The Glasgow consensus on the delineation between pesticide

advertisement
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Supporting Information:
The Glasgow consensus on the delineation between pesticide emission inventory
and impact assessment for LCA
Ralph K. Rosenbaum1,2, Assumpció Anton3, Xavier Bengoa4, Anders Bjørn1, Richard Brain5, Cécile Bulle6,
Nuno Cosme1, Teunis J. Dijkman1, Peter Fantke1, Mwema Felix7, Trudyanne S. Geoghegan8, Bernhard
Gottesbüren9, Carolyn Hammer10, Sebastien Humbert4, Olivier Jolliet11, Ronnie Juraske12,13, Fraser Lewis14,
Dominique Maxime6, Thomas Nemecek15, Jérôme Payet16, Kati Räsänen17, Philippe Roux2, Erwin M. Schau18,
Sandrine Sourisseau19, Rosalie van Zelm20, Bettina von Streit21, Magdalena Wallman22
1
Technical University of Denmark, Department of Management Engineering, Produktionstorvet, Building
426, 2800 Kgs. Lyngby, Denmark
2
IRSTEA, UMR ITAP, ELSA-PACT – Industrial Chair for Environmental and Social Sustainability Assessment,
361 rue Jean-François Breton, BP 5095, F-34196 Montpellier Cedex 5, France
3
IRTA, Ctra Cabrils km 2, 08348 Cabrils, Barcelona, Spain
4
Quantis, EPFL Innovation Park, 1015 Lausanne, Switzerland
5
Syngenta Crop Protection LLC. 410 Swing Rd. Greensboro, 27419 NC, USA
6
CIRAIG, Department of Chemical Engineering, P.O. Box 6079, École Polytechnique de Montréal (Qc),
H3C3A7, Canada
7
Tropical Pesticides Research Institute, P. O. Box 3024, Arusha, Tanzania
8
University of Otago, Department of Chemistry PO Box 56 Dunedin 9016 New Zealand
9
BASF SE; Agricultural Products, 67117 Limburgerhof, Germany
10
Environmental Resources Management Ltd, Eaton House Wallbrook Court North Hinksey Lane, OX2 0QS
Oxford, UK
11
University of Michigan, School of Public Health, Department of Environmental Health Sciences, Ann Arbor,
MI 48109, USA
12
Group for Ecological Systems Design, Institute of Environmental Engineering, ETH Zurich, 8093 Zurich,
Switzerland
13
Dr. Knoell Consult GmbH, 68165 Mannheim, Germany
14
Syngenta, Jealott's Hill International Research Centre, Bracknell, RG42 6EY Berkshire, UK
15
Agroscope, Institute for Sustainability Sciences, CH-8046 Zurich, Switzerland
16
Cycleco, 1011 Avenue Léon Blum, 01500 Ambérieu-en-Bugey, France
17
MTT Agrifood Research Finland, 31600 Jokioinen, Finland
18
European Commission, Joint Research Centre, Institute for Environment and Sustainability, 21027 Ispra,
Italy
19
Veolia Environnement Recherche & Innovation, Centre de Recherche-Maisons-Laffitte, Chemin de la
Digue B.P. 76, 78603 Maisons-Laffitte Cedex, France
20
Radboud University Nijmegen, Department of Environmental Science, Institute for Water and Wetland
Research, 6500 GL Nijmegen, The Netherlands
21
Bayer Technology Services GmbH, BTS-TD-TI-Sustainability Consulting, Leverkusen, Germany
22
SIK – Swedish Institute for Food and Biotechnology, Box 5401, SE-402 29 Göteborg, Sweden
Workshop schedule
08:00 - 08:30
Arrival of participants and tea/coffee
1
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
08:30 - 09:00
Welcome and introduction to the workshop (Ralph Rosenbaum, DTU)
09:00 - 10:30
Input presentations for the discussion (ca. 20 min each incl. Q&A):
- A LCA practitioner perspective: Current issues when conducting LCA including pesticide
application (Sebastien Humbert, Quantis)
-
A LCI (emission inventory) modelling perspective: Pesticide emission modelling in
PestLCI 2.0 (Teunis Dijkman, DTU)
- A LCIA (impact assessment) modelling perspective: Impact pathways of pesticides and
their current modelling in LCIA (Peter Fantke, DTU; Ronnie Juraske, ETHZ)
- A case-study covering LCI and LCIA for pesticides (Rosalie van Zelm, Radboud University;
Philippe Roux, IRSTEA/ELSA)
10:30 - 11:00
Tea/coffee break
11:00 - 12:00
Plenary discussion (moderator Ralph Rosenbaum)
12:00 - 12:30
Definition of 2-3 break-out groups and subjects
12:30 - 13:30
Lunch (all participants are invited to a buffet)
13:30 - 16:00
Discussion in break-out groups
16:00 - 16:30
Tea/coffee break
16:30 - 18:00
Presentations from the break-out groups and discussion
18:00 - 19:00
Wrap up and next steps (moderator Ralph Rosenbaum):
19:00
-
Is a 2nd workshop needed or have we reached a publishable consensus?
-
What’s next? Who does what?
Closure of the workshop
On-site participants
Name
Affiliation
Country
Assumpció Anton
IRTA
Spain
Anders Bjørn
Technical University of Denmark
Denmark
Richard Brain
Syngenta
Switzerland
Cécile Bulle
CIRAIG
Canada
Nuno Cosme
Technical University of Denmark
Denmark
Teunis Dijkman
Technical University of Denmark
Denmark
Peter Fantke
Technical University of Denmark
Denmark
Trudy Geoghegan
University of Otago
New Zealand
Bernhard Gottesbueren
BASF
Germany
Carolyn Hammer
Environmental Resources Management (ERM)
UK
Israel Herrera
CIEMAT
Spain
Sébastien Humbert
Quantis
Switzerland
Olivier Jolliet
University of Michigan
USA
Ronnie Juraske
ETHZ
Switzerland
2
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Fraser Lewis
Syngenta
Switzerland
Thomas Nemecek
Agroscope
Switzerland
Jérôme Payet
CYCLECO
France
Kati Räsänen
MTT Agrifood Research Finland
Finland
Ralph Rosenbaum
Technical University of Denmark
Denmark
Philippe Roux
IRSTEA
France
Alison Sapiets
Syngenta
Switzerland
Erwin Schau
EC Joint Research Centre
EU
Sandrine Sourisseau
Veolia Environment Research & Innovation
France
Rosalie van Zelm
Radboud University
The Netherlands
Bettina von Streit
Bayer
Germany
Magdalena Wallman
Swedish Institute for Food and Biotechnology
Sweden
Name
Affiliation
Country
Vlad Coroama
Universidade de Coimbra
Portugal
Mwema Felix
Tropical Pesticides Research Institute
Tanzania
Filipa Figueiredo
Universidade de Coimbra
Portugal
Fausto Freire
Universidade de Coimbra
Portugal
Clare Howard
International Nitrogen Initiative (INI) / GPNM
Global
Joe Lane
University of Queensland
Australia
João Malça
Universidade de Coimbra
Portugal
Carolina Passeira
Universidade de Coimbra
Portugal
Remote participants
Workshop minutes
08:40 Welcome and introduction to the workshop (by Ralph Rosenbaum)




Programme and presentations to come;
Objectives of the workshop and the focus on achieving a consensus;
Target groups;
Expected results.
Points for discussion





Classification of the ecosystem in the agricultural field;
Double counting from land use and ecotoxicity;
Are the recommendations applicable to all emissions (pesticides, fertilizers and plant growth
regulators)?
The relevance of the emissions compartments;
The relevance of the impact pathways.
09:50 Presentation: A LCA practitioner perspective: Current issues when conducting LCA including
pesticide application (by Sébastien Humbert – Quantis)
3
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Topics covered



LCA practitioner perspective;
Current issues when conducting an LCA, including intended application:
– Should be practical and efficient solutions (time and resource wise);
– Fertilizers – should they be allocated to soil or to water compartment?
– Pesticides – what kind of compartments should be included in an assessment?
Quantis perspective on LCI/LCIA fractions;
Default values should be applied to generic conditions and model should be flexible to adapt values to local
conditions.
Questions and comments
Participant A: Why modelling the impacts, if by following regulations no thresholds are crossed?
Participant B: That is the approach in regulatory Environmental Risk Assessment (ERA), but in Life Cycle
Assessment (LCA) impacts are calculated even if they fall below threshold values, in order to compare
different product systems for their relative impact potentials. It is not about assessing the impact from one
single compound in defined conditions, but from the combined impact of a mix of compounds resulting
from the activity.
Participant C: Are the impacts from fertilizer application always negative?
Participant D: It is difficult to assess positive impacts on biodiversity for example (some species may benefit
while others are harmed by pesticide or nutrient applications, hence an increase in present species or
individuals per species may – or may not – lead to losses of other species in the same ecosystem), but these
are out of the scope of this meeting. The focus is currently on the negative impacts on emissions. Positive
impacts may come to be included in the future.
Participant E: How should the differences be reflected in practice? Should inventory data be based on best
practice or average practice? (open question)
09:50 Presentation: A LCI (emission inventory) modelling perspective: Pesticide emission modelling in
PestLCI 2.0 (by Teunis Dijkman – DTU)
Topics covered




LCI perspective;
Compared three available LCI approaches: Ecoinvent, USLCI, PestLCI;
Conclusions: Different LCI approaches and different LCI boundaries lead to non-comparable LCA
studies;
Clear LCI/LCIA boundaries should be defined consistently for all emissions and LCIA impact
categories.
Question and comments
4
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Participant B: LCI and LCIA system boundaries should be aligned. When choosing a particular LCI model one
needs to look for an LCIA model with compatible boundaries – and vice versa.
Participant F: Field cannot be classified as either ecosphere or technosphere. On the one hand it is highly
manipulated and on the other the government subsidizes environmental protection of fields.
Participant B: However, we have different regulations for agricultural fields compared to what we call
ecosphere.
Participant G: PestLCI is a fast tool when used to compute, but gathering input data takes time!
Participant A: I have tested PestLCI and it is OK if you work with predefined scenarios, but If I am testing
different pesticides I would like to have the opportunity myself to include them instead of running it via
DTU.
Participant H: Unfortunately that is a disadvantage of the software program we chose. New compounds can
only be added if users use Analytica software or by requesting Teunis Dijkman or Morten Birkved to be
included and sent back. We can do it pretty quickly but it would be better if the users could do that by
themselves.
Participant A: So the access to the software and database is the key point for dissemination, distribution
and acceptance of PestLCI.
Participant I: In your example you said that 98% is taken up or degraded in the system. From my
perspective if I want to calculate concentrations or intake fractions for further evaluation of human toxicity
I would need to have other factors (intersections) and if I don’t use your model I would get different results.
In my view there is some inconsistency there, comparing results from someone using your model with the
intersections fractions that you derived, and others (the dynamiCrop plant uptake model, for instance). An
important step would be the harmonisation/coupling of existing models.
10:00 Presentation: A LCIA (impact assessment) modelling perspective: Impact pathways of pesticides
and their current modelling in LCIA (Peter Fantke – DTU; Ronnie Juraske – ETHZ)
Topics covered
System boundaries in LCIA:



Focus on translating LCI outputs (emissions inventories) into impacts;
Focus on three areas of protection: Human health, ecosystem health, resources;
We should account for: Substances, dimension (time and space), input and outputs (mass balance)
to avoid double counting.
Questions and comments
Participant J: Just a comment I forgot to mention from the practitioner’s perspective – in milk and meat
production the manure (as an organic fertilizer) should be included in the inventory, considering the cow as
5
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
part of the technosphere emitting manure to the field as ecosphere, and characterized from there on as
nitrogen-based pollutants.
Participant A: What exposure pathways should be included in the model?
Participant B: An initial screening decides which pathways are negligible, and then addressing the most
relevant ones.
Participant K: A lot of this data is already available, so methods for human exposure, particularly for
pesticides, for all the different groups - for operators, for bystanders, for consumers, information is already
generated and already available so why would we go to an additional model, particularly as practitioners,
when this data can be got freely and published in websites.
Participant B: To my knowledge data are NOT available – what is out there is that pesticides are safe to use,
but from that information and residues in the plant I cannot conclude on what chemical would be a better
choice to be incorporated into an overall production system, where there are hundreds of other chemicals.
I cannot draw a conclusion from that kind of conservative assumption assessment based on arbitrary
thresholds like maximum residues, and that is why we remodel it, simplify it, globalize it, average it based
on best estimates rather than conservative assumptions to make it comparable – this is actually the basis of
a comparative assessment and therefore very different from Risk Assessment.
Participant K: How does LCA handle the modelling of impacts on different species?
Participant B + Participant L + Participant J + Participant M: Within ecotox category species are equally
weighted in geometric mean function based on EC50. Between impact categories weighting and
normalization is applied. In theory most important species for ecosystem functioning should receive higher
weight, but this is impractical since it is very data demanding.
10:30 Presentation: A case-study covering LCI and LCIA for pesticides (by Rosalie Van Zelm - Radboud
University; Philippe Roux - IRSTEA/ELSA)
Topics:




Gap between LCI and LCIA for toxicological impacts from pesticides;
Banana production case study;
Limitations: inconsistencies in LCI, risk of overlapping with LCIA due to inadequate space and time
modelling, non-exhaustive models and mass balances;
Compared different LCI approaches and showed how fractions to environmental compartments
varied.
Questions and comments:
Participant I: The approach that comes closer to yours is Ecoinvent approach with same order of magnitude
without including uncertainty, so I wonder why so much hassle about producing a new method.
Participant E + Participant I: The banana case study is special because the applied pesticides (either by
plane or backpack sprayers) will end up in the peel and not in the pulp, hence relatively low human health
6
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
impact related to ingestion of fruits. Bananas are covered with bags (impregnated with pesticides) during
growth. This issue was not covered at all in the case study which makes this case study problematic.
Participant D: I remind you all that we should focus today on finding not the highest achievable precision
but to make it operational, based on parsimony to get a fairly good solution. Different approaches still need
to be explored and their influence on results quantified.
- Tea/coffee break 11:00: Plenary discussion
Topic launched by Ralph Rosenbaum: Ecosphere/technosphere boundary and temporal distinction - how to
reflect differences in practice.
Participant J: For me, the agricultural field is part of the ecosphere, as it is something we are trying to
protect and leave to the following generations. Otherwise unethical, since people enjoy ‘nature’ in fields,
and we are using something from nature (crops, fruits, etc.) and only giving it a help by applying fertilizers
and pesticides. With the exception of greenhouses, that should be technosphere. But let’s take this opinion
as a discussion proposal.
Participant M: Crop fields should be considered as the same as greenhouses, therefore technosphere, but
there is no single A or B option to choose from. Some fields have no ‘natural value’ for humans.
Participant I: starting from the endpoints we have human toxicity, freshwater ecotoxicity, terrestrial
ecotoxicity, but we actually miss the category of soil quality (to assess the field) and that might be the key
issue in why we have this problem drawing the boundary in the first place – in a perspective of the field as a
resource. Not possible to answer directly if it belongs to ecosphere or technosphere.
Participant F: Temporal aspects vary a lot from process to process. This makes the distinction between
ecosphere and technosphere in LCI/LCIA very difficult and we should not consider the same boundary for
all impacts – different environmental processes might require different temporal boundaries.
Participant K: As an environmentalist I could not understand why we would apply the same kind of
protection criteria to an agricultural production system as to a pristine ecosystem, so the distinction
between technosphere and ecosphere is perhaps too simplistic, because by definition they will have
different protection goals. In this sense, the ecosystem service approach tries to categorize what the field is
trying to do, so things like soil quality in an agricultural field are important otherwise the service of
producing crops in the future would not happen so it is important to clearly understand what we are
generalizing in LCA and deciding what we are trying to evaluate in terms of impact. An absolute distinction
is very difficult to attain without actually considering what the purpose of the agricultural field is.
Participant D: We are not here to define a framework for something new, but rather to find a consensus on
what is already being done. The distinction is done on a daily basis by LCA practitioners and it is a limiting
factor we face (people are using it), so unless we come up with a better solution we have to cope with such
current practice, and here today we should aim at a standard that works with what is currently done.
7
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Participant A: Adding the temporal dimension to the problem, the technosphere soil becomes ecosphere
after one season. It should be back to the original state after one year, meaning that any compounds
remaining after one year should be counted as emissions to the ecosphere.
Participant L: Land use also plays an important role. We can consider agriculture as an invasive species
forcing a monoculture in an agriculture field (in comparison to what was the natural environment), being
helped by pesticide and fertilizers application. The problem is that there is still an ecosystem living there
(insects in the soil, for instance) and where do we account for the effect on that ecosystem? In the land use
indicators?
Participant K: So the difficulty here is the definition of “adverse” and what to compare it to? Simply by
growing a crop we are modifying the “natural” ecosystem even if we spray nothing on it. Because the
organisms that go in there would not be the organisms that would normally occur in that ecosystem
therefore they should not be issues of protection. Otherwise we would be double counting. So if we
account for that by having agriculture there I think double counting it again, by spraying a herbicide or
insecticide to control the pests, is not correct because many of those organisms would not be there if you
have not planted the crop in the first place - so what are we comparing it to? To the area around it, which is
completely different in terms of its habitat? If you have to live with the current approach and looking for a
simplistic approach and if land use is already accounting to what then you cannot help but to ignore any
subsequent impact within that field caused by pesticides because comparing to what happens off-crop
because that habitat is already completely changed from what was before and even the soil environment in
terms of its impact from introducing the crop is much higher than the impact from using an herbicide. So
avoiding the double counting is important in LCA in this sense.
Participant A: Natural vegetation in most of Europe is swamps and wood with limited biodiversity.
Agriculture therefore artificially introduces additional biodiversity compared to natural state.
Participant E: I think that in LCA we do not want to compare the agricultural field to a natural system. In the
future of LCIA we need to prevent double counting between land use and land use change and e.g.
terrestrial ecotoxicity.
Participant L: Double counting is only relevant for impact categories dealing with impacts on the field and
terrestrial ecotoxicity. There is not such a big problem for freshwater ecotoxicity.
Participant N: If we can just identify exactly what is similar between those different scenarios then that
could be our baseline and if you’re looking at a GMO crop versus a crop with traditional pesticide use then
whatever are the dissimilarities between them that would be the technosphere and everything else beyond
that would be what you would have to look for an impact – like anticipating potential scenarios of what
people might be using this for.
Participant O: In addition to where the technosphere ends and the ecosphere begin, there is also where
does modelling in LCI ends and where does modelling in LCIA begins. Even if you clearly define where the
limit is, we are still modelling part of the fate in the inventory and part in the impact assessment and those
are not fully consistent. Even defining the field as part of the technosphere you could define different types
of fields and model different ways, including the fate in the field in LCIA and only adding some inventory
flows that differ with agricultural practices.
8
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Participant C: Most likely there are no natural environments between different agricultural fields. So
emissions from fields end up again in the technosphere (another field). Why include soil as the
technosphere, when it is sprayed accidentally?
Participant G: Sometimes soil is meant for spraying, e.g. for herbicides.
Participant B: Side remark: We should focus on the current practice – practitioners already use
technosphere/ecosphere boundaries, inventory modelling, and impact modelling, so this should be the
starting point. To me it would be beneficial in the long-run if we respect what is being proposed today and
be potentially implemented sooner or later into current practice in LCA. Not exclusively focus on what is
there and what has been done and defined and then when we have reached a consensus it will not be
applicable once we implement something else. So look into the future of what LCIA should look like and
make sure this is valid in the long term.
Participant D: I would also subscribe to that comment. It is important to consider what future LCI and LCIA
may look like, because we don’t want to come up with a consensus that will be out-dated in three years
basically. It is also important to consider new pathways or new impact categories that we know that are
developed scientifically and might be implemented in the future and we should consider that as well.
Participant J: From a practical perspective I would feel that people when they do LCA, the analysis or
practitioner, we asked them to do an inventory and then this inventory is multiplied automatically by CFs
(the impact assessment). The LCIA can be expected to be done by experts in that field. The inventory
sometimes is done by experts but often is done by someone who is not an expert in that specific
technology. I would say that if you say that the pesticide is emitted into the field, how much is end up in air
and water, then that gets multiplied by a CF and if the impact assessment, or the philosophy behind the
impact assessment, is that we don’t want to double count the impact on the field you could actually have in
your modelling of the CF taken off the impacts of the field, so if you emit a certain amount to the field it
takes for example 10 minutes to go off by the air and so you could take the impact of these 10 minutes
from the model and then you can start to count the impact only outside the field and avoiding double
counting the impact in the field. However, you could still have the connection between the inventory and
the CF at the emission level in the field even if you do not want to count the impact in the field. As I see it,
this would be the most practical way for practitioners and with this approach you leave the freedom of the
impact assessment to still advance on that without having to solve that today. Advances in what are
defined as an impact area and not, and also to find out where is the emissions taking place and not asking
the inventory modeller to know how the pesticide evolves in the field.
Participant D: Would you still include the impact to the field (terrestrial ecotoxicity, etc.) after a certain
time?
Participant J: I would leave that to the freedom of the impact assessment developer (ReCiPe, CML, USEtox,
or any other method) to make scenarios to choose the approach and not for us as a group to decide that
for all cases and would not leave it to the inventory modeller.
Participant D: Meaning not fixing the temporal boundary between the inventory and the impact
assessment?
9
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Participant J: Exactly.
Participant I: That means defining time-dependent CFs?
Participant J: For example if you say I do not want to account for the impact on my field from my herbicide
because it is already included in the land occupation impact category, as a practitioner I know that I have
emitted a certain amount of a chemical to that field but you as a USEtox CF developer you tell me if that
chemical amount will have so much impact in the way that the impact is excluded from the field for which I
emitted it.
Participant L: There is a paradox that when the environment is completely degraded, then additional
emissions count as zero impact because there is nothing to protect and only land use is to be considered. In
reality this is not the case and something always live in the field and therefore there is something to
protect. We end up defining whether the impact is modelled in the land use for a certain practice or does it
makes sense to look practice of the herbicide use at the biodiversity of what still remains and how do we
impact it.
Participant K: The key problem is what questions are we trying to ask? Sometimes crops should be included
in the ecosphere and other times in the technosphere (if it is fixed in the functional unit).
Participant H: Soil quality can be included in the technosphere, since it is needed for the biological factory,
like in PestLCI. It is important to keep the soil quality, it is a value to keep it productive, and therefore when
it is important for productivity it should also be a subject of protection.
Participant D: The information that should be gathered is dependent on the question to be answered, so
maybe it is not feasible to establish a single definition (or a recommendation, because it is too complex),
but one varying with the application (or clusters of applications). On the other hand, if there are no
recommendations then we will continue to see huge inconsistencies based on current practice.
Participant F: It always depends on the study and the goal of the study, so maybe you have to build
different approaches and adapt. In practice, choices are often made by the database designers and these
are not always transparent. We need a robust and practical approach which allows to be integrated in
databases and freedom to change according to different applications.
Participant M: Time issues in general were also discussed in the past and they are not specific for
agricultural use (it can also be discussed for any other land use). Time issues should be solved for all types
of land use (e.g. land use can change every year for agricultural use but not the same for industrial use that
can take 50 years), so the solutions that we might come up with should be also applicable for the other land
uses (and not only agricultural fields). Technosphere can always become ecosphere at some point in the
future.
Participant D: Land use is important, but so is toxicity. 1) Biodiversity on the field, with potential double
counting of biodiversity loss between land use and toxicity. 2) Biodiversity off field, where anything off the
fields is something we need to apply CFs and a related emission scenario. Do we need different boundary
definitions between different impact categories in order to avoid this double counting? So for land use the
field would always be technosphere, whereas for toxicity there should be a distinction?
10
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Participant J: We should not have different approaches to ecosphere/technosphere between impact
categories. This would be really difficult to operationalize in LCI databases.
Participant A and Participant K: Terminology confusion. In risk assessment Fate*Exposure*Hazard=Risk. In
LCA Fate*Exposure*Toxicity=Toxic Impact Potential.
Participant M and Participant D: Risk has to do with thresholds and in LCA we assume linearity. Therefore
we call it impact potential.
Participant O: Suggests using pre-determined CFs as impact potential per kg applied to field. This would
eliminate boundary issues avoiding different approaches by practitioners as it leaves that decision to the
LCIA methods.
Participant B: Promising, but not complying with ISO-standard, since inventory step (emissions from field)
would disappear as all the modelling would be at the LCIA phase.
Participant E: Also problematic when technologies advance or different parameterisation, then CFs need to
be recalculated and that computation could not be done by practitioners.
Participant J: From the practitioners perspective the inventory would refer to the application.
Participant L: The link between LCI and LCIA must be consistent and that is crucial. In that respect Ecoinvent
deals with application as the inventory input so I would favour leaving the modelling to LCIA (which is not
an emission, strictly speaking) but for land use ISO would have to be applied in a flexible way. As a
recommendation I would suggest that every inventory should report the amount applied and the mode of
application and that way would connect to any LCIA model.
Participant M: Another issue would be the emission of mixtures. In LCA the application of two substances is
supposed to be followed until they cross the boundary of the agricultural field and then multiplied by CFs –
so the modelling work has to be consistent for both substances to ensure consistency.
Participant D (on a side note about GMO): In the case of GMO crops (incorporation in the seed/plant) this is
not compatible with current practice. New sets of CFs would need to be developed due to the impact
pathways/mechanisms involved.
Participant B: Yes, because pesticides can be applied at different crop development stages (impacts to
human health and the environment) or even at post-harvest (impact only to environment). The term
‘connector’ is basically a fraction of distribution to another compartment in e.g. the dynamiCrop plant
uptake model.
Participant O: Both approaches (modelling substances at LCI or LCIA) already exist, so ‘just’ a matter of
combining models and calling the combination LCIA.
Participant D: Important to make default key parameters in models explicit so they can be adjusted if case
specific values are known.
12:30-13:30: Lunch break
11
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Participant M: With this proposed definition of technosphere/ecosphere boundary, should this discussion
be extended to e.g. waste water treatment plants?
Participant L: WWTP are for sure technosphere, because you clearly intervene before it is emitted via the
effluent.
Participant D (breaking in, and suggesting points to discuss in break-out groups): We need to agree on an
easy way for people to do the same (consensus):


Dimensions, time and space;
Key parameters that should be left explicit towards practitioners (for instance among the
parameters currently included in PestLCI 2.0).
Ralph Rosenbaum presented a proposal for framework, defining what different LCA stages should cover:
Inventory input (provided by the practitioner), inventory results, impact assessment (temporal aspects built
here).
Participant K (at the white paper board – Diagram 1): Pesticides application and characterisation of inputs
based on %s for air drift (e.g. 5%) and air to freshwater (e.g. 5%), crop (e.g. 70%), soil (% drainage + % runoff). So the mass to water at any given time and environment would be determined by those fractions over
the application plus a degradation or dissipation in the freshwater body. Some of the parameters would be
determined by fate and some by emission.
Diagram 1
5%
application
drift
volatil. 5%
90% remaining
surface
fresh water
A
25% soil
drainage %
70% crop
B
run-off %
C
Mass at any given time:
degradation
(A + B + C) - degradation


Participant H: This is basically what PestLCI modelling covers – the fractions emitted to the
environment from the first compartment and then the fate (degradation for instance) is dealt in the
LCIA model.
A discussion followed about the diagram in order to improve it and avoiding double counting on
fate processes.
12
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use

Location, type of crop and application are also important to define the fractions.
Participant L: There are three important points here: (1) total mass applied should be reported with
different conditions (documentation of the inventory), (2) there should be a clear view of the fractions
reaching the different environmental compartments and the sum should add 100%, (3) avoid double
counting on the fractions transferred.
Participant B: All fractions can only sum up to 100% under the assumption that degradation is not
relevant/not considered over the given time frame.
Participant M: Ensure that the delimitation of compartments is well documented assuring consistency
between methods (inventory and characterisation).
Participant K: Local practices should always be taken into account (Brazil example).
Participant J (in the white paper board – Diagram 2): Useful for practitioners - average percentages ending
up in “plant”, “soil”, “water”, “air”: default %s and other local/regional conditions for all the fertilizers
applied.
Diagram 2
Practical proposal
Application




1 kg
average
Dutch conditions
other conditions
plant
70%
50%
…
soil
10%
10%
…
water
10%
20%
…
air
10%
20%
…
These %s are based on average parameters and might be extended to supplementary times after
application if relevant for the crop, conditions, etc. to provide pre-calculated information to the
practitioner.
The term location can also mean scenario or site-specific conditions.
Agreement on the usefulness of tables such as the above for a practical application in LCA but
currently it is difficult because of the lack of definition of spatial boundaries to account for
emissions.
Practices and practice scenarios in different zones of the world are also important to define
different %s (or alternatively use default values to represent common average practice or best
practice scenario compared to not-best scenario).
13
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Participant P: Keep practices scenarios to practitioners in the LCI model and go for an average condition
called “good agricultural practices” as standard default - leave to the practitioner to decide if it is not the
case.
Participant J: LCA deals with a global default value scenario and then possibly per country scenarios and in
the future it is possibly easy to build wet or dry scenarios, etc.
Participant L: Then changes can be incorporated by reporting the differences from average conditions.
Participant K: There are different aspects to consider when defining scenarios. Is LCA really interested in
building so many different scenarios?
Participant J: LCA deals with a global default value scenario and if possible per country scenarios and this is
already a good resolution. If possible (and possibly easy to attain), build scenarios for wet and dry areas for
instance built on average scenarios (conditions) – e.g. average city, average country, average conditions,
and then specify further in the inventory when applicable.
Participant D: Regarding the time dimension – in LCI means the time of application and determines the
fractions, whereas in LCIA is the time frame for the fate modelling starting with the impacts to the
environment.
Participant B: The chemical type used also defines the time of application (default application times to
reduce the number of scenarios).
Participant C: Are the differences in good agricultural practices from developing state of the country
important? Will it make a difference?
Participant E: Yes, big differences should be expected.
Participant J: Often the country average practice is known for the inventory.
Participant D: Maybe using the location to calibrate the different fractions of the media distribution %s.
Participant J: What should we agree on when there is no “plant” compartment?
Participant B: Then don’t use that % and the unit flow is not characterized, acknowledging that the mass
balance is not complete (or report mass as loss). Like an elementary flow that is not characterised but
leaving the door open for method to be developed.





Inventory fate has a role in determining that amount, so it is recommended to include the plant
compartment to accommodate that %, and reporting flows that are not characterised if no plant
compartment is defined.
Depending on the inventory models – it will then determine the CFs to apply.
It is not feasible to exclude the plant compartment AND reporting the elementary flow – otherwise
it is assumed that all the pesticide ends in the soil rendering not effective at all.
Either the plant or the flow must be included to assure the mass balance is complete.
It should be a clear view of the mass balance. If the pesticide is degraded in the plant or transferred
to other media then it is loss.
14
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use

In conclusion the plant compartment should be included in the inventory modelling. Residues in
plant can then be further modelled or plant fraction re-distributed later into other
media/compartments.
A discussion arose on whether the metabolites should be addressed as well as sometimes they can be more
toxic than the applied parent compound. Current practice in LCA does not include by-products and fillers.
The conclusion is that LCA (and its toxicity methods) does not look into breakout products, so they are not
relevant for this discussion, although future research can be addressed to this issue.
Ultimately this discussion contributed to the topics in the PROPOSAL after working out wordings and
definitions.
Participant K: Regarding direct emissions after-spray drift, deposition is the only direct emission. Run-off
and drainage are further affected by fate.
Participant L: Time from application is important here as it defines short-term transport from where the
fate can take over.
Participant K stepped in to the white paper board (Diagram 3). The buffer zone dimension is dependent on
the application method but should be considered part of the on-field (technosphere)
Diagram 3
volatile fraction to air
A%
application
direct emissions
first steps
X% to plant
buffer zone
Y% to soil
Z% to water
part of infield
infield (technosphere)
First steps
Second steps -> FATE
X% plant
Y% soil/water onfield
Z% to sfw
A% to air
drainage
Run-off
degradation
volatilisation
surface
fresh water
off-field
Participant A: suggested inclusion of a wash-off from plant component. Additionally, emissions are defined
as for the soil surface – top soil (whichever the depth of the compartment is). Then it can leach or infiltrate
(subsoil layers are covered by fate).
- Tea/coffee break -
15
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
17:00 Afternoon session II
(Still commenting on Diagram 3)
Participant H: We end up with a very high number of CFs by the approach suggested, because there are too
many processes involved depending on location and other conditions. Is it practical to implement?
Participant D: We are facing these challenges with regional spatial and chemicals toxicity differentiation
anyways.
Participant L: Building archetypes would help application.
Participant D: Does our suggestion circumvent the problem of defining boundaries between ecosphere and
technosphere?
Participant H: Boundaries are implicitly assumed in this approach. By moving secondary steps to LCIA (i.e. in
the ecosphere) then we have a boundary there.
Participant D: Concluded that spatial boundaries are similar to the characterization, and no aerial boundary
is defined for now as no need is verified (but may depend on the application). And what do we state about
buffer zones?
Participant K: Buffer zones should be part of the production zone (i.e. field, technosphere), since they are
there to protect the adjacent environment.
Participant E and others: Buffer zones can be regulated by technology and application and are accounted
for through technosphere modelling - it comes down to how much it reduces the yield. It should become a
parameter (maybe stated as a function of another) in the LCI input requirements.
Participant K: Buffer zones depend very much on site specific conditions. Therefore modelling average
buffer zones makes little sense. Generally buffer zone regulations are based on worst case scenario, which
is not totally compatible with LCA.
Discussion on change in pesticide use + land use + transfer fraction per FU as a result of buffer zones,
concluded in buffer zones should be included as part of the production field (technosphere).
Participant P: Yes, it simply changes (reduces) the yield. It should also be referred as “good agricultural
practices” and be accounted for regional specificities.
Participant K (in the white paper board – Diagram 4): Diagram of a field of 1 ha with buffer zone (inside the
boundary). Less crop area means less mass applied and reduced yield.
16
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Diagram 4
1 ha field
crop
no crops (grass)
Buffer zone means
less mass applied --> yield is reduced
surface
fresh water
Participant B: and implication in the impact assessment by modifying the area needed for functional unit.
Participant L: For the run-off in USEtox it may be possible to implement a switch to be activated if there is a
buffer zone lowering the run-off coefficient.
Participant D: Concluded on two different CFs (with or without buffer zones).
The discussion of specific points to recommend contributed to further topics in the PROPOSAL, followed by
a discussion of different terminology and wording.
Side discussion on whether or not pesticide effects are included in land use indicators in Eco-indicator and
ReCiPe and if so, how important this double counting is.
Immediate / future action points

Write draft paper presenting the outcomes and circulate among participants.
Participant J: Interaction with data base developers regarding proposals (e.g. the formation of a ‘plant
compartment’)
Participant L: Likewise impact assessment developers should also be included in the future work
Participant J: Suggests a meeting in 6 months to check the progress and who is on board.
Participant L: Suggestions of a “how to” workshop, to get more into modelling the emissions to specific
compartments.
Participant E: In that workshop we should also look into which scenarios are relevant to the models, what
should be “stated in tables” and what should be left for the practitioners to decide.
Closing remarks
17
Glasgow workshop - Towards consensus about the delimitation between life cycle inventory and
impact assessment in LCAs with pesticide and fertilizer use
Ralph Rosenbaum: Closing remarks. A draft for commenting and revision will circulate.
18
Download