysis to study the dynamics of rural intensification in China. in
preparation.
Kay, J.J., 1991. A nonequilibrium thermodynamic framework for discussing ecosystem integrity. Environmental
Management 15, 483±495.
Kay, J.J., Schneider, E., 1994. Embracing complexity. Alternatives 20, 33±39.
Landais, E., 1994. SysteÁmes d'eÂlevage. D'une intuition holiste aÁ une meÂthode de recherche, le cheminement
d'un concept. In: Dynamique des systeÁmes agraires. A la croiseÂe des chemins: pasteurs, eÂleveurs,
cultivateurs. ORSTOM, Paris, pp. 15±49.
Lancelot, R., Faye, B., Lescourret, F., 1997. Factors affecting the distribution of clinical mastitis among udder
quarters in french dairy cows. Vet. Res. 28, 45±53.
Lescourret, F., PeÂrochon, L., Coulon, J.B., Faye, B., Landais, E., 1992. Modelling an information system using
the Merise method for agricultural research: the example of a database for a study on performances in dairy
cows. Agric. Syst. 38, 149±173.
Lescourret, F., Genest, M., Barnouin, J., Chassagne, M., Faye, B., 1993. Data modelling for data base design in
production and health monitoring systems for dairy herds. J. Dairy Sci. 76, 1053±1062.
Mallows, C., Tukey, J.W., 1982. An overview of techniques of data analysis emphasizing its exploratory aspects.
In: Tiago de Oliviera, J., Epstein, B. (Eds.), Some recent advances in statistics. Academic Press, London,
pp. 111±172.
Roling, N., 1996. Towards an interactive agricultural science. European J. Agric. Educ. and Ext. 2, 35±48.
Rowley, T., Gallopin, G., Waltner-Toews, D., Raez-Luna, E., 1997. Developing an integrated conceptual
framework to guide research on agricultural sustainability for tropical agroecosystems: CIAT-University of
Guelph Project. Ecosystem Health, in press.
Rosner, G., 1993. The ecopatho-transfert programme of the Centre d'Ecopathologie Animale: an example of
development and knowledge transfert. Proc. of Int. Symp. on ecopathology and animal health management
(abstract.). INRA, Clermont-Ferrand, France, p. 43.
Sabatier, P., Forestier, J., Marzin, P., 1994. L'eÂlevage, le conseil et l'eÂcopathologie. ReÂsultats d'une approche
didactique et situations de diagnostic d'eÂlevage en production porcine. Vet. Res. 25, 290±299.
Schwabe, C.W., 1982. The current epidemiological revolution in veterinary medicine. Prev. Vet. Med. 1, 5±15.
SeÂbillotte, M., 1978. ItineÂraires techniques et eÂvolution de la penseÂe agronomique. Cahier ReÂcherche Acad.
Agric. France 1416, 903±906.
Tillon, J.P., 1982. EnqueÂte permanente sur la pathologie porcine: choix des populations eÂtudieÂes. Bull. Assoc.
Epidemiol. Mal. Anim. 1, 32±35.
Tourte, R., 1965. Suggestions pour une politique d'application de la recherche agronomique dans les Pays en
voie de DeÂveloppement. L'agronomie Tropicale 11, 1163±1176.
128 B. Faye et al. / Preventive Veterinary Medicine 39 (1999) 111±128

Tuffery G., 1977. Recherches sur la bucephalose aÁ Bucephalus polymorphus Buer, 1827. Introduction aÁ
l'eÂcopathologie des systeÁmes piscicoles. TheÁse Doctorat d'eÂcologie. Univ. P. and M. Curie, Paris VI,
France, pp. 137.
Von Bertalanffy, L., 1970. La theÂorie des systeÁmes. Dunod, Paris, p. 298.
Waltner-Toews, D., 1996. Ecosystem health: a framework for implementing sustainability in agriculture.
Bioscience 46, 686±689.
Waltner-Toews, D., Nielsen, N.O., 1995. Assessing Agroecosystem Health. Discussion Paper 23. Agroecosys-
tem Health Project, University of Guelph.
Waltner-Toews, D., Wall, E., 1997. Emergent perplexity: in search of post-normal questions for community and
agroecosystem health. Social Science and Medicine, in press.