This paper presents a new calculation framework that explicitly takes into account full food web complexity and shows that the resulting SPPR for toothed whales in the Icelandic marine e
Trang 1Re-evaluating Primary Biotic Resource Use for Marine Biomass
Production: A New Calculation Framework
Anh D Luong, *,†,§ Thomas Schaubroeck,† Jo Dewulf,†,∥ and Frederik De Laender‡
†Department of Sustainable Organic Chemistry and Technology, Research Group EnVOC, Ghent University, Coupure Links 653, Ghent B-9000, Belgium
‡Research Unit in Environmental and Evolutionary Biology, Université de Namur, Rue de Bruxelles, 61, Namur 5000, Belgium
§Department of Environmental Management, Faculty of Environment, Vietnam National University of Agriculture, Hanoi 10000, Vietnam
∥European Commission, Joint Research Centre, Institute for Environment and Sustainability, Sustainability Assessment Unit, Via E Fermi 2749, I-21027 Ispra, Varese, Italy
*S Supporting Information
ABSTRACT: The environmental impacts of biomass
harvest-ing can be quantified through the amount of net primary
production required to produce one unit of harvested biomass
(SPPR-specific primary production required) This paper
presents a new calculation framework that explicitly takes into
account full food web complexity and shows that the resulting
SPPR for toothed whales in the Icelandic marine ecosystem is
2.8 times higher than the existing approach based on food web
simplification In addition, we show that our new framework can
be coupled to food web modeling to examine how uncertainty
on ecological data and processes can be accounted for while
estimating SPPR This approach reveals that an increase in the
degree of heterotrophy byflagellates from 0% to 100% results in
a two-fold increase in SPPR estimates in the Barents Sea It also shows that the estimated SPPR is between 3.9 (herring) and 5.0 (capelin) times higher than that estimated when adopting food chain theory SPPR resulting from our new approach is only valid for the given time period for which the food web is modeled and cannot be used to infer changes in SPPR when the food web is altered by changes in human exploitation or environmental changes
■ INTRODUCTION
The production of biomass by natural or artificial systems has
many well-documented environmental impacts Aquaculture
leads to for example, resource consumption,1eutrophication of
local water bodies,2and the spread of farm-origin diseases and
parasites to wildfish.3
Fisheries induce direct impacts to target species’ stock,4
substantial loss in fish biomass through mortality by discard,5 and disturbance of the benthic
community.6As a result of these impacts, various stakeholders
raise concerns about the sustainability of seafood production.7,8
One concern regarding biomass production in general is the
extraction of biotic resources A crucial biotic resource is net
primary production (NPP), i.e., the net amount of mass/energy
synthesized by primary producers that provides the basis for
higher trophic levels the Earth can sustain.9,10Because human
appropriation of NPP through biomass
consumption/extrac-tion can prevent other species to be sustained by NPP, the net
primary production both directly and indirectly required to
produce biomass extracted by man (PPR) can be used to assess
the environmental impact of biomass consumption/extraction
As a result, PPR is used in different environmental sustainability
assessment methodologies such as ecological footprint analysis, and life cycle assessment (LCA).11−19
The total PPR for biomass production of a given species is the product of the specific primary production required (SPPR), which is the amount of NPP required to produce one unit biomass of harvested species, and the amount of harvested biomass Current environmental impact assess-ments15−18 and ecological footprint studies14 calculate SPPR based on food chain theory, i.e., as SPPR = TE1‑TL, assuming a default trophic transfer efficiency (TE) of 10% for all ecosystems14,15,18,20 or a specific value for each ecosystem types16,17 and using the mean trophic level of the harvested species (TL) as listed in available databases (e.g., Fishbase21) More explanation about TL and TE can be found in the
Supporting Information, SI, (section A) We argue that there are two limitations to this approach First, this approach by
Received: May 22, 2015
Revised: September 7, 2015
Accepted: September 8, 2015
Published: September 8, 2015
pubs.acs.org/est
© 2015 American Chemical Society 11586 DOI: 10.1021/acs.est.5b02515
Trang 2definition assumes that energy/material transfer in a food web
can be described using a set of linear food chains In the SI
section D, we demonstrate that this assumption is likely to be
invalid in many cases Second, the TL of a given species may
change among ecosystems,22 and differences in TE are
observed among and within ecosystem types (Figure S1,
section A)
A more realistic method for SPPR calculation obtains
species-specific SPPR from a food web flow matrix,23
i.e., a square matrix defining feeding interactions between species in a
food web Before calculating the SPPR, the food web structure
is typically simplified to a set of intertwined food chains starting
from primary producers or detritus after removing all cycles
from this food web flow matrix.24
Next, two distinguished components of the SPPR coming from primary producers and
from detritus are calculated from these intertwined food chains
This approach has a limitation that is subtler than the approach
solely based on food chain theory, as explained in the previous
paragraph That is, the detritus component of the SPPR makes
no distinction between the origins of the detritus that can either
originates from primary producers (e.g., dead algae) or from
consumers (e.g., feces or dead grazers) This leads to an
overestimation of the SPPR since only a fraction of the detritus
originating from primary producers, which is part of NPP,
actually contributes to the SPPR By removing cycles from food
webs, this approach can distort SPPR estimation depending on
the degree of energy/material cycling, but to a yet to be
quantified extent.24
More details about these two limitations can be found in theSI section G
In this paper, we present a new mathematical calculation
framework that accounts for full food web complexity (i.e., it
does not remove food web cycles) to estimate the SPPR We
demonstrate the possibilities this approach offers by comparing
the SPPR obtained using the new framework with the SPPR
obtained after simplifying food web structure for the cases of
the Icelandic marine ecosystem and the northern Gulf of St
Lawrence, which differ in their degree of cycling Next, we
demonstrate how the new framework can be coupled to a food
web modeling technique to infer how uncertainty on ecological
data and processes translates to uncertainty of SPPR estimates,
illustrated for the case of Barents Sea
■ MATERIALS AND METHODS
Required (SPPR) Calculation: Overview The new
calcu-lation framework is based on input−output analysis as
conceived by Leontief25 and first applied in ecology by
Hannon.26 For more information and explanation regarding
ecological input−output analysis, we refer to the work of
Hannon26 and the SI section E A key element in the new
framework is the food web flow matrix, which expresses
transfers within the ecosystem, and between the ecosystem and
its surrounding environment, using a given currency (e.g., mass,
energy) This food webflow matrix can be derived via various
approaches, of which steady-state food web modeling is one
For simplicity, all primary producers will be gathered into one
group and the energy/materialflows from this group to other
groups will be changed accordingly From the food web flow
matrix, three core matrices in our new calculation framework
(i.e., the biotic transaction matrix, the production-normalized
transaction matrix and the production requirement matrix) are
derived to calculate the specific net primary production
required (SPPR) (Figure 1) These matrices and how to obtain them will be described in detail in the next sections
Required (SPPR) Calculation: Subdivision of Flows and Matrices Identification and Subdivision of Food Web Flows
In our new calculation framework, we consider n living compartments of a food web as components of the production system Each component receives inputs from its biotic environment (i.e., flows from the living compartments, imported biomass via immigration) and also from the abiotic environment (i.e., uptakes from detritus, dissolved organic or inorganic material pools) Subsequently, we subdivide the energy/materialflows (here collectively called “material flows”) leaving each component of the production system into“useful flows” and “waste flows”, following the convention of Hirata and Ulanowicz.27 Useful flows consist of flows to living compartments and net biomass growth that either causes biomass accumulation or biomass export (i.e., via emigration or
compartments Theseflows are illustrated inFigure 2 Biotic Transaction Matrix (Z) From the food web flow matrix, the transfers among n living compartments will be extracted to form the biotic transaction matrix, Z = (zij)nxn Each element zijrepresents the materialflow from the ithto the
jthliving compartment Net primary production (NPP) can be transferred to higher trophic levels indirectly through nonliving compartments, i.e., dissolved organic matter (DOM) and detritus (DET), via the consumption of bacteria and other detritivores, respectively This is because a portion of NPP (1)
is excreted to the DOM pool that is available for bacteria and/
Figure 1 New mathematical framework for calculation of specific primary production required (SPPR) z ij , a ij , l ij : the elements in the ith row and jth column of biotic transaction matrix (Z), production-normalized transaction matrix (A), and production requirement matrix (L), respectively; p j : production of the jth living compartment; I: identity matrix SPPR of the jth living compartment is the element corresponding to the primary producers in the jthcolumn of L The new calculation framework can be coupled with linear inverse modeling to provide a more ecologically realistic estimation of SPPR.
DOI: 10.1021/acs.est.5b02515
11587
Trang 3or (2) goes to the DET pool before being grazed by
detritivores As the constructed biotic transaction matrix only
consists of living compartments, such indirect transfers of NPP
through nonliving compartments to higher trophic levels are
not accounted for To account for these transfers in the biotic
transaction matrix, the values of the flows from primary
producers to the jthliving compartment consuming DOM and/
or DET will be adjusted as follows
Let pDET,j and pDOM,j be the fractions of DET and DOM
originating from primary producers of the total flows from
these nonliving compartments to the jth compartment,
respectively The values of the adjusted flows from primary
producers to the jthcompartment (zp‑j.new) as calculated from
the originalflow (zp‑j) are thus as follows:
z p j z p j p z p z
where zDET‑jand zDOM‑jare theflows from DET and DOM to
the jthcompartment, respectively The old zp‑jare replaced by
these newly obtained zp‑j·new values in the biotic transaction
matrix
Production-Normalized Transaction Matrix (A) The
production of the jth living compartment (pj) is defined as a
sum of all material flows from this compartment to living
compartments and net biomass growth ( fj) minus the imported
biomassflow (z0j)
∑
=
p j z f z
k
n
jk j j
1
0
(2)
As a result, the production (P) column vector for n living
compartments can be represented in matrix form as follows:
= × + −
P Z i F Z0 (3)
where i is the summation operator, a column vector with all
elements equal to 1, F is the column vector of net biomass
growths, and Z0is the vector of biomass imported to n living
compartments
The production-normalized transaction matrix (A) is then
calculated by normalizing the elements of each column of Z by
the production of the living compartment corresponding to
that column The matrix A can be represented in matrix form as
follows:
= × ̂−
where P̂−1is the inverse of a diagonalized matrix of vector P Each element aijof A represents the amount of material from the ithcompartment directly required to produce one unit of the jth compartment’s production Hence, each column of A represents the direct requirements for producing one production unit of the corresponding jth living compartment This normalization per produced unit is novel and different from the conventional normalization per total output amount (seeSI section E)
Production Requirement Matrix (L) In order to calculate SPPR, we also need to account for the indirect required net primary production per unit of production However, the A matrix only represents the direct flows between all living compartments per unit of production To resolve this issue, we will construct the production requirement matrix To derive its calculation, we will start with a simple equation accounting for the way in which n living compartments distribute their production to living compartments and to net biomass growth ( fj)
∑
= + + + + = +
=
p j z j z j z jn f j z f
k
n
jk j
CombiningEquation 4andEquation 5 yields the following:
p a p a p a p f
Equation 6can be represented in matrix form as follows:
or:
= − −
P (I A) 1F (8) where I is the identity matrix The matrix L = (I− A)−1is the production requirement matrix and each element lijrepresents the amount of material from the ithcompartment that is directly and indirectly required to produce one unit of the jth compartment’s production.26 , 28
Note that in A, only the direct requirements are specified, while L represents direct and indirect requirements The specific primary production required (SPPR) of the jthcompartment will then be calculated
as the element in the jth column of L corresponding to the primary producers
L is calculated through inverse matrix calculation The requirement for matrix inversion is that the determinant of matrix (I − A) differs from zero Hence, in some cases, modifications need to be made If one living compartment has a production equal to zero, then the row and column corresponding to this compartment will be eliminated prior
to inversion Subsequently, compartments of which the exported biomassflows are zero, and all their predators have been removed from previous step, will be eliminated as well However, these eliminations do not affect the result of SPPRs
of the other compartments
Comparing the New Calculation Framework with the Existing Approach Based on Food Web Simplification
In the existing approach to calculate the SPPR for a given (group of) species from a food webflow matrix, all food chains which start from either primary producers or detrital matter to the respective (group of) species in the flow network are identified after removing all cycles, using the algorithm
Figure 2 A flow diagram represents inflows and outflows of a living
compartment zij and zjk: material flows from the i th and j th living
compartments to another j th and k th living compartments, respectively;
zjj: self-consumption flow of the j th living compartment; z 0j : imported
biomass flow from the surrounding systems; z DET‑j , zDOM‑j, zDIM‑j:
uptake from detritus, dissolved organic matter and dissolved inorganic
matter, respectively; zj‑DET, zj‑DOM, zj‑DIM: waste flows from the j th
compartment due to egestion, excretion and respiration processes,
respectively; f j : net biomass growth of the jthliving compartment The
dashed line represents the system boundary between the considered
ecological community and its surrounding environment.
DOI: 10.1021/acs.est.5b02515
11588
Trang 4proposed by Ulanowicz.24 Subsequently, the corresponding
SPPR is calculated by the following equation:29
∑ ∏
=
Q P
SPPR
paths predator,prey
predator predator predator
predator,prey
(9) where P is production, Q is consumption, and DC is the diet
composition for each predator/prey constellation in each path
EE is the ecotrophic efficiency, i.e., the proportion of the total
production of a group that is consumed by the predation,
emigration, biomass accumulation andfisheries
As mentioned previously, this existing approach can lead to
overestimation of SPPR because only part of the detrital matter
coming from primary producers (i.e., part of net primary
production) should be included in the SPPR calculation To
overcome this limitation, one can multiply the detritus
component of SPPR with the fractions of detritus originating
from primary producers in the total contribution of detritus to
the diets of consumers
We have compared SPPR estimates between this existing
approach and the new framework for the cases of the Icelandic
marine ecosystem (low cycling) and the northern Gulf of St
Lawrence ecosystem (high cycling) For these two systems,
published food web models were available so that the food web
flow matrices (measured in ton of wet weight·km−2·year−1)
could be extracted.30,31 Because the fractions of detritus
originating from primary producers in the total flows from
detritus to detritivores were unknown, we assumed that these
were equal to the respective fractions of primary
producer-originated detritus in the total inflow to detritus compartments
The SPPR results based on the existing approach were obtained
using the ecological network analysis package in ECOPATH
software and Equation 9.29,32,33 The detritus components of
estimated SPPRs were then adjusted according to the
procedure mentioned above Subsequently, the new calculation
framework was applied to obtain the SPPRs from the extracted
food webflow matrices Detailed descriptions of the two food
web models and the calculations can be found in SI Section
C&D
Integrating Ecological Data and Uncertainty into the
New Framework Using Linear Inverse Modeling Linear
inverse modeling is a well-known and widely used tool to
quantify energy or materialflows in predefined food webs.34 − 38
One output of a linear inverse model (LIM) is the food web
flow matrix, which can directly be coupled to the new
calculation framework we present here We illustrate this
coupling for the Barents Sea case study In the case study, the
SPPRs of adult cod, young cod, herring, and capelin in the
Barents Sea ecosystem were calculated LIMs were already
available for spring 1998 and summer 1999 of the Barents
Sea,37 a highly productive marine high latitude ecosystem.39
These models include the compartments: dissolved organic
carbon (DOC), dissolved inorganic carbon (DIC), detritus,
bacteria, heterotrophic flagellates, heterotrophic ciliates,
phytoplankton (pico- and nanoplankton, diatom, and Phaeocytis
sp.), mezozooplankton (copepods), macrozooplankton (krill
and chaetognaths), cod Gadus morhua (split into adult and
young groups), herring Clupea Hargengus, and capelin Mallotus
villosus Because the fraction of flagellates that was
hetero-trophic in the spring was not available, the standing stock of
heterotrophicflagellates and pico- and nanoplankton (including
autotrophic flagellates) were unknown Therefore, three
scenarios: HF0, HF50, and HF100 were run, in which we
assumed 0, 50, and 100% of flagellates to be heterotrophic,
respectively As a result, we constructed and solved three LIMs (HF0, HF50, and HF100) for spring and one LIM (HF30) for summer using the package LIM in the software R version 3.0.1.40,41Compared to the published LIMs for this system,37
we added a constraint on the minimum copepod production rate (0.007day−1)42 to obtain a better constraint on the resulting carbonflows
All four LIMs were underdetermined, meaning that unknown carbon flows could be quantified with a certain range only Therefore, we used a Markov Chain Monte Carlo procedure to sample 1000 possible food web realizations Per LIM, SPPR was calculated for each species and for each of the 1000 food web realizations For each species (i.e., adult cod, young cod, herring, and capelin), we report the mean value and the 95% confidence interval of the mean SPPR
The SPPR values from our new calculation framework coupled with LIMs were compared to SPPR calculations based
on food chain theory, i.e., using SPPR = TE1‑TLto examine the
influence of including food web theory in SPPR calculations
To estimate the SPPR using food chain theory, the trophic levels (TL) of the above species were taken from the Fishbase database.21We found trophic levels of cod, herring and capelin
of 4.42, 3.23, and 3.15, respectively The trophic level of young cod was not available in this database; hence, we retrieved the value for trophic level of young cod of 4.0 based on the work of Blanchard et al.43We choose a transfer efficiency (TE) of 14%, which is the mean TE for temperate shelves and sea reported
by Libralato et al.44
■ RESULTS AND DISCUSSION
Comparison between the New Calculation Frame-work and the Existing Approach Based on Food Web Simplification The ratio SPPRcomplex (SPPR of our frame-work)/ SPPRsimple (SPPR of the approach with a simplified food web structure) was highly species-specific, ranging between 1 (e.g., herring, capelin) and 2.8 (toothed whales) for the Icelandic marine ecosystem and from 1 (molluscs) to 2.3 for redfish in the Northern Gulf of St Lawrence For adult cod (Gadus morhua), the species harvested most intensely in both ecosystems (>45% of total catches), SPPRcomplex/ SPPRsimple was 1.7 (the northern Gulf of St Lawrence) and 1.1 (the Icelandic marine ecosystem) (Figure 3) These
differences illustrate that food web complexity can greatly
influence biotic resource use estimation and that removing cycles can lead to the underestimation of SPPR However, for species not participating in any cycles our new calculation framework gave the same results as the simplified framework (e.g., herring, capelin in the Icelandic marine ecosystem, and molluscs in the Northern Gulf of St Lawrence)
The extent to which removing cycles affects the SPPR calculation depends on the degree of cycling in the food webs, which can be measured by the Finn’s cycling index.45 − 47
The degree of cycling in the northern Gulf of St Lawrence was much higher than in the Icelandic marine ecosystem, as indicated by the difference in the Finn’s cycling index (0.147 and 0.0023, respectively) Consequently, the effect of removing cycles on SPPR was more pronounced for the former ecosystem
The influence of food web simplification on SPPR estimates also depends on how the origin of detritus was accounted for in SPPR calculation For the Icelandic marine ecosystem, SPPRcomplex/SPPRsimple changed slightly when the fraction of detritus from consumers was also taken into account in SPPR
DOI: 10.1021/acs.est.5b02515
11589
Trang 5calculation due to the very high contribution of detritus from
primary producer (up to 98%) to the total inflow to the detritus
pool (Figure S3) Larger differences, ranging from 13%
(Capelin) to 38% (Flounders) were noted for the Northern
Gulf of St Lawrence because of a lower contribution (60%) of
primary producer-originated detritus in this system (Figure S5)
However, the SPPRcomplex/SPPRsimple (where SPPRsimple does
not correct for the origin of detritus) was higher than 1 (except
for American plaice, skates,flounders, molluscs, large demersal
fish, and large crustaceans in the northern Gulf of St Lawrence)
(Figures S3 and S5) Hence, this illustrates thatin most
casesremoving cycles leads to an underestimation of SPPR
rather than an overestimation by inclusion of detritus from
consumers in the two studied food webs
Coupling the New Framework with Linear Inverse
Modeling for the Barents Sea The uncertainty surrounding
ecological data and processes greatly influenced the estimated
SPPR Here, the uncertainty considered was the degree of
heterotrophy by flagellates As the fraction of heterotrophic
flagellates in the flagellate’s community rose from 0 to 100% in
the spring models, the mean SPPR increased by a factor of 2 for
allfish species These increases can be attributed to an increase
in conversion of organic carbon to CO2 (and therefore decreasing carbon transfer efficiencies) as a result of increasing heterotrophy In summer, mean SPPRs decreased by 23% (herring), 46% (adult cod), and 63% (young cod) compared to the lowest means values for spring (HF00 model) These decreases can be explained by the migration of capelin (CAP) out of the ecosystem, which reduces the number of carbon transfers in the food web, and thus leads to increase in carbon transfer efficiency.37
Hence, coupling linear inverse modeling with our new calculation framework allows one to quantify SPPR and to assess the propagation of uncertainty in ecological data and processes into SPPR estimates
Mean SPPRs for capelin and herring calculated from combining LIMs with our new calculation framework were always higher than those resulting from the food chain approach, which assumed a constant transfer efficiency of 14% Differences between both approaches mounted to factors
of 3.9 (herring) and 5.0 (capelin) Conversely, the mean SPPRs
of young cod and cod calculated from the new calculation framework were lower than those provided by the food chain approach, except for the models assuming 50% and 100% flagellate heterotrophy Interestingly, our new calculation framework indicated that the mean SPPRs for herring were always lower than that for capelin, while the food chain approach suggested the opposite This result can be explained
by the higher mean trophic level attributed to herring in the Fishbase database compared to capelin while the transfer
efficiency is assumed to be constant for all species in an ecosystem Hence, this implies that a constant transfer
efficiency for all species is not adequate for quantifying SPPRs of different (groups of) species in an ecosystem Our
Figure 3 Ratios of SPPR calculated by the new calculation framework
(SPPRcomplex) and by an approach that simplifies food web structure
(SPPRsimple) for the Icelandic marine ecosystem and the northern Gulf
of St Lawrence The two approaches give the same results when
SPPRcomplex/SPPRsimpleequals to 1 (blue horizontal line). Figure 4.Specific primary production required (SPPR) for adult cod
(COD), young cod (YCO), herring (HER), and capelin (CAP), as calculated from food web models for spring (HF00, HF50, HF100) and summer (HF30) in the Barents Sea compared to results from a food chain approach with transfer efficiency of 14% (TE14).
DOI: 10.1021/acs.est.5b02515
11590
Trang 6new calculation framework allows one to relax this assumption
and to calculate species-specific SPPR of all species in an
ecosystem from its food webflow matrix
Implications and Future Perspectives The framework
we present adds ecological realism to the environmental impact
assessment of biomass production Most notably, it allows one
to explicitly account for material cycling within ecosystems
while calculating SPPR Indeed, within most ecosystems,
material tends to cycle as opposed toflowing unidirectionally,24
for example due to the presence of omnivores or cannibalistic
species.48,49 Our results suggest that eliminating cycles from
food web flow matrix may underestimate the primary biotic
resource use This problem has been mentioned before by
Ulanowicz;24however, our present contribution is the first to
demonstrate this quantitatively Also the exchange of material
between ecosystems is explicitly included in the SPPR
calculation method we propose In many ecosystems, including
aquatic ecosystems, biomass dispersal can occur due to currents
and/or the active migration of individuals In the presented case
study of the Barents Sea, there was an advection of copepod
biomass with Atlantic Ocean water We assumed that the
requirements per capita production of this imported biomass
were the same as for locally produced biomass, in absence of
data for surrounding food webs However, we note that,
technically, multiple food webs can be easily combined into a
landscape of food webs
Transfer efficiency, representing the efficiency of energy/
material transfers among community’s components in
ecological studies, is defined as the ratio of the sum of exports
and predation to the total ingestion for a given trophic level in
the food chain approach.50As such, dead biomass caused by
natural nonpredatory mortality is considered as waste or
inefficient fraction of energy/material transfer Furthermore,
fishing or harvesting activities can be considered as “predation”
in an ecosystem Hence, if dead biomass is harvested by human
and considered as a product, then it will represent a useful
production and be subtracted from the biomass lost as detritus
In our new framework, we therefore clearly consider the part of
production that lost as detritus (caused by natural
non-predatory mortality and not harvested by man) as waste and
the rest as useful product As a result, inputs of a living
compartment are only allocated to the parts of production that
are utilized/accumulated in the system through predation/
biomass accumulation or exported out of system by emigration
andfisheries (called exploitable production) This is based on
the similar distinction between product, to which the
environmental burdens are attributed, and waste, to which
none is assigned, in LCA, which is widely used tool for
quantifying environmental impacts of a product or service
throughout its life cycle.51In an ecological production system,
waste (i.e., DOM, DET) released from consumers can be
“treated” by bacteria and detritivores without extra requirement
of NPP, meaning that there is no burden for waste treatement
We consider exploitable production as the only product of each
living compartment; hence there is no allocation needed among
products like in multifunctional processes in LCA The way that
our new framework considers natural nonpredatory mortality
and makes allocation to exploitable production is also
consistent with the existing approach that calculates SPPR
based on food web simplification By using ecotrophic
efficiencies inEquation 9, it is clear that the input requirements
(Qpredator) are allocated to only the parts of production
exploitable by living organism and humans (Ppredator ×
EEpredator)
The interpretation of SPPR estimates from our new calculation framework should be approached with care In our calculation framework, SPPRs are calculated from a food webflow matrix that represents a snapshot of a food web for over a fixed time period and a given degree of exploitation Thus, the SPPRs our framework calculates are only valid for this exploitation scenario As such, SPPRs should not be used
to assess resource use for scenarios that represent substantially higher or lower exploitation as these may alter the food web and thus the SPPR calculated by the framework In addition to human exploitation, also environmental changes (e.g., climate change, nutrient enrichment, toxic chemicals) can cause changes in the food web’s structure or the magnitude of its flows,38 , 52
thus also possibly leading to the changes in SPPR estimates If the resulting changes in food webflows are known, then our framework can be used to calculate the corresponding changes in SPPR
Steady-state food web modeling (e.g., LIM) is one of the most commonly used techniques to derive energy/material transfers in food webs Although steady-state models provide only snapshots of food webs, temporal dynamics can be inferred by constructing and solving these models at different discrete intervals of time.38 As such, the changes in SPPR estimates over time can be assessed by applying our new framework on the resulting food web flow matrices For example, the SPPR for adult cod in Barents Sea was reduced by 46% in Summer in comparison with Spring Schaubroeck et
al.53 (re)introduced a framework to also construct nonsteady stateflow matrices by considering depletion and increment of biotic food web compartments as part of import (immigrated) and net biomass growth, respectively Our presented framework
to quantify SPPR is also applicable to nonsteady state biotic transaction matrices derived by their approach However, it should be noted that our new calculation framework is independent of food web modeling techniques by which food webflow matrices are derived
The approach we propose for SPPR calculation can be extended to calculate land occupation or ecological footprint (EF) per biomass amount (ha·yr·kg biomass−1) by dividing SPPR (kg NPP·kg biomass−1) by total areal net primary production (kg NPP·ha−1·yr−1) of the studied system In the conventional EF calculation, one usually applies the food chain approach to calculate the SPPR value.14Due to the limitations associated with the food chain approach in estimating SPPR, alternatives using quantitative ecosystem modeling, which allow inclusion of a large amount of information, are preferrable.29,44 Our new approach is such an alternative approach and, for the ecosystem examined here, proved to be more conservative than classical methods because the SPPRs were always higher than those obtained after food web simplification For example, the SPPR for adult cod can be 1.7 times higher than the result from the existing approach (the Northern Gulf of St Lawrence); hence, the respective ecological footprint will likewise be 1.7 times larger In addition, the new calculation framework calculates the species-specific SPPR for all species in the systems, as such allowing one to account for the footprint of fishing associated with by-catch and mortality by discard, i.e of nontarget species In the field of LCA, Huijbregts et al.54 suggested using ecological footprint as an alternative single score indicator Our new framework contributes to better quantification of the ecological footprint of fishery-related
DOI: 10.1021/acs.est.5b02515
11591
Trang 7product (e.g., fish meal and fish oil), particularly in terms of
marine ecological footprint which is left unaccounted for in
their analysis For example, the ecological footprint or direct
marine land occupation of 1 kg herring harvested in Barents Sea
(summer) is about 17 × 10−5 ha·year (se = 0.7 × 10−5, SI
section F) This value can be then transformed to standardized
unit of ecological footprint (global hectares·year−gha·year) by
using equivalent factor (which accounts for the difference in
productivity of different land use types) of 0.4 for marine
land.54,55
Overall, our results demonstrate that explicitly incorporating
food web theory allows refining estimates of primary biotic
resource use in sustainability assessment of a product When
combined with food web modeling, it can propagate
uncertainty on ecological processes to such estimates Even
though we focused on marine ecosystems, the presented
framework is applicable to any ecosystem, including freshwater
and/or terrestrial ones, for which a food web flow matrix is
available or can be estimated We therefore expect this
framework to advance the use and development of SPPR in
environmental impact assessment and ecological footprint
accounting studies
■ ASSOCIATED CONTENT
*S Supporting Information
The Supporting Information is available free of charge on the
ACS Publications websiteat DOI:10.1021/acs.est.5b02515
Additional information on trophic level (TL) and
transfer efficiency (TE) of different marine ecosystem
types (section A), case studies’ system description and
calculation (section B,C), mathematical proof (section
D), introduction to input−output analysis (section E),
marine land occupation calculation (section F), and
numerical examples (section G) (PDF)
■ AUTHOR INFORMATION
Corresponding Author
*Tel.: +32-9-2649927; fax: +32-9-2646243; e-mail: Ducanh
luong@ugent.be(A.D.L.)
Notes
The authors declare no competingfinancial interest
■ ACKNOWLEDGMENTS
A.D.L is a PhD research fellow supported by Special Research
Fund (BOF) of Ghent University T.S was granted by a
research project (number 3G092310) of the Research
FoundationFlanders (FWO-Vlaanderen) We thank Bui
Xuan Dieu for discussion on mathematics and the EwE
development team for giving us the permission to access the
source code of ECOPATH with which SPPR based on food
web simplification was calculated
■ REFERENCES
(1) Huysveld, S.; Schaubroeck, T.; De Meester, S.; Sorgeloos, P.; Van
Langenhove, H.; Van linden, V.; Dewulf, J , Resource use analysis of
Pangasius aquaculture in the Mekong Delta in Vietnam using Exergetic
Life Cycle Assessment J Cleaner Prod 2013, 51, 225−233.
(2) Folke, C.; Kautsky, N.; Troell, M The cost of eutrophication
from salmon farming: Implications for policy J Environ Manage.
1994, 40, 173−182.
(3) Krkošek, M.; Lewis, M A.; Morton, A.; Frazer, L N.; Volpe, J P.
Epizootics of wild fish induced by farm fish Proc Natl Acad Sci U S.
A 2006, 103, 15506−15510.
(4) Pauly, D.; Christensen, V.; Guénette, S.; Pitcher, T.; Sumaila, U.; Walters, C.; Watson, R.; Weller, D Towards sustainability in world fisheries Nature 2002, 418, 689−695.
(5) Alverson, D.; Freeberg, M.; Murawski, S.; Pope, J A Global Assessment of Fisheries Bycatch and Discards; FAO: Rome, 1994 (6) Johnson, K Review of National and International Literature on the Effects of Fishing on Benthic Habitats; U.S Department of Commerce, National Oceanic and Atmospheric Administration, National Marine Fisheries Service: Maryland, 2002.
(7) Henriksson, P J G.; Guinee, J B.; Kleijn, R.; de Snoo, G R Life cycle assessment of aquaculture systems - a review of methodologies Int J Life Cycle Assess 2012, 17 (3), 304−213.
(8) Pelletier, N L.; Ayer, N W.; Tyedmers, P H.; Kruse, S A.; Flysjo, A.; Robillard, G.; Ziegler, F.; Scholz, A J.; Sonesson, U Impact categories for life cycle assessment research of seafood production systems: Review and prospectus Int J Life Cycle Assess 2007, 12 (6), 414−421.
(9) Vitousek, P M.; Ehrlich, P R.; Ehrlich, A H.; Matson, P A Human Appropriation of the Products of Photosynthesis BioScience
1986, 36 (6), 368−373.
(10) Haberl, H Human appropriation of net primary production as
an environmental indicator: Implications for sustainable development Ambio 1997, 26 (3), 143−146.
(11) Kautsky, N.; Berg, H.; Folke, C.; Larsson, J.; Troell, M Ecological footprint for assessment of resource use and development limitations in shrimp and tilapia aquaculture Aquacult Res 1997, 28, 753−766.
(12) Larsson, J.; Folke, C.; Kautsky, N Ecological limitations and appropriation of ecosystem support by shrimp farming in Columbia Environ Manage 1994, 18 (5), 663−676.
(13) Swartz, W.; Sala, E.; Tracey, S.; Watson, R.; Pauly, D The Spatial Expansion and Ecological Footprint of Fisheries (1950 to Present) PLoS One 2010, 5 (12), e15143.
(14) Borucke, M.; Moore, D.; Cranston, G.; Gracey, K.; Iha, K.; Larson, J.; Lazarus, E.; Morales, J C.; Wackernagel, M.; Galli, A Accounting for demand and supply of the biosphere’s regenerative capacity: The National Footprint Accounts’ underlying methodology and framework Ecol Indic 2013, 24, 518−533.
(15) Papatryphon, E.; Petit, J.; Kaushik, S J.; van der Werf, H M G Environmental Impact Assessment of Salmonid Feeds Using Life Cycle Assessment (LCA) Ambio 2004, 33 (6), 316−323.
(16) Langlois, J.; Fréon, P.; Delgenes, J.-P.; Steyer, J.-P.; Hélias, A New methods for impact assessment of biotic-resource depletion in LCA of fisheries: theory and application J Cleaner Prod 2014, 73 (0),
63 −71.
(17) Langlois, J.; Fre ́on, P.; Delgenes, J P.; Steyer, J P.; Hélias, A., Biotic resources extraction impact assessment in LCA fisheries In Proceedings of the 8th International Conference on Life Cycle Assessment in the Agri-Food Sector (LCA Food 2012); Corson, M S.; van der Werf, H.
M G., Eds INRA: Rennes, 2012; pp 517 −522.
(18) Pelletier, N.; Tyedmers, P Feeding farmed salmon: Is organic better? Aquaculture 2007, 272, 399−416.
(19) Pelletier, N.; Tyedmers, P.; Sonesson, U.; Scholz, A.; Ziegler, F.; Flysjo, A.; Kruse, S.; Cancino, B.; Silverman, H Not All Salmon Are Created Equal: Life Cycle Assessment (LCA) of Global Salmon Farming Systems Environ Sci Technol 2009, 43 (23), 8730 −8736 (20) Pauly, D.; Christensen, V Primary production required to sustain global fisheries Nature 1995, 374, 255−257.
(21) Froese, R.; Pauly, D FishBase 2000: Concepts, Design and Data Sources; ICLARM: Los Barnos, Laguna, Philippines, 2000.
(22) Pedersen, T.; Nilsen, M.; Nilssen, E M.; Berg, E.; Reigstad, M Trophic model of a lightly exploited cod-dominated ecosystem Ecol Modell 2008, 214 (2 −4), 95−111.
(23) Christensen, V.; Walters, C J Ecopath with Ecosim: methods, capabilities and limitations Ecol Modell 2004, 172, 109−139 (24) Ulanowicz, R E Ecosystem trophic foundations: Lindeman Exonerata In Complex Ecology: The Part −Whole Relation in Ecosystems; Patten, B C.; Jorgensen, S E., Eds.; Prentice Hall: New York, 1995.
DOI: 10.1021/acs.est.5b02515
11592
Trang 8(25) Leontief, W W Input-Output Economics; Oxford University
Press: New York, 1966.
(26) Hannon, B The structure of ecosystems J Theor Biol 1973, 41,
535 −546.
(27) Hirata, H.; Ulanowicz, R E Information theoritical analysis of
ecological networks Int J Syst Sci 1984, 15 (3), 261 −270.
(28) Leontief, W W Quantitative Input and Output Relations in the
Economic Systems of the United States Rev Econ Stat 1936, 18 (3),
105−125.
(29) Cury, P M.; Shannon, L J.; Roux, J.-P.; Daskalov, G M.; Jarre,
A.; Moloney, C L.; Pauly, D Trophodynamic indicators for an
ecosystem approach to fisheries ICES J Mar Sci 2005, 62 (3), 430 −
442.
(30) Buchary, E A., Preliminary reconstruction of the Icelandic
marine ecosystem in 1950 and some predictions with time series data.
In Fisheries Impacts on North Atlantic Ecosystems: Models and Analyses;
FCRR: 2001; Vol 9.
(31) Morissette, L.; Despatie, S.-P.; Savenkoff, C.; Hammill, M O.;
Bourdages, H.; Chabot, D Data gathering and input parameters to
construct ecosystem models for the Northern Gulf of the St Lawrence
(mid-1980’s) In Canadian Technical Report of Fisheries and Aquatic
Sciences; Department of Fisheries and Oceans: Quebec, 2003; Vol.
2497.
(32) Christensen, V.; Pauly, D ECOPATH II a software for
balancing steady-state ecosystem models and calculating network
characteristics Ecol Modell 1992, 61 (3 −4), 169−185.
(33) Christensen, V.; Pauly, D Ecopath with Ecosim: A User ’s Guide;
Fisheries Centre, University of British Columbia: Vancouver, 2005.
(34) Vézina, A F.; Platt, T Food web dynamics in the ocean I
Best-estimates of flow networks using inverse methods Mar Ecol.: Prog Ser.
1988, 42, 269−287.
(35) Kones, J K.; Soetaert, K.; van Oevelen, D.; Owino, J O.;
Mavuti, K Gaining insight into food webs reconstructed by the inverse
method J Mar Syst 2006, 60 (1 −2), 153−166.
(36) Van Oevelen, K.; Van den Meersche, K.; Meysman, F J R.;
Soetaert, K.; Middelburg, J J.; Vezina, A F Quantifying food web
flows using linear inverse models Ecosystems 2010, 13, 32−45.
(37) De Laender, F.; Van Oevelen, D.; Soetaert, K.; Middelburg, J J.
Carbon transfer in a herbivore- and microbial loop-dominated pelagic
food webs in the southern Barents Sea during spring and summer.
Mar Ecol.: Prog Ser 2010, 398, 93 −107.
(38) Luong, A D.; De Laender, F.; Olsen, Y.; Vadstein, O.; Dewulf,
J.; Janssen, C R Inferring time-variable effects of nutrient enrichment
on marine ecosystems using inverse modelling and ecological network
analysis Sci Total Environ 2014, 493 (0), 708−718.
(39) Rat’kova, T N.; Wassmann, P Seasonal variation and spatial
distribution of phyto- and protozooplankton in the central Barents
Sea J Mar Syst 2002, 38 (1 −2), 47−75.
(40) R Development Core Team, R: A Language and Environment for
Statistical Computing; R Foundation for Statistical Computing: Vienna,
2009.
(41) Van Oevelen, D.; Van den Meersche, K.; Meysman, F J R.;
Soetaert, K.; Middelburg, J J.; Vezina, A Package LIM, Implementing
Linear Inverse Models in R; R package version 1.4.2 2009.
(42) Eiane, K.; Tande, K S Meso and macrozooplankton In
Ecosystem Barents Sea; Sakshaug, E., Johnsen, G., Kovacs, K., Eds.;
Tapir Academic Press: Trondheim, Norway, 2009.
(43) Blanchard, J L.; Pinnegar, J K.; Mackinson, S Exploring marine
mamal-fishery interactions using “Ecopath with Ecosim”: Modelling
the Barents Sea ecosystem; Sci Ser Tech Rep 117: CEFAS Lowestoft,
2002.
(44) Libralato, S.; Coll, M.; Tudela, S.; Palomera, I.; Pranovi, F.
Novel index for quantification of ecosystem effects of fishing as
removal of secondary production Mar Ecol.: Prog Ser 2008, 355,
107−129.
(45) Finn, J T Measure of ecosystem structure and functioning
derived from analysis of flow J Theor Biol 1976, 56, 363−380.
(46) Ulanowicz, R E Identifying the structure of cycling in
ecosystems Math Biosci 1983, 65, 219−237.
(47) Latham, L G Network flow analysis algorithms Ecol Modell.
2006, 192, 586−600.
(48) Smith, C.; Reay, P Cannibalism in teleost fish Reviews in Fish Biology and Fishery 1991, 1, 41−64.
(49) Suh, S Theory of materials and energy flow analysis in ecology and economics Ecol Modell 2005, 189 (3−4), 251−269.
(50) Christensen, V.; Pauly, D In Flow Characteristics of Aquatic Ecosystems, Trophic models of aquatic ecosystems ICLARM Conf Proc 26; Christensen, V., Pauly, D., Eds.; 1993, pp 338−352 (51) ISO, ISO:14000: Environmental Managementlife Cycle Assess-mentprinciples and Framework 2006.
(52) De Laender, F.; Van den Brink, P J.; Janssen, C R Functional redundancy and food web functioning in linuron exposed ecosystems Environ Pollut 2011, 159, 3009−3017.
(53) Schaubroeck, T.; Staelens, J.; Verheyen, K.; Muys, B.; Dewulf, J Improved ecological network analysis for environmental sustainability assessment; a case study on a forest ecosystem Ecol Modell 2012, 247, 144−156.
(54) Huijbregts, M A J.; Hellweg, S.; Frischknecht, R.; Hungerbu ̈hler, K.; Hendriks, A J Ecological footprint accounting in the life cycle assessment of products Ecol Econ 2008, 64 (4), 798− 807.
(55) Wackernagel, M.; Mondreda, C.; Moran, D.; Wermer, P.; Goldfinger, S.; Deumling, D.; Murray, M National Footprint and Biocapacity Accounts 2005: The Underlying Calculation Method; Global Foodprint Network: Oakland, 2005.
DOI: 10.1021/acs.est.5b02515
11593