Food for Thought
Advice under uncertainty in the marine system
Dorothy J. Dankel 1*, Robert Aps2, Gurpreet Padda3, Christine Ro¨ckmann4, Jeroen P. van der Sluijs 5,
Douglas C. Wilson6, and Poul Degnbol 7
1Institute of Marine Research, PO Box 1870, N-5817 Bergen, Norway
2Estonian Marine Institute, 14 Ma¨ealuse Street, EE-126 18 Tallinn, Estonia
3Department for Environment, Food and Rural Affairs, Nobel House, Area 2D, 17 Smith Square, SW1P 3JR London, UK
4Wageningen IMARES, PO Box 68, NL-1970 AB IJmuiden, The Netherlands
5Utrecht University, PO Box 80125, NL-3508 TC Utrecht, The Netherlands
6Department of Development and Planning, Innovative Fisheries Management, Aalborg University, Fibigerstræde 13, DK-9220 Aalborg Ø, Denmark
7ICES, H.C. Andersens Boulevard 44-46, DK-1553 Copenhagen V, Denmark
*Corresponding Author: tel: +47 55 23 85 56; fax: +47 55 23 86 87; e-mail: dorothy@imr.no
Dankel, D. J., Aps, R., Padda, G., Ro¨ckmann, C., van der Sluijs, J. P., Wilson, D. C., and Degnbol, P. 2012. Advice under uncertainty in the marine
system. – ICES Journal of Marine Science, 69: 3–7.
Received 4 February 2011; accepted 10 October 2011; advance access publication 22 November 2011.
There is some uncertainty in the fisheries science–policy interface. Although progress has been made towards more transparency and
participation in fisheries science in ICES Areas, routine use of state-of-the-art quantitative and qualitative tools to address uncertainty
systematically is still lacking. Fisheries science that gives advice to policy-making is plagued by uncertainties; the stakes of the policies
are high and value-laden and need therefore to be treated as an example of “post-normal science” (PNS). To achieve robust govern-
ance, understanding of the characteristics and implications of the scientific uncertainties for management strategies need to come to
the centre of the table. This can be achieved using state-of-the-art tools such as pedigree matrices and uncertainty matrices, as devel-
oped by PNS scholars and used in similar science–policy arenas on other complex issues. An explicit extension of the peer community
within maritime systems will be required to put these new tools in place. These new competences become even more important as
many countries within the ICES Area are now embarking on new policies.
Keywords: fisheries advice, fishery system, maritime system, pedigree matrix, post-normal science, uncertainty.
Introduction
The Working Group on Fisheries Systems (WGFS) was established
by the International Council for the Exploration of the Sea (ICES)
10 years ago, inter alia “to develop a framework and methodology
for the analysis of fishery system performance” and “propose . . .
interdisciplinary research which will advance ICES future capabil-
ity in fishery systems analysis” (ICES, 2000).
The WGFS created a forum to bring social scientists into ICES
to help describe the socio-ecological system around fisheries and
to use this perspective to intensify ICES’ effectiveness in marine
science advice for policy. The main conclusions from 10 years of
interdisciplinary WGFS work/expertise are that many interests
should be represented when science and policy meet and that an
understanding of the characteristics and the implications of scien-
tific uncertainty (including data quality) need to be placed at the
centre of the discussion (ICES, 2008b). The latter is seen as the
key to governance in the science–policy interface of complex
issues. Others have argued, for good reason, that overemphasizing
uncertainty in fisheries advice can lead to policy paralysis
(Rosenberg, 2007). However, broader experience in science and
policy indicates that underemphasizing uncertainty is even more
dangerous, because it can do lasting damage to the credibility of
the science (Keepin and Wynne, 1984; Kloprogge and van der
Sluijs, 2006; van der Sluijs, 2007; van der Sluijs et al., 2008). An
effective approach for dealing with science in situations of high
stakes and high systems uncertainty is through “post-normal
science” (PNS), as developed by Funtowicz and Ravetz (1993).
Succinctly, the concept of PNS—as opposed to “normal
science”—suggests that in situations of high uncertainty and
high stakes, imperfect (and sometimes subjective) knowledge
needs to be used in providing advice to policy-makers. An import-
ant pillar of PNS is the inclusion of an extended peer community
# 2011 International Council for the Exploration of the Sea. Published by Oxford University Press. All rights reserved.
For Permissions, please email: journals.permissions@oup.com
ICES Journal of
Marine Science
ICES Journal of Marine Science (2012), 69(1), 3–7. doi:10.1093/icesjms/fsr179
at U
niversity of California, Santa Cruz on D
ecem
ber 19, 2011
http://icesjms.oxfordjournals.org/
D
ow
nloaded from

(EPC), including stakeholders with different types of expertise.
These EPCs need to acknowledge, analyse, and communicate
uncertainty and quality in science for policy in the extended
peer-review process (Funtowicz and Ravetz, 1990, 1993; van der
Sluijs, 2002; Kloprogge and van der Sluijs, 2006; Petersen, 2008).
Hence, the EPCs need to become the foundation for credible,
legitimate, and salient science for policy advice. Cash et al.
(2003) found these three variables to have the most influence on
the extent to which a scientific result will be incorporated in
policy. The PNS approach characterizes a method of enquiry for
situations in which decisions need to be made before conclusive
scientific evidence is available (Funtowicz and Ravetz, 1990,
1993). Often, in fact, a single scientific answer will never be avail-
able for complex systems such as fisheries. In such cases, more
research does not lead to less uncertainty, but can lead instead
to unforeseen complexities (van der Sluijs et al., 2005, 2010;
Trenberth, 2010). Concurrently, the potential impacts of decisions
based on uncertain science have very large consequences (biologic-
al and/or social), so values are in dispute.
We highlight two examples from outside fisheries science to
illustrate the importance of advancing uncertainty early and in a
transparent manner. The first is the recent “Climategate” contro-
versy from late 2009 (van der Sluijs et al., 2010), which originated
at the Climatic Research Unit at the University of East Anglia
(UK), after a hacker published e-mails and files from CRU scien-
tists on the Internet. The media coverage that followed instigated
an independent review panel of the CRU (Oxburgh et al., 2010).
The e-mail leak exposed the decision by some climate scientists
to exclude a specific (tree ring) dataset from the historical
climate reconstruction presented in the latest IPCC report.
Although there are good scientific reasons not to include the
data post-1960 (Briffa et al., 1998), and although these reasons
are widely accepted as valid in the scientific community working
on temperature reconstructions from the past, this was not
made clear in the final report of the IPCC, nor did the report
mention that these data were excluded from the climate recon-
structions presented. This prompted climate sceptics to question
the credibility and legitimacy of the report, which is based on sci-
entific consensus within the IPCC. Consecutive independent
reviews of both the science (Oxburg et al., 2010) and the process
(Russell, 2010) concluded that the scientific approach had been
justified, proper scientific process such as peer review had been
in place, but the problem had been elsewhere, because “there has
been a consistent pattern of failing to display the proper degree
of openness” (Russell, 2010). The case demonstrates how oversel-
ling certainty creates vulnerability in the credibility and legitimacy
of the scientific basis for policy. Such vulnerabilities can and will be
exploited easily, to obstruct and delay policy intervention (van der
Sluijs et al., 2010).
An example of a severe scientific credibility crisis of advice for
policy support is the controversy originated by a whistle-blower at
the Netherlands Environmental Assessment Agency in early 1999
(van der Sluijs, 2002). The media was the last way out for senior
statistician de Kwaadsteniet who decried his institution for using
poorly validated computer models as the scientific basis for
advice given to the Dutch government for far-reaching environ-
mental policy decisions. De Kwaadsteniet was also critical that
the agency’s advice was presented as point values, with spurious
precision and opaque uncertainties (van der Sluijs, 2002). A
6-month credibility-ravaging media storm fuelled vehement
debate in parliament on the credibility and validity of
environmental numbers that form the basis of Dutch environmen-
tal policy. It ultimately resulted in rigorous reforms of the quality-
control procedures inside the Netherlands Environmental
Assessment Agency and in how scientific evidence is analysed,
reviewed, and communicated in their assessments, and a new
focus that copes with uncertainty (van der Sluijs et al., 2003;
Beck, 2007; Petersen et al., 2011).
These dangers from overselling certainty are relevant to ICES,
because ICES also communicates results from uncertain and
imperfect data. Most ICES assessment scientists have experienced
being asked to find certainty that is not really there: the achievable
state of knowledge does not allow one to deliver the degree of cer-
tainty that policy-makers seem to expect from science (Wilson,
2009; Kraak et al., 2010). Even when the uncertainties in science
are presented through caveats in the advice, policy-makers have
little choice but to take and use the uncertain numbers. Under
the current Common Fisheries Policy (CFP), policies do neglect
uncertainties surrounding problem-framing, model structures,
assumptions, system boundaries, indeterminacies, and the extent
to which a policy is value- (or theory-) laden (Hauge, 2010; van
der Sluijs et al., 2010).
Tools to address uncertainty
One tool to address unquantifiable uncertainties is the numeral
unit spread assessment pedigree (NUSAP) analytical and nota-
tional system (van der Sluijs et al., 2005). It extends the classic
notational system for quantitative scientific information (usually
provided as a number, a unit, and a standard deviation) with
two additional qualifiers: expert judgement of the reliability (the
assessment) and a multicriteria characterization reflecting the
origin and status of the information (the pedigree). The classical
notational system does not reveal the distinction between nearly
perfect information (such as the speed of light) and highly imper-
fect information (such as the size of a marine fish stock). The two
additional qualifiers, assessment and pedigree, attempt to remedy
this problem. The pedigree analysis is a qualitative structural
process to clarify the knowledge base on which scientists and sta-
keholders frame their perceptions of a problem, by appraising the
information underpinning the numbers and theories that form the
basis of scientific advice, often model-derived. In PNS, the trad-
itional search for robust scientific findings, ideally based on scien-
tific consensus, is replaced by a search for robust policy strategies,
which are useful regardless of which of the diverging scientific
interpretations of the knowledge is correct (van der Sluijs et al.,
2010). The qualitative approach of pedigree analysis helps to
assess different aspects of the knowledge base, such as its empirical
basis, the level of theoretical understanding, the rigour of the sci-
entific methods used, the extent to which the findings have been
validated, and the extent of scientific consensus among peers
and among the wider scientific and stakeholder communities
(van der Sluijs et al., 2005). An example of a pedigree matrix is pre-
sented in Table 1. Results from sensitivity analysis and pedigree
analysis can be combined in a so-called diagnostic diagram that
aims to reveal the weakest, i.e. the most uncertain, elements of a
scientific assessment or a model (van der Sluijs, 2005). It is
based on the notion that neither sensitivity alone nor pedigree
alone is a sufficient measure for whether uncertainty is critical
for the outcome of an assessment. For example, if the spread in
a model parameter has a negligible effect on model output (low
sensitivity), the robustness of model output to parameter uncer-
tainties could be good even if uncertainty around that parameter
4 D. J. Dankel et al.
at U
niversity of California, Santa Cruz on D
ecem
ber 19, 2011
http://icesjms.oxfordjournals.org/
D
ow
nloaded from