a multi-disciplinary framework for bio-economic modeling in

This article was downloaded by: [80.200.203.7]
On: 14 January 2013, At: 01:24
Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered
office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK
Aquaculture Economics & Management
Publication details, including instructions for authors and
subscription information:
http://www.tandfonline.com/loi/uaqm20
A MULTI-DISCIPLINARY FRAMEWORK
FOR BIO-ECONOMIC MODELING IN
AQUACULTURE: A WELFARE CASE STUDY
a
b
c
d
Chris Noble , Iain K. Berrill , Bob Waller , Markus Kankainen
d
a
e
, Jari Setälä , Pirjo Honkanen , Cecilie M. Mejdell , James F.
b
a
f
a
Turnbull , Børge Damsgård , Oliver Schneider , Hilde Toften ,
g
Adriaan P. W. Kole & Sunil Kadri
a
h
Nofima, Tromsø, Norway
b
Institute of Aquaculture, University of Stirling, Stirling, United
Kingdom
c
Freedom Food Ltd., Horsham, West Sussex, United Kingdom
d
Finnish Game and Fisheries Research Institute, Turku Game and
Fisheries Research, Turku, Finland
e
National Veterinary Institute, Oslo, Norway
f
IMARES, Yerseke, Netherlands
g
Wageningen University and Research Centre - Centre for Innovative
Consumer Studies, Wageningen, The Netherlands
h
Trans-National Consulting Partnership, Glasgow, United Kingdom
Version of record first published: 30 Nov 2012.
To cite this article: Chris Noble , Iain K. Berrill , Bob Waller , Markus Kankainen , Jari Setälä , Pirjo
Honkanen , Cecilie M. Mejdell , James F. Turnbull , Børge Damsgård , Oliver Schneider , Hilde Toften ,
Adriaan P. W. Kole & Sunil Kadri (2012): A MULTI-DISCIPLINARY FRAMEWORK FOR BIO-ECONOMIC
MODELING IN AQUACULTURE: A WELFARE CASE STUDY, Aquaculture Economics & Management, 16:4,
297-314
To link to this article: http://dx.doi.org/10.1080/13657305.2012.729250
PLEASE SCROLL DOWN FOR ARTICLE
Full terms and conditions of use: http://www.tandfonline.com/page/terms-and-conditions
This article may be used for research, teaching, and private study purposes. Any
substantial or systematic reproduction, redistribution, reselling, loan, sub-licensing,
systematic supply, or distribution in any form to anyone is expressly forbidden.
Downloaded by [80.200.203.7] at 01:24 14 January 2013
The publisher does not give any warranty express or implied or make any representation
that the contents will be complete or accurate or up to date. The accuracy of any
instructions, formulae, and drug doses should be independently verified with primary
sources. The publisher shall not be liable for any loss, actions, claims, proceedings,
demand, or costs or damages whatsoever or howsoever caused arising directly or
indirectly in connection with or arising out of the use of this material.
Aquaculture Economics & Management, 16:297–314, 2012
Copyright # Taylor & Francis Group, LLC
ISSN: 1365-7305 print/1551-8663 online
DOI: 10.1080/13657305.2012.729250
Downloaded by [80.200.203.7] at 01:24 14 January 2013
A MULTI-DISCIPLINARY FRAMEWORK FOR BIO-ECONOMIC
MODELING IN AQUACULTURE: A WELFARE CASE STUDY
Chris Noble1, Iain K. Berrill2, Bob Waller3, Markus Kankainen4, Jari Setälä4,
Pirjo Honkanen1, Cecilie M. Mejdell5, James F. Turnbull2, Børge Damsgård1,
Oliver Schneider6, Hilde Toften1, Adriaan P. W. Kole7, and Sunil Kadri8
1
Nofima, Tromsø, Norway
2
Institute of Aquaculture, University of Stirling, Stirling, United Kingdom
3
Freedom Food Ltd., Horsham, West Sussex, United Kingdom
4
Finnish Game and Fisheries Research Institute, Turku Game and Fisheries Research,
Turku, Finland
5
National Veterinary Institute, Oslo, Norway
6
IMARES, Yerseke, Netherlands
7
Wageningen University and Research Centre - Centre for Innovative Consumer Studies,
Wageningen, The Netherlands
8
Trans-National Consulting Partnership, Glasgow, United Kingdom
& This article summarizes the framework that translated data from multiple disciplines into a
bio-economic decision tool for modeling the costs and benefits of improving fish welfare in
commercial aquaculture. This decision tool formed the basis of a recent EU research project,
BENEFISH which was funded via the European Commission’s Sixth Framework (FP6) initiative.
The bio-economic decision model can incorporate biological data, productivity data, micro (farm)
and macro (industry) level economic data, and consumer marketing and business to business data.
It can identify areas for potential added value that might be achieved by improving fish welfare
across a range of species and husbandry systems within European aquaculture. This article
provides a brief overview of the minimum data requirements for successfully modeling the
bio-economic impacts of improvements in farmed fish welfare using the model developed during
the BENEFISH project. It also highlights potential bottlenecks and the minimum prerequisites
for each potential data set to be used for successful modeling.
Keywords aquaculture, bio-economic modeling, framework, multi-disciplinary, welfare
Address correspondence to Chris Noble, Nofima, Muninbakken 9-13, P.O. Box 6122, NO-9291,
Tromsø, Norway. E-mail: [email protected]
298
C. Noble et al.
Downloaded by [80.200.203.7] at 01:24 14 January 2013
DEVISING A BIO-ECONOMIC DECISION TOOL FOR FISH
WELFARE
To develop a robust bio-economic decision analysis tool, stakeholders,
researchers and any other potential end users must adopt a stringent
framework for using and utilizing potential data sources. However, assessing the bio-economic implications of implementing fish welfare actions
is fraught with problems, primarily due to drawing, assembling and managing inputs from a wide range of disciplines including fish biology, aquaculture production, welfare monitoring, ethics, marketing and economics (see
Berrill et al., 2009 for an in-depth summary). It also requires inputs and
guidance from welfare assurance schemes and standards agencies, across
a range of fish species and husbandry systems.
The primary objectives of this article are to describe a multi-disciplinary
modeling framework for the BENEFISH decision tool (Kankainen et al.,
2012) for quantifying the costs and benefits of improving fish welfare in
aquaculture. It will provide the reader with a brief overview of the minimum data requirements and guidance on best practice, highlighting potential bottlenecks and the minimum prerequisites for each potential data set
to be used. We then discuss the implications and limitations of these
requirements for the successful utilization of the model.
Prerequisites for Robust Modeling
Potential Welfare Threats and Appropriate Welfare Actions
To model the bio-economic consequences of actions or practices that
improve welfare, one must (i) identify and quantify potential welfare
threats, and (ii) develop intervention strategies to reduce or eliminate
these threats. However, the identification of such threats requires some
means of assessing and quantifying welfare on the farm and how animal
welfare might change as a result of a particular action. For terrestrial animals, welfare assessments are generally based upon information that is
drawn from a range of studies measuring the effects of husbandry
systems and rearing methods (e.g., Fraser, 2003). But in fish, there are only
a limited number of studies that attempt to bring such information
together.
In general, any welfare assessment can utilize several different ways of
defining and measuring the welfare of an animal (or population). Assessment may be needs-based, aiming to ensure that the specific needs of a
particular farmed species are met, whilst measuring the effects of nonfulfilment of these needs. Alternatively aspects of welfare assessment may
be based upon measuring deviations from normality for a specific animal
in a specific rearing system (for example deviation from normal feeding
Downloaded by [80.200.203.7] at 01:24 14 January 2013
Bio-Economic Modeling of Fish Welfare Interventions in Aquaculture
299
might be indicative of a welfare problem). To quantify deviations from
normality does not necessarily involve measuring deviations from what
might be expected in wild fish (as a benchmark) and instead quantifies
deviations from baseline levels recorded in comparable farm animals.
A welfare assessment can also be indicator based, using either animalbased (e.g., body condition, levels of fin damage; Ellis et al., 2008) or
resource-based indicators i.e., measures from the environment or rearing
system (e.g., water quality parameters, tank size, feeding method). However, for on-farm welfare assessment it is evident that whilst many animaland resource-based measures can be used to quantify aspects of welfare,
some of those indicators cannot be easily measured on a farm. Consequently, many scientists and farmers investigating animal welfare on farms
focus on so-called Operational Welfare Indicators (OWIs). OWIs are
defined as indicators that must provide a valid reflection of welfare, must
be repeatable and comparable, and must be relatively inexpensive and easy
to measure on a farm. Examples of OWIs include mortality, fin damage,
feed efficiency and growth. Changes in measures such as histopathology
or gene expression, for example, may act as good indicators of welfare,
but should not be considered as OWIs.
Any welfare assessment that uses an indicator based approach can be
based upon a suite of species and system specific OWIs, which may be combined into welfare indices, where each OWI is weighted according to its
relative importance. In practice, it is likely that a welfare assessment will
concentrate on few key welfare indicators and their accompanying parameters, which should be relatively easy to measure under field conditions.
A welfare action or intervention can be defined as any technological or
methodological step that is taken to improve fish welfare. For example
increased water flow rate in flow-through rearing systems could be an intervention to alleviate poor water quality, which might be negatively impacting
welfare, as identified by high mortality. Not all welfare interventions are
necessarily viewed as beneficial by everyone, and the action itself could
be regarded as either right or wrong, even if an OWI is improved, depending on a particular individual’s ethical stance.
This can be illustrated by some welfare interventions from the poultry
industry. Serious feather pecking is a widespread problem during egg production, leading to increased mortality in those birds that are attacked.
One possible welfare intervention to address this problem is to farm a blind
strain of hen, which is not able to peck other members of its population.
Although the hens are blind, they are otherwise healthy and produce just
as many eggs as sighted birds. Some people would argue that it is not ethically right to deliberately change a sighted creature into a blind one, to
improve another aspect of welfare, and this would be regarded as an
offense to the birds’ integrity (Lassen et al., 2006).
Downloaded by [80.200.203.7] at 01:24 14 January 2013
300
C. Noble et al.
A welfare intervention may also improve one OWI, but have negative
impact on other aspects of welfare. In the 2004 article ‘‘Laying hen welfare
standards: A classic case of ‘power to the people,’’ Savory argued that cages are
more beneficial for the bird’s welfare, because of better health and lower
mortality. However, the public does not generally regard battery cages as
an appropriate welfare action to reduce mortality, and in many consumers’
view of animal welfare, freedom of hens to move around and flap their
wings, and display natural behaviors like dust bathing, nesting behavior,
perching and ground pecking, outweighs the health benefits of being
reared in a battery cage.
Thus, cage rearing is not regarded as an appropriate welfare action to
reduce mortality. A similar example can be presented for the fish
farming industry. The use of triploid fish (instead of diploids) has been
identified as a possible intervention to reduce mortality in the UK rainbow
trout farming industry (see Berrill et al., 2012). Further, because triploids
are sterile escapees from farms, they are a reduced threat to wild fish stocks
and do not experience some of the negative traits often associated with
maturation, which can themselves negatively impact welfare (e.g., reduced
immuno-competence, Suzuki et al., 1997).
However, for some individuals (including consumers) triploid fish may
be perceived as unnatural as they are not able to reproduce, which in itself
might be seen as reduced welfare. Such variations in public opinion are
likely to have a significant impact on consumers’ willingness to pay for
products that have been produced using a specific intervention to improve
welfare. This has implications for bio-economic modeling the consequences of specific welfare interventions.
In practice, modeling the bio-economic consequences of welfare
interventions will involve quantifying the relationships between any
welfare threat, establishing a proposed welfare action or intervention
to address that threat, and quantifying any resultant improvement in welfare, as measured (in the case of our bio-economic model) by a change
in a particular OWI. To do this, it is necessary to establish suitable data
sets that can be analyzed to identify welfare threats and interventions.
Such data sets can be either experimental or commercial in origin, providing they are sufficiently robust for analysis and to identify welfare
threats that are applicable to on-farm conditions. Data sets used for this
process may be novel, in that they are specifically collected for the
purpose of identifying, and quantifying the effects of a specific welfare
action.
Alternatively, existing data resources can be analysed to identify a welfare action. The use of existing data resources is not without its difficulties.
For example, if a data set is collected for a specific purpose and then made
available to study fish welfare, data may be difficult to re-analyze to answer
Downloaded by [80.200.203.7] at 01:24 14 January 2013
Bio-Economic Modeling of Fish Welfare Interventions in Aquaculture
301
the alternative research question, particularly if there is insufficient
supporting metadata. Further, before any existing data resources can be
analyzed, it will be necessary to gain approval to use those data from the
relevant owner(s). In practice this can take some time.
However, there are also some benefits to using existing data resources
compared with collecting new data. Collecting novel data can be expensive,
whereas analyzing existing data resources can be relatively inexpensive.
Also, for studies which focus on animal welfare, the use of existing data
resources does not require any additional experiments where the welfare
of animals might be impacted. Finally, the use of existing data sets can give
those data considerable added value.
Although some threats to welfare are broadly generic and relate to
more than one species, rearing system or life stage, in most cases threats
to welfare will differ between species, life stages and rearing systems. For
example low dissolved oxygen is a threat to welfare for most species of fish,
but the actual levels that are harmful will vary between species and life
stages. Consequently any welfare action, and the data set used to identify
that action, must be specific i.e., it is not possible to provide a generic
welfare action for all farmed fish.
The process BENEFISH used to identify and quantify welfare actions
from various data sets (either novel or existing) first involves mathematical
or statistical analysis of that data set to identify aspects of the farming system that might impact welfare (i.e., identifying welfare risk factors). For this
process, OWIs provide a good, practical outcome measure of welfare in any
analyses. To be effective, and successfully modelled in the bio-economic
model, the identified threats and their corresponding welfare actions must
be clear and unequivocal.
Actions to directly address those welfare threats, and thus improve welfare as measured through the chosen OWI, must then be developed. The
development of welfare actions is not necessarily a product of data analysis.
Instead it involves scientific interpretation of the welfare threats that have
been identified and combined, where necessary, with consultation with
farmers=welfare experts to develop the most effective and practical interventions to address the threat.
Once a suitable intervention has been established, it is also important
(for bio-economic modeling) to establish the efficiency with which that
action improves welfare (in the form of the change to the OWI). This information may be available from the data set used to identify the welfare
action, or it may be available elsewhere, but it is vital for bio-economic
modeling.
Finally, although the above process will have identified suitable welfare
actions for bio-economic modeling, the economic consequences of those
actions can only be quantified if suitable supporting farm (micro) and
302
C. Noble et al.
industry (macro) data is available to parameterize the model. Without that
data, the action will still be effective but we will not be able to quantify its
biological and economic consequences. These data requirements will be
addressed in the following sections of this article.
Downloaded by [80.200.203.7] at 01:24 14 January 2013
Quantifying Welfare Threats and Potential Actions in Relation to
Productivity Parameters
To quantify any potential welfare actions upon the productivity of an
aquaculture system, modellers must gather robust data on their potential
effects upon key productivity indicators. The productivity indicators must
be commonly used amongst aquaculturists, researchers and policy makers
and the BENEFISH project thus chose the following three parameters:
growth, feed conversion ratio (FCR) and mortality. By using these widely
used and applicable productivity indicators, the operational context of
the model can be improved.
The relationship between the three chosen productivity indicators and
any potential welfare improvement measures should be identifiable and
quantifiable. These three indicators have a direct impact upon the economic performance of the farm. Poor growth limits a farms’ profitability, primarily by extending the length of time required to grow fish to a market
size, and also increased risks associated with escapes and disease losses.
FCR is a measure of the efficiency with which food is turned into body
tissues. With feed costs accounting for up to 60% of a farms total cost outlay
(Sveier and Lied, 1998), any inefficiency in feed use can rapidly impact the
profitability of a farm.
Any mortality clearly impacts overall production biomass and therefore
productivity. However, the impacts of mortality on productivity are variable
depending on when mortalities occur in the production cycle. The longer a
fish has been in a production system before it dies, the more effort and
financial resources (e.g., feed, labor, general farm running costs) that have
gone towards producing that fish, and the larger the impact its death will
have on productivity.
Establishing the relationships between specific OWI’s and these productivity indicators is an essential component of the bio-economic modeling process and can also enhance the commercial applicability of the data
set. In addition, any advances in aquacultural productivity are crucial drivers for the uptake of and successful implementation of any welfare actions.
Therefore, a minimum requirement of any potential data set or literature
source that quantifies the effects of welfare threats and actions upon OWIs,
must be to also provide information on the consequences for growth,
feeding efficiency and mortality.
Bio-Economic Modeling of Fish Welfare Interventions in Aquaculture
303
MODELING
Downloaded by [80.200.203.7] at 01:24 14 January 2013
To obtain suitable data for the decision analysis tool, an end user must
utilize a value chain approach, identifying welfare actions and indicators in
various European aquaculture systems, analyzing the cost-benefit consequences for production, fish quality and consumer perception, and formulate these inputs for bio-economic modeling. This data must be audited for
quality and any uncertainties in the robustness of the data must be highlighted. Metadata must also be provided to increase the uptake and understanding of the data sets produced from each of the various disciplines (see
also Olson et al., 1999).
Sourcing Suitable Farm (Micro Level) Data
Sourcing sufficient and suitable data covering farm productivity is an
essential pre-requisite for successfully modeling (Piedrahita, 1988). As stated earlier, the three major productivity parameters are: growth (rate), feed
conversion efficiency and mortality. To quantify their on-farm effects and to
increase the commercial applicability of the data sets, sourcing applied
farm data is essential. For instance, minimum data requirements include
accurate information on i) farm production volume and ii) the % of production & production volume this welfare intervention will affect. Based
upon this data, the potential impacts of the welfare action can be estimated. In addition to this, some other operational data is needed, such
as average starting and end weight and the duration of production cycle
(see also Table 1).
The majority of farms (if not all of them) record data on fish growth,
feeding efficiency and mortality as part of their daily management practices. However, some of these records may be incomplete for part of the
production cycle, especially on poorly managed farms with labor and cash
flow issues. When these factors are taken into consideration, sourcing commercial farm level data can be very challenging. Irrespective of this, farm
level data should be:
a. Specific: for a given species, its origin, age, grade, husbandry system;
b. Measurable: there needs to be sufficient data to detect the effects of any
given welfare action upon the value chain;
c. Attainable: farmers need to be able to record data efficiently and
effortlessly;
d. Realistic: the farmer or their recording system must collect and record
the necessary data;
e. Time bound: data needs to be time bound in terms of start and end
date;
304
C. Noble et al.
f. Efficient: monitoring, recording and analyzing systems need to be
effective and efficient; and
g. Auditable: Bench mark lines need to be established and re-evaluated
often to avoid impacts of the specific factors mentioned earlier.
TABLE 1 Summary of Parameters, Values and Uncertainties for Robust Modeling Covering: a) Data
Requirements for Robust Modeling at the Farm (Micro) Level and b) Data Requirements for Robust
Modeling at the Industry (Macro) Level
Downloaded by [80.200.203.7] at 01:24 14 January 2013
Name of Input
System description
Production volume
Welfare intervention affected volume
Average starting weight
Average end weight in period
Average Production cycle
Average mortality
(Cumulative mort biomass for period)
Ave. FCR
System economy
Producer price
Cost=benefit factor share of producer price
Fingerlings
Feed
Other
Work
Investments
Capital
Profit
Welfare intervention effects
Costs for implementation for
average company
Equipment investment
Energy=other costs
Change in capital costs
Added work
Intervention efficacy
Utility for productivity factors
Growth effect
Feed effect
Survival effect
Supply Chain effects
Processing premium
of which targeted to case
Consumer premium
of which targeted to case
Change in total demand
Changed share of welfare
labelled product
Value=
Averages:
Unit
Quality of
Uncertainty Reference
Tons
%=of total
grams
grams
weeks
%=total pieces=period
%=production volume
Feed kg=fish kg
S
S
E
E
E
S
4($)=kg
S
S
E
%
%
%
%
%
%
%
E, C
E, C
E, C
E
4($)=year/tank
4($)=year
4($)=year
4($)=year
%
%
%
%
SD
SD
SD
R
R
R
%
%
%
%
%
% units
þ 20% TRI
E
S
R
S
E
E
SD
Uncertainty is determined according to quality of reference; R ¼ Research variance, Statistics ¼
10%, Commercial ¼ 5% and Expert statement ¼ 20%, uncertainty is given to variables not constant
values.
Bio-Economic Modeling of Fish Welfare Interventions in Aquaculture
305
Downloaded by [80.200.203.7] at 01:24 14 January 2013
Sourcing Suitable Industry (Macro Level) Data
The robustness of industry (macro) level economic data depends, to a
large extent, on the strength of national surveillance and monitoring programs implemented by either the government, farm stakeholder associations or other interested groups. As stated in the previous section, if the
influence of a welfare intervention on a target market is to be accurately
evaluated, economic data must first be sourced from individual farms and
incorporated into the bio-economic model. This farm level information is
then scaled up to the industry level using aggregated industry data.
However, if the welfare measures are not evaluated at the industry level
for their impact, their effect is somewhat unclear. For example, if a fish welfare intervention is implemented in a regional segment of the industry, but
is not implemented at the national level, this can be difficult to model. It is
therefore imperative that accurate national and transnational data is available to allow the model to expand the outputs to the macro scale and to
judge the overall impact of the welfare intervention on the industry. A summary of macro level data requirements is shown in Table 1.
Sourcing industrial (macro) level data in a national context can be
rather difficult. In some of the more mature aquaculture industries, there
are robust national programmes for auditing and monitoring farmers, and
economic data on factors such as market price and other cost factors can be
sourced relatively easily. The model can also treat this data with minimal
uncertainty, as it is drawn from a large sample of producers. Other data that
may not be readily available (such as average FCR or average market size,
see Table 1 for a summary) can be sourced from industrial partners or some
of the larger aquaculture companies.
This is an added benefit of having a large aquaculture producer
involved as an industrial advisor to the modeling process, as they can provide the project with applicable data from the outset. For less mature aquacultural markets, with a small number of producers and potentially poor
national surveillance programs, sourcing economic data may prove problematic. As a lack of data can have a serious and detrimental effect upon
the robustness of the model outputs, the value of modeling welfare
interventions in such markets must be evaluated from the outset. If it is
envisaged that there will be a lack of data, or data required for Table 1 will
be of poor quality, a pragmatic decision on whether to proceed with the
chosen data set should be taken immediately.
Sourcing Suitable Consumer Perception and Added Value Data
One key aspect of BENEFISH was to deliver quantitative data to model
potential market gains that could be derived from welfare actions in the
Downloaded by [80.200.203.7] at 01:24 14 January 2013
306
C. Noble et al.
husbandry system. To identify key added value factors, an added value
component of the data set must address the needs of both businesses
and consumers. It must address and quantify the values attributed to welfare indicators and any potential welfare improvement measures. Actual
added value at the market side of the value chain was quantified as a
‘‘willingness to buy’’ and a ‘‘willingness to pay,’’ either by business-tobusiness customers, consumers or both.
‘‘Willingness to buy’’ refers to sales volume that should rise as a proof of
added value, whereas ‘‘willingness to pay’’ refers to the willingness to pay a
premium compared to regular prices. Both are likely to be triggered by the
added value as perceived by customers and=or consumers. The price one is
willing to pay depends on many factors, most of them subjective. Objective
differences between fish raised to a welfare standard and those raised
under regular farming conditions are often difficult to prove, meaning
potential buyers must be convinced about any potential added value.
Communication is crucial here, and one of the most well-known aspects
of marketing. Further, two key players were identified in the marketing side
of the value chain: business to business customers and consumers. They are
likely to judge the qualities of a product differently and to value them differently. Data collection on costs and benefits at the market side therefore
concentrated on the following issues:
. Perceived added values in fish raised to improved welfare standards over
regular farmed fish, both for B2B customers and consumers;
. The effectiveness of communication about the welfare indicators;
. Their willingness to buy and to pay for welfare friendly farmed fish; and
. The costs involved in marketing and communication
With respect to perceived added value, these can roughly be related to
intrinsic product qualities (e.g., appearance, taste, organoleptic characteristics, shelf life), and to extrinsic product qualities (e.g., price, labels, package). Modern marketing theory and practices show that the market value of
products may vary enormously depending on the interaction between
intrinsic and extrinsic product qualities. The perception of benefits affects
the perception and value of (intrinsic and extrinsic) qualities of the fish
products. The chosen welfare actions and their potential consequences
need to be communicated to buyers.
The effectiveness of animal welfare communication depends on the
personal weight that consumer will put on animal welfare in their purchase
decisions, and on the perceived reliability of the information source. Other
factors that determine the effectiveness of the welfare communication are
knowledge level (the more abstract the information, the more difficult it
is to process), involvement and motivation to process animal welfare
Bio-Economic Modeling of Fish Welfare Interventions in Aquaculture
307
Downloaded by [80.200.203.7] at 01:24 14 January 2013
information, and the type of information (emotional or factual). To model
costs and benefits, the robustness of the model improves with the amount
of data it incorporates. However, in the end there are only two dependent
variables that matter: the willingness to buy and the willingness to pay –
balanced against the costs involved.
Customers
To produce data for the business-to-business added value, a series of
data collection experiments be carried out. First, qualitative data from
industry representatives are the quickest and cheapest way to get an
impression of stakeholders’ perceptions of benefits and costs of welfare
actions. For modeling, quantitative data are needed. In BENEFISH, a short
questionnaire was designed for collecting data about stakeholder perceptions of the expected benefits and costs of increased fish welfare at trade
shows, conferences and exhibitions.
However, this proved to be a difficult method for collecting data as it
was difficult to get the industry representatives to return a completed questionnaire. In order to get data on how sales volumes are affected by welfare
products, industry data are needed. To achieve this, one might approach
potential customers directly. On the other hand, these data might be
sourced elsewhere, e.g., from welfare assurance and monitoring schemes
such as Freedom Foods Ltd., UK. In BENEFISH, B2B customers were likely
to be most focused on the impact that welfare actions might have on
processing qualities of the flesh.
Consumers
The data that was delivered to the bio-economical model was consumers’ willingness to pay for improved fish welfare (price premium: %
change compared to a reference price) and willingness to buy (that is,
how much of the welfare product consumers would buy at different
prices—assuming that such a product will be more expensive for the consumer). To get these data, establishing the most effective communication
route about welfare actions in fish farming is crucial. Next, one should
be aware that it is often hard to predict actual consumer purchase behavior
from stated preferences and attitudes. Experimental data are required,
preferably from real shopping environments.
To successfully model potential market gains, consumer studies can be
carried out using several approaches. Again, the fastest and cheapest way to
get preliminary information is through qualitative data, e.g., interviews. In
BENEFISH, a literature review was also carried out to find any indicators of
consumer perceptions and reactions to animal welfare, to find variables to
Downloaded by [80.200.203.7] at 01:24 14 January 2013
308
C. Noble et al.
be tested in later stages. The review showed that there were few studies on
consumer and business perceptions of fish welfare.
There are some studies on general animal and fish welfare issues and
sustainability (Frewer et al., 2005; Honkanen & Olsen, 2009), but data on
specific welfare actions is lacking. Following the literature study, qualitative
studies in the form of focus groups and interviews were conducted to provide meta-data for quantitative measures. Third, the most useful data comes
from consumer experiments testing consumers’ willingness to pay and buy
welfare friendly fish. Fourth, experiments conducted in shops where consumers buy fish can be considered to be among the most reliable.
In BENEFISH, the literature study and the qualitative studies indicated
that it would be difficult to test the importance of OWIs separately to consumers as they lack knowledge of fish farming practices. Instead a general
welfare indicator is used in the final bio-economic model (welfare=no welfare) in a limited or extended version. The extended communication did
provide additional information on the specific OWIs.
Other studies have shown that animal welfare is not the most important
factor deciding consumer purchases, especially not for consumers that tend
to focus on intrinsic qualities. Table 2 gives an overview of potential consumer values that were used in a conjoint experimental design to test for
their effect on (self-reported) willingness to buy and to pay. Thus the most
effective communication format can be established. More advanced
marketing methodology would account for different market segments:
the purchase behavior of different types of consumers might be affected
differently by welfare communication. In BENEFISH we segmented this
based upon attitudinal variables.
TABLE 2
The Setup of the Consumer Conjoint Experiments
Communication Variables
No. of Levels
Welfare indication
6
Information sources
2
Dependent variables
Willingness to pay
Willingness to buy
2
Content
-
No indication
Positive picture (emotional)
Negative picture (emotional)
Simple text: ‘‘welfare is taken care of’’
Simple text, extended with text on OWIs
Text connecting welfare with better sensory quality
Producers (organization)
Governmental body
Euros=USD for 200 grams of the product
- How many grams would you buy of this product for
[price at 15% premium]
- How many grams would you buy of this product for
[price at 25% premium]
Bio-Economic Modeling of Fish Welfare Interventions in Aquaculture
309
Downloaded by [80.200.203.7] at 01:24 14 January 2013
Monitoring Costs and Protocols
Any modeling of potential improvements in fish welfare must incorporate the costs of documenting, monitoring and regulating these welfare gains, from individual farm initiatives (micro level) through to the
whole industry, including national and international schemes and initiatives (macro level). The costs of such schemes must be addressed in the
bio-economic model, in addition to any potential economic benefits due
to increased exposure and improvements in market perception. To
address this component, links to aquaculture organizations, quality mark
providers and eco-label controllers will be invaluable in the modeling
process. The costs of any potential remedial actions, when welfare monitoring discovers deviations from the welfare standard must be also be
estimated and incorporated into the model. Any non-compliance identified at this assessment will need to be corrected and then continually
adhered to.
When joining a scheme, there will be input standards that need to be
implemented and these may or may not have a cost. Once a producer
has decided to proceed with a welfare monitoring scheme, a full independent assessment (inspection) is undertaken covering the entire scope
of the scheme, from either (i) birth to slaughter or (ii) to a specific area
of production and therefore costs will vary. Traditionally these assessments
(inspections) are pre-arranged, and therefore the producer is aware that
they are happening and has an opportunity to prepare for them. These
visits are charged for by the scheme owners and can be on a membership
basis or animals on the scheme or animals sold carrying a particular logo.
For any scheme to be successful, the producer must adhere to the assurance standard at all times. Some kind of monitoring scheme is therefore
required to audit the welfare action and these are best done unannounced
or with a minimal pre-warning. Various quality schemes for improving the
quality, welfare, and environmental sustainability of farmed fish now exist
and these include:
i. Industry-led schemes, e.g., Scottish Quality Salmon;
ii. Retailer-led schemes, e.g., UK supermarket quality schemes; the Label
Rouge government-led scheme in France (www.agriculture.gouv.fr);
and
iii. Niche market schemes (e.g., organic salmon schemes, welfare schemes
such as Freedom Foods).
Users of the model must pick a suitable scheme for the desired species and
incorporate monitoring and membership costs into the bio-economic
decision tool.
310
C. Noble et al.
Downloaded by [80.200.203.7] at 01:24 14 January 2013
DATA CONSOLIDATION INTO BIO-ECONOMIC MODEL
Once data is consolidated from each discipline, inputs for the bioeconomical modeling of each welfare action were collected from the biologists and industry experts with structured spreadsheets. First, average farm
size and average production costs were collected for the micro economic
analysis, and production volumes of the whole industry were collected for
macro level estimations. Second, estimates of the influence of the intervention implementation and monitoring to different cost categories were
acquired. These estimates could be based on industry data or expert
statements. Third, estimates of productivity changes were needed. This
information could be sourced from experimental or commercial data. Productivity improvements normally lead to savings in overall production costs.
The fourth step is to estimate the effects of the intervention on the
value chain. This requires an understanding of how improvements in biological productivity affect the added value of fish in different parts of the
value chain. For example, when fish are raised to specific welfare standards,
there may be a price premium from the first hand buyer or consumer (see
Olesen et al., 2010). It can also have an effect upon the volume of demand.
Both business-to-business and business-to-consumer markets may gain from
changes in intrinsic (i.e., appearance, taste, shelf life) and extrinsic product
qualities (i.e., price, labelling, packaging) of the product. Furthermore,
one has to evaluate how the consumer price premium is divided between
the stakeholders in the value chain in order to calculate benefits at the
producer level.
Once collected and audited, input values are then fed in the bioeconomic model (Kankainen et al., 2012), which calculates the costs of
implementation and monitoring, the gains of the productivity improvements and any potential added value for the production chain. The model
also merges all these costs and benefits into the overall utility estimations at
the micro and macro levels. The uncertainty in input values could be taken
into account through dispersion in estimates, which consequently means
that the final results can be shown as probability distribution. This makes
it easy for the decision maker to evaluate the uncertainty of results and
the risks in the decision making.
MODEL DEVELOPMENT
The BENEFISH model uses various scientific and industrial data
sources to model practical welfare actions in different production systems.
As the structure of the bio-economical decision tool is outlined in
Kankainen et al. (2012), it is only briefly highlighted here. One important
function of the decision tool is to illustrate the nature of the decision
Downloaded by [80.200.203.7] at 01:24 14 January 2013
Bio-Economic Modeling of Fish Welfare Interventions in Aquaculture
311
problem. Decision tools are developed to give structure and clarity to
complex decision objectives. The problem is normally divided into
smaller components, which are much easier to approach in a stepwise
fashion. However, it is vital to understand the dependencies between the
chosen model components. Thus, the most important phase of
bio-economical modeling is to decide the overall structure of the
model. This decision has an effect on all the other stages in the processing
of data.
In brief, the main driver of the costs and benefits is the welfare intervention. Once a valid intervention has been found, it is relatively easy to
structure the economic consequences in the modeling. Economic consequences were organized as separate utility components in the bio-economic
model, and thereupon the costs and benefits concerning these components can be separately modelled. The formulas and functions for utility
calculations were developed in cooperation between economists and biologists on a case-by-case basis. The inter-dependencies between different variables were partly based on experimental results, published literature
sources and expert statements.
The overall utility is the sum of these utilities. It is relevant to consider
both micro (farm) and macro (industry) level utilities. At the micro level,
one judges the utility of applying the welfare intervention for an average
company. Following on from this, one should then judge the utility for
the whole industry. This macro level utility gives a wider perspective to
the analysis and is essential for decision makers to estimate the relevance
of a welfare action for any given aquaculture industry. The overall structure
of the model and all its main components is fixed, irrespective of the aquaculture system or the species studied. However, the sub-modules including
all the functions behind the main components are case-specific and the
inputs have to be considered on a case-by-case basis. This structure makes
it possible to apply the model across various fish species and aquaculture
systems.
On the other hand, the use of the model demands a lot of specific
multi-disciplinary information. If these requirements can be fulfilled, the
bio-economical decision tool gives a reliable insight of the economic consequences of a welfare intervention. Altogether, profitability remains the
main factor driving the business processes, and therefore a tool helping
the economic evaluation is an important supplement to contribute rational
decision making.
Road Map and Framework Structure
For a simplified schematic of the BENEFISH framework see Figure 1.
Downloaded by [80.200.203.7] at 01:24 14 January 2013
312
C. Noble et al.
FIGURE 1 The framework of the bio-economic decision tool for evaluating the effects of fish welfare
actions on the entire value chain, summarizing each component of the model. Based upon the framework outlined in Berrill et al. (2009). (Color figure available online.)
CONCLUDING SUMMARY
Limitations
As is the case with numerous types of operational models and decision
tools, the BENEFISH bio-economic model makes assumptions and has its
limitations. The scope of the model is primarily based upon the factors
outlined in Table 1. If any of this micro or macro level data is missing or
is of poor quality, the efficacy and accuracy of the model will be reduced.
A modeller should, therefore, be transparent about the source of their
data, any assumptions they make, and adjust potential uncertainties accordingly. This will allow for transparent auditing of the results and will provide
the end user with enough meta-data to formulate their own conclusions on
the potential robustness of the model outputs.
Where there are confounding factors, the model can also use conservative measures to improve the potential accuracy of the model (De Ionno
et al., 2006). Where possible, the factors included in the model should
be sourced directly from individual farmers or national aquaculture
Bio-Economic Modeling of Fish Welfare Interventions in Aquaculture
313
Downloaded by [80.200.203.7] at 01:24 14 January 2013
surveillance and monitoring programs. This can increase the industrial
applicability of the model, and means its findings can be directly translated
into each aquaculture sector. However, if there are problems with data
reporting, from the farm level upwards, or if the model requires data that
is poorly reported at the national level e.g., average farm size, this can
reduce the models’ scope and efficacy.
Another limitation of multi-disciplinary models can be the lack of understanding within and between researchers from each discipline and the
potential end user. This can be remedied through transparency and clear
documentation throughout the modeling process, allowing the end user
to audit the model and make an informed decision on its applicability to
their chosen case study (see Scholten et al., 2007 for an in-depth summary).
Future Applications
The multi-disciplinary nature of the model allows its use in numerous
other fields and disciplines. For example, it can be used to model the efficacy of welfare interventions in other species and industries e.g., dairy cattle
or agricultural meat production. It can also be used to model the effects of
other non-welfare-related interventions upon the productivity of various
aqua- and agricultural ventures. It can also be used in many other avenues
beyond food production, such as epidemiology, sustainability and water
management. All each discipline requires is a quantifiable problem and a
robust operational intervention to reduce or alleviate it.
The decision tool can also be tailored to fit the needs of numerous
end-users. Within aquaculture, it can be used by farm managers and senior
management personnel to model the effects of a stated welfare intervention
upon different farms within their company. National surveillance and monitoring data is therefore not required to help the end user make an informed,
balanced decision on whether to apply a stated welfare intervention. The ability to model numerous options e.g., implement a welfare intervention on (i)
all farms, (ii) some farms or (iii) no farms means members of stakeholder associations or co-operatives can test the efficacy of different levels of implementation and can potentially adjust this for different regions within their
aquacultural sector. Policy makers and legislators can also use this tool to help
them produce industrially achievable guidelines for fish welfare, and base
these guidelines upon robust operational data from the entire value chain.
ACKNOWLEDGMENTS
This project was funded by the European Commission’s FP6 research
programme ‘‘Scientific Support to Policies’’ project No. 044118. It does
314
C. Noble et al.
not necessarily reflect its views and in no way anticipates the Commission’s
future policy in this area. The authors would like to thank all members of
the BENEFISH Industrial Advisory Group, especially Alan Dykes of Lighthouse Caledonia and Ian Michie of the Findus Group for providing advice
and a valuable industrial perspective throughout the BENEFISH project.
Downloaded by [80.200.203.7] at 01:24 14 January 2013
REFERENCES
Berrill, I.K., S. Kadri, K. Ruohonen, M. Kankainen, B. Damsgård, H. Toften, . . . J.F. Turnbull (2009)
BENEFISH: A European project to put a cost on fish welfare actions. Fish Veterinary Journal, 11,
23–38.
Berrill, I.K., C.M. MacIntyre, C. Noble, M. Kankainen, & J.F. Turnbull (2012) Bio-economic costs and
benefits of using triploid rainbow trout in aquaculture: Reduced mortality. Aquaculture Economics
& Management, 16(4), 365–383.
De Ionno, P., G. Wines, P. Jones, & R. Collins (2006) A bioeconomic evaluation of a commercial scale
recirculating finfish growout system—An Australian perspective. Aquaculture, 259(1=4), 315–327.
Ellis, T., B. Oidtmann, S. St-Hilaire, J.F. Turnbull, B. North, C. MacIntyre, . . . T. Knowles (2008) Fin
erosion in farmed fish. In: Fish Welfare (ed. E. Branson), pp. 121–149. Blackwell, Oxford, UK.
Fraser, D. (2003) Assessing animal welfare at the farm and group level: the interplay of science and
values. Animal Welfare, 12, 433–443.
Frewer, L., A. Kole, M.A. van der Kroon, & C. de Lauwere (2005) Consumer attitudes towards the
development of animal-friendly husbandry systems. Journal of Agricultural and Environmental Ethics,
18, 345–367.
Honkanen, P., & S.O. Olsen (2009) Environmental and animal welfare issues in food choice - the case of
farmed fish. British Food Journal, 111, 293–309.
Kankainen, M., I.K. Berrill, C. Noble, K. Ruohonen, J. Setälä, A. Kole, . . . J.F. Turnbull (2012) Modeling
the economic impact of welfare interventions in fish farming—A case study from the UK rainbow
trout industry. Aquaculture Economics & Management, 16(4), 315–340.
Lassen, J., M. Gjerris, & P. Sandøe (2006) After Dolly—Ethical limits to the use of biotechnology on farm
animals. Theriogenology, 65, 992–1004.
Olesen, I., F. Alfnes, M.B. Røra, & K. Kolstad (2010) Eliciting consumers’ willingness to pay for organic
and welfare-labelled salmon in a non-hypothetical choice experiment. Livestock Science, in press.
doi:10.1016=j.livsci.2009.10.001.
Olson, R.J., J.M. Briggs, J.H. Porter, G.R. Mah, & S.G. Stafford (1999) Managing data from multiple
disciplines, scales, and sites to support synthesis and modeling. Remote Sensing Environment, 70,
99–107.
Piedrahita, R.H. (1988) Introduction to computer modeling of aquaculture pond ecosystems. Aquaculture and Fisheries Management, 19, 1–12.
Savory, C.J. (2004) Laying hen welfare standards: A classic case of ‘power to the people’. Animal Welfare,
13, 153–158.
Scholten, H., A. Kassahun, J.C. Refsgaard, T. Kargas, C. Gavardinas, & A.J.M. Beulens (2007) A
methodology to support multidisciplinary model-based water management. Environmental
Modeling & Software, 22, 743–759.
Suzuki, Y., T. Otaka, S. Sato, Y.Y. Hou, & K. Aida (1997) Reproduction related immunoglobulin changes
in rainbow trout. Fish Physiology and Biochemistry, 17, 415–421.
Sveier, H., & E. Lied (1998) The effect of feeding regime on growth, feed utilisation and weight
dispersion in large Atlantic salmon (Salmo salar) reared in seawater. Aquaculture, 165, 333–345.