Content uploaded by Peter Krost
Author content
All content in this area was uploaded by Peter Krost on Jan 14, 2020
Content may be subject to copyright.
Content uploaded by Jonne Kotta
Author content
All content in this area was uploaded by Jonne Kotta on Jan 01, 2020
Content may be subject to copyright.
Cleaning up seas using blue growth initiatives: Mussel farming for
eutrophication control in the Baltic Sea
Jonne Kotta
a,
⁎,MartynFutter
b
, Ants Kaasik
a
, Kiran Liversage
a
, Merli Rätsep
a
, Francisco R. Barboza
c
,
Lena Bergström
d
, Per Bergström
e
, Ivo Bobsien
c
,EliecerDíaz
f,1
, Kristjan Herkül
a
, Per R. Jonsson
e,q
,
Samuli Korpinen
g
, Patrik Kraufvelin
h,2
, Peter Krost
i
, Odd Lindahl
j
, Mats Lindegarth
e
,
Maren Moltke Lyngsgaard
k
, Martina Mühl
i
, Antonia Nyström Sandman
l
,HelenOrav-Kotta
a
,MarinaOrlova
m
,
Henrik Skov
n
, Jouko Rissanen
g
, Andrius Šiaulys
o
, Aleksandar Vidakovic
p
, Elina Virtanen
g
a
Estonian Marine Institute, University of Tartu, Mäealuse 14, EE-12618 Tallinn, Estonia
b
Department of Aquatic Sciences and Assessment, Swedish University of Agricultural Sciences, Box 7050, SE-75007 Uppsala, Sweden
c
GEOMAR Helmholtz Centre for Ocean Research Kiel, Düsternbrooker Weg 20, DE-24105 Kiel, Germany
d
Department of Aquatic Resources, Swedish University of Agricultural Sciences, Skolgatan 6, SE-74242 Öregrund, Sweden
e
Department of Marine Sciences –Tjärnö Marine Laboratory, University of Gothenburg, Tjärnö, SE-45296 Strömstad, Sweden
f
Novia University of Applied Sciences, Raseborgsvägen 9, 10600 Ekenäs, Finland
g
Marine Research Centre, Finnish Environment Institute,FIN-00790Helsinki, Finland
h
Novia University of Applied Sciences, Raseborgsvägen 9, 10600 Ekenäs, Finland
i
Coastal Research and Management,Tiessenkai12, D-24159 Kiel, Germany
j
Musselfeed AB, Hallgrens väg 3, SE-47431 Ellös, Sweden
k
Orbicon, Department for Nature and Environment, Jens Juuls vej16, 8260 VibyJ., Denmark
l
AquaBiota Water Research, Löjtnantsgatan 25, SE-11550 Stockholm, Sweden
m
Sankt-Petersburg Research Centre of Russian Academy of Science, University embankment 5, 199034 St.-Petersburg, Russia
n
DHI, Agern Alle 5, 2970 Hørsholm, Denmark
o
Marine Research Institute, Klaipeda University, Universiteto ave. 17, LT-92294 Klaipėda, Lithuania
p
Department of Animal Nutrition and Management, Swedish University of Agricultural Sciences, Box 7024, SE-75007 Uppsala, Sweden
q
Environmental and Marine Biology, Åbo Akademi University, Finland
HIGHLIGHTS
•Mussel farming is a viable internal mea-
sure to address Baltic Sea eutrophica-
tion.
•Rates of nutrient removal depend on sa-
linity at the regional scale and food
availability at the local scale.
•Cost effectiveness of nutrient removal
by mussel farming depends also on
farm type.
•Total farm area needed for achieving
HELCOM nutrient reduction targets is
realistic.
GRAPHICAL ABSTRACT
abstractarticle info
Science of the Total Environment 709 (2020) 136144
⁎Corresponding author.
E-mail address: jonne@sea.ee (J. Kotta).
1
Present address: Department of Environmental Sciences, University of Helsinki, Finland.
2
Present address: Department of Aquatic Resources, Institute of Coastal Research, Swedish University of Agricultural Sciences, Öregrund, Sweden.
https://doi.org/10.1016/j.scitotenv.2019.136144
0048-9697/© 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Contents lists available at ScienceDirect
Science of the Total Environment
journal homepage: www.elsevier.com/locate/scitotenv
Article history:
Received 18 October 2019
Received in revised form 13 December 2019
Accepted 14 December 2019
Available online 19 December 2019
Editor: Ashantha Goonetilleke
Keywords:
Aquaculture
Blue growth
Eutrophication control
Internal measures
Mussel farming
Baltic Sea
Eutrophication is a serious threat to aquatic ecosystems globally with pronounced negative effects in the Baltic
and other semi-enclosed estuaries and regional seas, where algal growth associated with excess nutrients causes
widespread oxygen free “dead zones”and other threats to sustainability. Decades of policy initiatives to reduce
external (land-based and atmospheric) nutrient loads have so far failed to control Baltic Sea eutrophication,
which is compounded by significant internal release of legacy phosphorus (P) and biological nitrogen
(N) fixation.Farming and harvestingof the native mussel species(Mytilus edulis/trossulus)is a promising internal
measure for eutrophication control in the brackish Baltic Sea. Mussels from the more saline outer Baltic had
higher N and P content than those fromeither the inner or centralBaltic. Despite theirrelatively low nutrient con-
tent, harvesting farmed mussels from the central Baltic can be a cost-effective complementto land-based mea-
sures needed to reach eutrophication status targets and is an important contributor to circularity. Cost
effectiveness of nutrient removal is more dependent on farm type than mussel nutrient content, suggesting
the need for additional development of farm technology. Furthermore, current regulations are not sufficiently
conducive to implementation of internal measures, and may constitute a bottleneck for reaching eutrophication
status targets in the Baltic Sea and elsewhere.
© 2019 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
1. Introduction
Eutrophication is a global threat to many aquatic ecosystems and its
negativeeffects are particularly pronounced in semi-enclosed estuaries
and regional seas (Diaz and Rosenberg, 1995;Conley et al., 2009a;
Rabalais et al., 2009). Excessive amounts of nitrogen (N) and phospho-
rus (P) from present-day and legacy sources support massive algal
blooms which results in widespread and increasing oxygen free “dead
zones”(Breitburg et al., 2018), increasing susceptibility to ocean acidifi-
cation (Cai et al., 2011), reduced biodiversity and loss of ecosystem
functions and services (Smith, 2003;Riedel et al., 2016). In the Baltic
Sea, a multi-jurisdictional water body, N40 years of international efforts
to reduce external nutrient (N and P) inputs have failed to solve the eu-
trophication problem (Fleming-Lehtinen et al., 2015;Andersen et al.,
2017). Today, 97% of the marine area is considered as degraded due to
eutrophication (HELCOM, 2018,Fig. 1) and despite significant reduction
in external loads, the total P pool in Baltic Sea waters continues to in-
crease (Savchuk, 2018) and internationally agreed upon water quality
targets are still not met. To date, management actions have primarily fo-
cused on minimizing external loads, i.e., terrestrial point sources and
diffuse nutrient inputs. Agriculture is targeted in many cases (Larsson
and Granstedt, 2010) but the internal loads of legacy P released from
marine sediments (Vahtera et al., 2007;Conley et al., 2009a) and atmo-
spherically fixed N are often neglected (Vahtera et al., 2007), as are non-
food nutrient sources (Hamilton et al., 2018).
Aquaculture is a key component of the EU Blue Growth strategy (EC,
2012) and can have significant positive and negative effects on water
quality. Aquaculture is the fastest growing food-producing sector and
currently represents nearly 50% of global fish, crustacean and mollusc
production (FAO, 2018). Marine bivalves, e.g., mussels, oysters, clams
and other shellfish, are often referred to as extractive species as these
filter feeding species act as nutrient sinks by ingesting particles
suspended in the water column. Importantly, harvesting of cultivated
mussels removes both N and P, thereby improving water quality in af-
fected areas (Carlsson et al., 2012;Kraufvelin and Díaz, 2015).
Aquaculture can also make a positive contribution to circularity and
nutrientrecycling. Most internal eutrophication control measuresmake
P unavailable for re-use through, e.g., bottom water oxygenation to
change sediment redox status and the binding of P to iron
(Stigebrandt et al., 2015) or aluminium treatment to effectively immo-
bilize P in the sediment (Rydin et al., 2017). Unlike the aforementioned
measures, harvesting of internally produced biomass, (i.e. farmed mus-
sels) offers the potential for efficient recirculation of nutrients from sea
to land. Harvested mussels can be used to produce feed for chickens
(McLaughlan et al., 2014)orfish (Vidakovic et al., 2015), as well as for
human consumption (Gren et al., 2009). Harvested mussels can also
be used for bioenergy production (Hu et al., 2011;Nkemka and Murto,
2013), or as a soil amendment.
The failure to control Baltic Seaeutrophication through external nu-
trient loadreduction measures has highlighted the needfor in-situ (in-
ternal) methods to lower nutrient concentrations in the water column,
e.g. through geoengineering (Stigebrandt et al., 2015;Rydin et al., 2017)
or biomass harvesting (Gren et al., 2009). Intensive fishing of commer-
cial or non-commercial fishes (e.g., three-spined stickleback, round
goby) has been proposed as an alternative means for removing nutri-
ents from theBaltic Sea. However, this could have unknown and poten-
tially catastrophic consequences for marine biodiversity due to the role
as top or intermediate predators that these species have in littoral hab-
itats. While internal measures for nutrient regulation are not a universal
means of controlling eutrophication, they should be considered when
feasible external measures have been tried and found to be inadequate
(Savchuk, 2018). It should be noted, however, that many internal mea-
sures havebeen associated with high costs for nutrient removal (Lurling
et al., 2016) aswell as undesirable secondary effects such as damage to
benthic habitats (Stadmark and Conley, 2011), potentially harmful
shifts in thermal regime (Conley, 2012) and/or food web impacts
(Naylor et al., 2001). The Baltic Sea region is an important test case
highlighting the opportunities and challenges of farming native bivalve
species as an internal measure to mitigate the adverse effects of coastal
eutrophication. The region has a long, well-documented history of eco-
system deterioration, high data density and multiple cross-border envi-
ronmental management actions to counter marine eutrophication
(Reusch et al., 2018).
Farming of the ubiquitous blue mussel species complex (Mytilus
edulis/trossulus,Stuckas et al., 2009) has been proposed as an internal
measure for eutrophication control in the brackish Baltic Sea (Lindahl
et al., 2005;Gren et al., 2009;Petersen et al., 2014;Schröder et al.,
2014;Ozoliņa, 2017;Kiessling et al., 2019). Mussel farming has also
been criticized as being not cost effective (Hedberg et al., 2018)and
harmful to the environment (Stadmark and Conley, 2011).
Blue mussels are marine species and form hybrid zones within the
Baltic Sea (Stuckas et al., 2009). While individuals are able to survive
down to salinities of 4–5 practical salinity units (PSU), they grow better
in high salinity conditions where they do not need to expend as much
energy on osmoregulation (Maar et al., 2015). Blue mussels are primary
consumers and usually the dominant species (i.e. main contributor to
abundance and biomass) in the environments where they occur, and
consequently their sustainable harvest is not expected to produce cas-
cading effects or other impacts on the stability of the food web.
Blue mussel farming relies on recruitment of free-swimming larvae
(veligers) from wild populations that are entrained into the water col-
umn and passively dispersed from natural mussel reefs. After dispersal,
2J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
veligers attach themselves to available substrates, including objects in
the water column, e.g., mussel farms. Thus, determining how to best al-
locate areas suitable for mussel farming requires consideration of the
connectivity between candidate farm sites and natural mussel reefs in
order to define areas that do not require artificial mussel seeding.
Farms usingropes with high surface area perunit length, i.e. ribbons,
Swedish bands or so-called “fuzzy ropes”promote higher rates of larval
settlement. After settlement, it is necessary to account for the way in
which salinity, food availability and wave action affect growth rates
and to select sites with the highest harvest potential. A typical Baltic
Sea mussel farm has an area of a b5 ha and consists of 10–100 km of
rope suspended at different depths (Holmer et al., 2015;Kraufvelin
and Díaz, 2015). Cost effectiveness of the farms is dependent on nutrient
and salinity levels as well as the type of equipment for culturing mus-
sels, with specialized ropes that optimize veliger recruitment being
the most effective for culturing the small mussels found in the Baltic.
Harvest rates are usually expressed in units of mass of mussels per
metre of rope. Mussels are harvested one to two years after recruitment,
depending on site productivity. As farmed mussels spend their entire
life suspended in the water column, they are less affected by contami-
nated sediments than benthic dwelling organisms but can be suscepti-
ble to contamination by algal toxins (Sipiä et al., 2001).
A synthesis of a large number of recent measurements of farmed
mussel growth in the Baltic Sea and a new model chain for predicting
growth and nutrient removal potential across key environmental gradi-
ents are presented. The relationship between wild mussel production
and predicted nutrient removal through harvest of farmed mussels
was quantified by modelling occurrence of wild mussels throughout
the whole Baltic Sea.A biophysical dispersal model wasused to analyse
direction and distance of larval drift from each natural mussel reef. Next,
spatially explicit and empirically modelled growth rates of farmed mus-
sels were combined with measured N and P concentrations in mussels
Fig. 1. Areas of the Baltic Sea (coloured; 97% of its total surface area) that currently have unacceptable water quality with respect to eutrophication. Different colours indicate different
water quality classes. The 2013 HELCOM Ministerial Meeting agreed on the amount of reduction in emissions for nitrogen (N) and phosphorus (P) in different sub-basins of the Baltic
Sea in order to meet goals of the Baltic Sea Action Plan. However, trend-based estimates demonstrate that the maximum allowable nutrient inputs are still exceeded in the Central and
Inner Baltic Sea. The excesses are shown as numbers in the different sub-basins. Blue hatched areas show the best mussel growth location separately for Outer, Central and Inner
regions predicted by the model. Within these regions, blue rectangles show the surface area of future mussel farms that are needed to meet the basin-specific goals of nutrient load
reduction defined in the Baltic Sea Action Plan. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
3J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
harvested from production-scale farms to quantify nutrientremoval po-
tential. Finally, farm-scale nutrient removal estimates were upscaled to
predict thetotal area of musselfarms needed to make a meaningfulcon-
tribution to reducing Baltic Sea eutrophication.
2. Material and methods
2.1. Study area
The Baltic Sea is shallow, brackish and has almost no tide but expe-
riences intense seasonality in temperature and inflow. It is heavily af-
fected by eutrophication with N and P concentrations showing
decreasing and increasing trends, respectively (HELCOM, 2018;
Savchuk, 2018).
For most of the analyses presented here, the Baltic Sea was divided
into Outer (Kattegat and Belt Sea), Central (Northern Baltic Proper,
Western and Eastern Gotland Basins, Gdansk Basin and Bornholm
Basin) and Inner regions (Bothnian Bay, Bothnian Sea, Archipelago
Sea, Åland Sea, Gulf of Finland andGulf of Riga), representing the gradi-
ent from the near-oceanic (Outer) to brackish-water conditions (Fig. 2).
Despite salinity constraints, several characteristics of the Baltic Sea
favour mussel farming for eutrophication control. First, the Baltic Sea
is very eutrophic and food is only rarely a limiting factor for mussels
(Kotta et al., 2015). Thus, within suitable habitat ranges, elevated re-
source availability can compensate for growth limitation associated
with reduced salinity (Kotta et al., 2015). Second, high nutrient concen-
trations in the water require in-situ removal actions for which mussel
farming is promising. Finally, more than forty years of international
agreements and land-based measures have failed to solve the problem
of Baltic Sea eutrophication; large amounts of money have been allo-
cated to reduce inputs from land with variable, often minimal, effects
(Helin, 2013).
2.2. Mussel farms
Harvest data are reported from three farms, one each in the Outer,
Central and Inner Baltic (Table 1 and Fig. 1). The Kumlinge farm (Inner
Baltic) is located in the Åland archipelago. It was established in spring
2010 and harvested in November 2012. The farm technology consisted
of four 120 × 3 m nets with a mesh size of 15 cm fastened to floating
Fig. 2. Location of samplingpoints for the distribution and growth of blue mussels. Colours depict different sub-regions of the Baltic Sea.(For interpretation of the references to colour in
this figure legend, the reader is referred to the web version of this article.)
4J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
plastic pipes. The average water depth was 8 m and average bottom
water current speeds were 3–4cms
−1
(Kraufvelin and Díaz, 2015).
The Sankt Anna farm (Central Baltic) is located on a sheltered site in
the Swedish Östergötland archipelago. The farm was established in
spring 2016 and harvested in October 2018. The farm technology
consisted of “spat catching”rope developed by Quality Equipment Ltd.
and optimised for settling of small mussels. Ropes were hung at a
depth of 2–12 m. Average water depth at the site was 20 m and bottom
water current speeds were low. The Kiel farm (Outer Baltic) is located
near Kiel,Germany. The farm technology consists of ropes with collector
bands and socks optimised for the production of large mussels. Ropes
were suspended at depths of 1–3.5 m and average water depths ranged
between 7 and 11 m (Schröder et al., 2014). The farm was established in
2011 and the first harvest took place in September 2012.
2.3. Spatial mapping
The blue mussel distribution data were combined from different
sources: benthos database of the Estonian Marine Institute, University
of Tartu (http://loch.ness.sea.ee/gisservices2/liikideinfoportaal/); the
VELMU database, Finnish Environment Institute (http://www.
ymparisto.fi/en-US/VELMU); the database of the Swedish National
Monitoring Programme (http://sharkdata.se/), benthic inventory data
collected by AquaBiota (http://www.aquabiota.se/en/researchservices/
inventories-using-underwater-video/), the database of Marine Re-
search Institute, Klaipeda University; EurOBIS (http://www.eurobis.
org/) and EMODnet (http://www.emodnet-biology.eu/portal/)
(Fig. 2). Altogether, data from 226,031 stations from coastal hard and
soft bottom habitats of the Baltic Sea were included in this study. This
dataset was based on a regional sampling and sample processing proto-
col developed for the HELCOM COMBINE programme (HELCOM, 2015).
The stations included were sampled at least once in summer (June to
August) between 2005 and 2015. On hard bottoms, blue mussels were
collected by divers using a standard bottom frame (0.04 m
2
) and/or a
hand-held drop camera operated from small motorboats with recording
devices operated on the surface. On soft bottoms, samples were col-
lected using different benthos grabs (sampling area 0.02–0.1 m
2
).
Quantitative samples were sieved in the field using 0.25 mm mesh
screens. The residues were stored at −20 °C and subsequent sorting,
counting, weighing and measuring of blue mussels were performed in
the laboratory.
Oxygen measurements under the farms were made with a JFE
Advantech optical DO sensor (https://www.jfe-advantech.co.jp/eng/
ocean/rinko/rinko3.html).
The majority of existing experimental measurements of mussel
growth in the Baltic Sea (n= 14,944) were used to model the potential
growth and yields across the key environmental gradients. This includes
the original data of the INTERREG Baltic EcoMussel and Baltic Blue
Growth projects as well as data from different national research initia-
tives from Estonia, Finland, Sweden, Denmark and Germany (Fig. 2).
2.4. Mussel tissue analysis
Nutrientcontent was analysedfor 124 samples of blue mussel tissue.
In each case, 100–150 g of fresh material (shells, soft tissue and associ-
ated water) were analysed in the following manner. Whole frozen mus-
sels were removed from the freezer and thawed. A portion of the
thawed mussels (shells, soft tissues and associated water) were manu-
ally crushed using a mortar and pestle. Between 100 and 150 g of the
crushed mussels were weighed. This weight is reported as the sample
wet weight. The samples were then freeze-dried at −80 °C and
weighed. They were then oven dried at 105 °C (to remove any residual
moisture) and weighed again to determine dry mass and dry matter
fractions.
Prior to the nutrient analysis, dried material was filtered through a
1 mm sieve. Total N measurements were performed by the laboratories
of Swedish University of Agricultural Sciences using the total Kjeldahl
nitrogen (TKN) method. Total phosphorus (P) concentrations were
analysed by Agrilab AB. Samples were acidified using sulfuric acid. P
concentrations were obtained using ICP-AES.
2.5. Statistical analysis of nutrient concentrations
In order to account for regional variability in the nutrient content of
mussels, samples were classified into those obtained in the Outer, Cen-
tral or Inner Baltic Sea. These three functionally different regions were
used to account for the regional-specific nutrient accumulation in mus-
sels when assessing the potential of nutrient removal through
harvesting.
Because the sampling design was unbalanced, i.e., the same number
of samples were not available for the different months across regions,
only the samples collected from the Outer Baltic were used to define
the best way of grouping the samples obtained in the different seasons
for subsequent analyses. A Tukey's Honest Significant Difference (HSD)
post-hoc test of an ANOVA predicting wet weight P concentrations as a
function of month indicated that samples from March, April, May and
June belonged to the same group (spring) and had no overlap with
the group of samples collected in other months (other). This grouping
was corroborated by the analysis of wet weight N concentrations.
To facilitate the comparison with mussel harvest values, which are
typically reported as total mass of mussels (i.e. shells, soft tissue and as-
sociated water), ANOVA analyses were performed on wet weight con-
centrations. Pairwise differences were assessed using the Tukey's HSD
test. The ANOVAs tested for the fixed effects of region (Outer, Central
and Inner), season (spring or other) and their interaction.
2.6. Modelled environmental variables
Care was taken to select the most relevant ecological variables in
order to reachthe most robust predictions about the role of the environ-
ment for blue mussel occurrence and growth. When the variable selec-
tion is inadequate, a model may include irrelevant variables and its
predictive power is low (MacNally, 2000). Earlier studies have shown
Table 1
Summary of environmental conditions, nutrient removal and economic factors for three
blue mussel farms in the Baltic Sea.
Variable Unit Kumlinge Sankt Anna Kiel
Latitude 60.2147°
N
58.3564° N 54.3755° N
Longitude 20.7524°
E
16.9368° E 10.1634° E
Region of the Baltic
Sea
Inner Central Outer
Average salinity 6 6 15
Chlorophyll a
(mean/max/min)
mg m
−3
2.0 2.0/3.5/1.0 2.3/4.5/0.9
Farm technology Nets “Spat
catching”
rope
Ropes with
collector bands
Nitrogen removal at
harvest
kg ha
−1
83 140 148
gm
−1
3.74 23.26 22.25
Phosphorus removal
at harvest
kg ha
−1
6.4 10.8 10.8
gm
−1
0.29 1.80 1.63
Farm size ha 0.90 4.0 0.30
Long line length km 20 24 2
Long line density km ha
−1
22.2 6.0 6.7
Observed biomass
yield
tonnes 14.4 81.50 5.00
kg m
−1
0.72 3.40 2.50
Modelled biomass
yield
kg m
−1
0.75 1.3 14.7
Investments €kg
−1
5.45 0.35 0.36
Operational expenses €kg
−1
3.08 0.17 1.49
Total costs €kg
−1
8.52 0.52 1.85
N removal cost €kg
−1
1638 76 208
P removal cost €kg
−1
21,300 981 2846
5J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
that water salinity, temperature conditions, and food availability (a
product of phytoplankton concentration and water flow) mostly
shape the distribution and growth of blue mussels at the Baltic Sea
scale (Kotta et al., 2015).
Model inputs for the physical and biogeochemical conditions in the
Baltic Sea were obtained from the products
BALTICSEA_ANALYSIS_FORECAST_PHY_003_006 and
BALTICSEA_ANALYSIS_FORECAST_BIO_003_007 at the Copernicus
open access data portal (http://marine.copernicus.eu/services-
portfolio/access-to-products/). These physical products covering the
whole Baltic Sea area contain data with hourly resolution and 25 vertical
levels. The biogeochemical data are served with 6-h resolution and 25
vertical levels. For both products, the horizontal grid step is regular in
latitude and longitude and is approximately 1 nautical mile. The physi-
cal product is based on simulations with the HBM ocean model code
(HIROMB-BOOS-Model). The biogeochemical product is based on simu-
lations with the BALMFC-ERGOM version of the biogeochemical model
ERGOM, originally developed at IOW, Germany. The BALMFC-ERGOM
version has been further developed at Danish Meteorological Institute
(DMI) and Bundesamt für Seeschifffahrt und Hydrographie (BSH). The
BALMFC-ERGOM model is run online coupled with the HBM ocean
model code. In the analyses presented here, annual averages of salinity
and current velocity and summer averages (June to August) of temper-
ature and chlorophyll aconcentration were used.
In addition to the aforementioned data layers, depth data acquired
from the Baltic Sea Bathymetry Database (Baltic Sea Hydrographic
Commission, 2013) were used as a modelling input variable for
predicting blue mussel presence and growth. The locations of hard bot-
tom areas were obtained from the EMODnet portal (http://www.
emodnet.eu/) and unpublished sediment data were collated from Finn-
ish Environment Institute, Geological Survey of Sweden, and the
Bundesamt für Seeschifffahrt und Hydrographie. Wave exposure data
were produced by Aquabiota, using the Simplified Wave Model method
(SWM; Wijkmark and Isæus, 2010). The SWM method calculates the
wave exposure for mean wind conditions using a nested-grids tech-
nique to take into account long distance wind effects on the local
wave exposure regime. This method results in a pattern where the
fetch values are smoothed out to the sides, and around islands in a sim-
ilar way that refraction and diffraction make waves deflect around
islands. Then a depth-attenuation correction was applied to the SWM
in order to estimate depth-attenuated wave exposure (Bekkby et al.,
2008). For maps of environmental variables, see Supplementary Fig. 1.
2.7. Modelling the occurrence of blue musselreefs along environmental gra-
dients of the Baltic Sea
In the case of distribution data, all samples having positive coverage
or biomass were considered as indicative of mussel presence and all
other samples were considered asabsences. The occurrence probability
of wild blue mussels on seafloor was modelled as a function of depth, sa-
linity, temperature, wave exposure and the presence of hard or mixed
substrate with sand, boulders and bedrock. These substrate types are
known to be good habitats for blue mussels in the Baltic Sea area (e.g.
Westerbom, 2006). A binomial Generalized Additive Model (GAM)
with logit link function was used for modelling occurrence. Possible
over-fitting was limited by constraining the degrees of freedom of
model covariates.
2.8. Hydrodynamic connectivity model
The connectivity structure among all mussel reefs in the Baltic Sea
area was estimated with a biophysical model of larval dispersal. Blue
mussel larvae may drift in the water column for up to 30 days (Bayne,
1965). Thebiophysical model combinedflow fields from an ocean circu-
lation model with a Lagrangian particle-tracking model simulating
transport of individual larvae from spawning to settling locations. The
ocean current velocity fields were produced with the three-
dimensional NEMO-Nordic model (Hordoir et al., 2013, 2015), a re-
gional configuration of the NEMO ocean engine (Madec, 2010)covering
the Baltic Sea, the Kattegat, the Skagerrak, and most of the North Sea.
The model has a horizontal spatial resolution of 3.7 km and 84 vertical
levels with depth intervals of 3 m at the surface and 23 m for the
deepest layers. The model has open boundaries between Cornwall and
Brittany, and between the Hebrides Islands and Norway with tidal har-
monics defining sea surface height (SSH) and velocities, and Levitus cli-
matology defining temperature and salinity (Levitus and Boyer, 1984).
The applied model had a free surface and the atmospheric forcing was
based on the re-analysis dataset ERA40 (Uppala et al., 2005). Runoff
was based on climatological data from several databases for the Baltic
Sea and the North Sea. Validation of the NEMO-Nordic showed that
the model correctly represents both tidally induced and wind driven
SSH anomalies (Hordoir et al., 2015).
To simulate larval drift trajectories, the Lagrangian particle-tracking
model TRACMASS (De Vries and Döös, 2001), that calculates transport
of particles using stored flow field data from the ocean model, was
used. The velocity, temperature and salinity were updated with a regu-
lar interval for all grid boxes in themodel domain(in this study- every
three hours), and thetrajectory calculations wereperformed with a 15-
min time step. Particles simulating larvae of blue mussels were released
from the model grid cells (3.7 × 3.7 km
2
) that overlapped with the mus-
sel reef areas. From each grid cell, 294 particles were released on three
occasions between June to July as this time corresponds to a planktonic
larval phase of blue mussels in the Baltic Sea region (Kautsky, 1982).
Each larva was forced to drift in one of three depth intervals:25% of lar-
vae between 0 and 10 m, 50% of larvae between 10 and 15, and 25% of
larvae between 15 and 30 m (Corell et al., 2012). The pelagic larval du-
ration (PLD) was set to either 20 or 30 days with equal probability, and
settlement was assumed at the location when the PLD was completed.
All these simulations were repeated for 8 years (1995–2002),
representing a range of North Atlantic oscillation index values (NAO;
Hurrel and Deser, 2009), which is known to correlate well with the var-
iability in circulation pattern, making a total of 670,000 released parti-
cles. A grid cell was considered to receive recruits if larvae spawned at
any of the reefs in the Baltic Sea range settled at the specified grid cell
(Supplementary Fig. 2).
2.9. Modelling the growthof blue mussels along environmental gradients of
the Baltic Sea
Blue mussel growth was modelled as statistical relationships be-
tween environmental variables and mussel growth yield experimen-
tally evaluated all over the Baltic Sea region. Only the environmental
variables known to affect regional patterns of Baltic Sea mussel growth
(salinity, temperature, chlorophyll a, exposure to waves) were included
in the model. It was assumed that new larvae can settle from 1st to 30th
of June and only in the grid cells that are connected to mussel reefs (see
previoussubsection). Thegrowth simulations were based on dry weight
of mussels as opposed to length (this allowed for negative growth dur-
ing periods of resource limitation and for greater flexibility when deal-
ing with gamete production). The model assumed that the new larvae
appeared in June. Yields were normalized with the total incubation
time (to produce data for yield per day) but a linear pattern was ob-
served within a year, thus allowing to extrapolate the predictions to
365 days. Two year predictions were calculated from one-year predic-
tions using a coefficient obtained from individual growth patterns.
Gaussian GAMs with an identity link function were used for model-
ling. Possible over-fitting was reduced by constraining the degrees of
freedom of model covariates. Final growth model included salinity and
interaction between wave exposure and chlorophyll a.Tworandomfac-
tors were used to model the dependence inherent in the growth data.
First, for a combination of farm area and year to allow for yearly varia-
tion in different farming areas. Second, for a combination of place
6J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
(within area) and yearto allow for yearly variation within areas. Thus it
was assumed that yields for two farms from thesame area or place from
the same year are more similar than yields for two farms from different
areas or places from the same year. To normalize the residual distribu-
tion, the yield per day was fourth-root transformed.
For prediction, it was assumed (based on empirical knowledge) that
no further gain can be obtained from wave exposure values above
200,000 which represents a transition from moderately exposed to ex-
posed areas (Kotta et al., 2015). Blue mussel growth was deemed im-
possible at salinity values below 3.5 (Riisgård et al., 2014). The
available growth data was more evenly spread along the north-
eastern and south-western coasts as compared to central coasts of the
Baltic Sea. Hence, in these sparsely sampled areas, growth for locations
far from any growth assessment was estimated by spatial extrapolation.
However, with respect to salinity, the main factor explaining mussel
growth in our model, extrapolations are not extensive, since thegrowth
data spans the salinity gradient.
To quantify meaningful effect sizes of the two components (salinity
vs chlorophyll aand exposure to waves) in the study area, predictions of
two-year yield were obtained for six different combinations of predictor
values, as follows. First, for each predictor the 2.5% quantile (low), the
median and the 97.5% quantile (high) were determined. To assess the
interaction effect of chlorophyll aand wave exposure, salinity was
kept at its median value while four different value combinations (low-
low, low-high, high-low, high-high) were assigned to the other two
predictors. To assess the effect of salinity, the other two predictors
were keptat their respective medians while two different value combi-
nations (low and high) were assigned to salinity.
2.10. Nutrient removal at harvest
The mass of N and P removed during harvest at the three farms was
estimated by multiplying reported wet weight harvest values and least
squares mean estimates for “other season”wet weight nutrient per-
centages for the three regions of the Baltic. These estimated percentages
were obtained from analyses of variance(ANOVA) of N and P tissue con-
centrations from 124 composite samples (Supplementary Tables 2 and
3).
3. Results
3.1. Analyses of farmed mussels
A total of 9478, 1516 and 4912 mussel samples were harvested and
measured from farms in the Outer, Central and Inner Baltic Sea (three
major analysis regions, Fig. 2). Average densities and individual wet
weights (±SE) of harvested mussels were 2654 ± 77 individuals m
−1
(individuals per metre of rope) and 0.50 ± 0.02 g ww (wet weight) in
the Outer Baltic; 4998 ± 329 individuals m
−1
and 0.20 ± 0.01 g ww
in the Central Baltic and 2326 ± 24 individuals m
−1
and 0.16 ±
0.001 g ww in the Inner Baltic, respectively.
In total, 124 composite samples ofwhole mussels (shell andsoft tis-
sue) were availablefor dry matter and nutrient analysis. Samples of blue
mussels from the Outer Baltic had significantly higher dry matter con-
tent (42.5%) than mussels from the Central (34.0%) or Inner (32.6%) Bal-
tic (Supplementary Table 1, Fig. 3). Region, season and their interaction
accounted for 62.3% and 67.7% of the total observed variation in mussel
tissue N and P percentages (Supplementary Tables 2 and 3). Nitrogen
concentrations were highest in Outer Baltic samples from spring
(1.23%), followed by those obtained for the same region in other sea-
sons (0.89%). There were no significant differences among spring Cen-
tral Baltic samples (0.70%), Central Baltic samples from other seasons
Fig. 3. Mean and standard errors (bars) of the mean dry weight proportion in the Outer,
Central and Inner Baltic Sea. Letters depict statistically significant differences between
groups (Tukey's HSD test results at pb0.05) and levels are statistically significantly
different if they do not have any letters in common.
Fig. 4. Mean (filledshapes) and standard errors(bars) of nitrogen(N) and phosphorus(P) content in farmed musselsexpressed as a percentage ofwet weight in springvs other seasonsin
the Outer,Central and Inner BalticSea. Letters depictstatistically significant differences betweengroups (Tukey'sHSD test results at p b0.05) with capital lettersdenoting N and lowercase
letters P groups. N and P levels are statistically significantly different if they do not have any letters in common.
7J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
(0.69%), and spring Inner Baltic samples (0.57%). Inner Baltic samples
from other seasons (0.52%) were significantly lower compared to all
other contexts except Inner Baltic spring samples (Supplementary
Table 2). Observed phosphorous concentrations followed similar trends
to those of N (Supplementary Table 3). The highest P concentrations
were measured in spring Outer Baltic samples (0.111%) and the lowest
in mussels from the Inner Baltic in other seasons (0.040%). Phosphorous
concentrations in spring Central (0.060%) and Inner Baltic (0.050%)
samples were significantly lower than in spring Outer Baltic samples
and did not differ from Outer Baltic samples obtained in other seasons
(0.065%). As for N, P concentrations did not statistically differ between
seasons in mussels from the Central (spring: 0.060%, other: 0.053) and
Inner (spring: 0.050%, other: 0.040%) Baltic (Fig. 4).
Biomass yield and economic information were available for three
production farms: Kumlinge (Inner Baltic), Sankt Anna (Central Baltic)
and Kiel (Outer Baltic) (Table 1). Although the higher salinity and chlo-
rophyll alevels at Kiel may suggest a greater potential for mussel bio-
mass production and hence nutrient removal, this difference was not
manifested. In fact, nutrient removal was higher in Sankt Anna
(23.3 g N m
−1
and 1.8 g P m
−1
line) than either Kiel (22.2 g N m
−1
and 1.6 g P m
−1
line) or Kumlinge (3.7 g N m
−1
and 0.3 g P m
−1
line).
Production costs were approximately four times higher at Kiel (1.85 €
kg biomass harvested
−1
) than at Sankt Anna (0.52 €kg biomass
harvested
−1
). Costs were much higher at Kumlinge (8.52 €kg biomass
harvested
−1
). Differences in production costs were the main driver of
the large difference in nutrient removal costs which were lowest at
Sankt Anna (76 €kg N)
−1
and 981 €kg P
−1
) and higher at the other
two farms.
3.2. Modelling of biomass yield and regional nutrient removal
Spatially explicit estimates of farm biomass yield were predicted
as a function of site salinity, exposure to waves and food availability
(i.e. chlorophyll aconcentration) (Fig. 5). The model explained 82.3%
of the variation in the data. Modelled patterns of biomass yield were
driven by salinity at the regional scale and food availability at the
Fig. 5. Overview of the blue musselbiomass yield model and obtainedresponse curves.Estimations referto biomass yields obtained two yearsafter the establishment of the farms.Panel A
shows the interactive effects of exposure to waves and chlorophyll aconcentration (i.e. food availability) at a salinity of 7.5 psu. Panels B-D show the interactive effects of exposure to
waves and salinity at low, medium and high chlorophyll aconcentrations. Panels E-G show the interactive effects of salinity and chlorophyll aat low, medium and high exposure
levels. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
8J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
local scale. On the model scale, the effect of salinity was estimated to
be linear (and positive) which translated to a quartic effect on the re-
sponse scale. The interaction between food concentration and expo-
sure to waves (a good proxy of water movement and exchange) was
more complex. At low water movement, elevated chlorophyll acon-
centrations were associated with low biomass yield of mussels,
whereas at moderate to good water exchange, increasing chloro-
phyll aresulted in the raised biomass yield. The overall effect size
of salinity was about 13 times as large as the effect size of the afore-
mentioned interaction. The random effects, accounting for the inter-
annual and spatial variation not explained by the mean trends in sa-
linity and the interaction between wave exposure and food availabil-
ity, explained approximately 50% of the total variance. The model
does not simulate disastrous loss of harvestable mussel biomass as-
sociated with severe storms or harmful algal blooms.
Response curves predicting mussel yield as a function of environ-
mental conditions (Fig. 5) were combined with spatial data on salinity,
wave exposure and surface chlorophyll aconcentrations to produce
pan-Baltic estimates of potential rates of biomass removal that can be
obtained using farmed blue mussels (Supplementary Fig. 3). The
model extrapolation power was assessed by predicting the average
yield in the Kiel mussel farm that was not used for model fitting. On
the model scale, the average yield was predicted to be only 10% smaller
than what was actually measured. Higher growth was predicted at
higher salinities and/or better food regimes, i.e. the Outer and Central
Baltic. Predicted biomass yield was highest in high-salinity areas of the
Fig. 6. Longline removed from the waterat Sankt Anna (A) showingmussel growth. Viewof mussel farm at Kiel showingfloats on which linesare suspended (B).Panels (C) and (D) show
the environment underneath thesame farms. Blue surfaces of panels(E) and (F) show variability inoxygen conditions(mg l
−1
) measuredat the sediment-water interface underneath the
farms and dottedlines indicatethe region of hypoxia and normoxia (for further data see http://www.sea.ee/bbg-odss/Ocean/OceanMain). (For interpretationof the references to colourin
this figure legend, the reader is referred to the web version of this article.)
9J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
Outer Baltic where it was estimated at 15 kg m
−1
per harvest. With de-
clining salinity, predicted biomass yields varied between 1 and
3kgm
−1
in the Central Baltic. In marginal (Inner Baltic) regions, pre-
dicted biomass yields never exceeded 1 kg m
−1
. Patterns in modelled
rates of biomass removal are similar to the observed patterns of N and
P content (Fig. 4), which are higher in the Outer Baltic than in either
the Central or Inner Baltic.
The harvest data from the three example farms(Table 1) can be put
into a regional context by comparing them to regional modelled rates of
biomass removal (Supplementary Fig. 3). The observed harvest at
Kumlinge is similar to model projections while the actual harvest at
Sankt Anna is 2.5 times larger than model projections and the Kiel har-
vest is 5 times smaller (Table 1).Mussels grow larger inwaters that are
more saline and therefore their substrate (i.e., the settlement rope) be-
comes quickly saturated and competition between mussels for space is
high. A multi-layered structure of large mussels is very unstable and
even moderate storms can remove outer layers from suspended ropes.
Many detached mussels were observed underneath the Kiel farm,
whereas nosuch losses were recorded in Sankt Anna (Fig. 6). Neverthe-
less, the full potential of high-salinity areas is demonstrated based on
harvest data from experimental fuzzy ropes deployed in Kiel farm
over a year and such substrates hosted nearly 10 kg mussels m
−1
(Sup-
plementary Fig. 4).
When evaluating the potential of mussel cultivation as a mitigation
tool to reach regional nutrient reduction targets, a surprisingly small
marine area would need to be used for mussel farms in order to close
the remaining nutrient reduction gaps, i.e., 900 km
2
for the Central Bal-
tic and 600 km
2
for the Inner Baltic (Fig. 1). When farms are established
at optimal growth locations and optimal density, then nutrient removal
during mussel harvest can compensate up to 100% of the local and a
large part of the regional nutrient loading. Although the modelling re-
sults presented here suggest a higher efficiency of mussel farms at
high salinity, the factual evidence suggests that Outer Baltic farms are
not necessarily more efficient in nutrient removal as compared to Cen-
tral Baltic farms. Importantly, the predicted total area of farms needed
for achieving HELCOM nutrient reduction targets could be achievable
under the current marine spatial planning regime in the Baltic Sea.
4. Discussion
Marine eutrophication is a pervasive and growing threat to global
sustainability (Conley et al., 2009b). While all reasonable efforts to re-
duce nutrient inputs from land to sea must continue, internal measures
are also needed to ensure the timely recovery of eutrophicated systems
(Savchuk, 2018). Extractive harvesting of farmed nativebivalve species,
including M. edulis/trossulus, is a sustainable, low-impact (Petersen
et al., 2014, 2019), circular (Spångberg et al., 2013) and potentially
cost-effective (Gren et al., 2009) internal measure for eutrophication
control (Suplicy, 2018). While arguments have been made against the
use of internal measures such as mussel farming (Stadmark and
Conley, 2011) or geo-engineering (Conley, 2012), there can be little
doubt that internal measures must be considered when all feasible ex-
ternal measures for nutrient load reduction have been explored, applied
and found to be inadequate or insufficient.
The BalticSea is a plausible representationof the likely future stateof
other coastal seas globally (Reusch et al., 2018) and the accumulated
knowledge for this region may serve as a useful future management
model for other internationally managed seas. Mitigation has already
been largely successful for recovery of Baltic Sea top predators
(Reusch et al., 2018) and some fish stocks (Eero et al., 2012). External
loads of both N and P from the surrounding catchment have declined
(Reusch et al., 2018;Savchuk, 2018) but average N concentrations are
decreasing slowly, if at all, while P concentrations continue to increase
(Savchuk, 2018). This mismatch between the successful reduction of
terrestrial nutrient inputs and failure to observe corresponding im-
provements in water column nutrient concentrations is due in part to
the ongoing release of nutrients accumulated in marine sediments
(Vahtera et al., 2007).
The predicted nutrient removal by mussel harvesting largely follows
the spatial patterns of mussel growth, i.e., farms in the Outer Baltic are
expected to have higher yields than in other Baltic Sea regions. Harvest
weight (kg m
−1
) is linearly related to mussel size (Nielsen et al., 2016)
and blue mussels do not grow as rapidly in brackish waters as they doin
fully marine environments. While the small size of harvested mussels
poses challenges for feed or food production, the data presented here
suggest that the overall potential for nutrient removal does notdiminish
along the salinity gradient, except for the innermost parts of the Baltic
Sea (Table 1, Supplementary Fig. 3). While it is important to prioritize
high salinity sites in order to enhance the yield, even at reduced salin-
ities in the central Baltic Sea, a one hectare mussel farm with a density
of appropriate ropes may yield hundreds of tons of biomass per harvest
cycle. Furthermore, the harvest strategy can be optimised as smaller
mussels may be more efficient at nutrient removal due to lower detach-
ment rates asdensity dependentlosses can reach 50% in oceanic regions
with high biomass production (Haamer, 1996).
Unlike earlier studies (Dahlbäck and Gunnarsson, 1981;Hartstein
and Stevens, 2005), the monitoring of all existing mussel farms in the
Baltic Sea region offers no evidence to suggest that blue mussel farms
in the BalticSea have any negative effects on the local oxygen conditions
at the sediment–water interface (Aigars et al., 2019). While others have
suggested that mussel farms can cause promote lower sediment oxygen
concentrations associated with a reduction in bioturbation or excessive
accumulation of organic matter(Stadmark and Conley, 2011), the oppo-
site phenomenon was observed at the Kiel mussel farm where an in-
crease in bioturbation led to higher sediment oxygen concentration
(Aigars et al., 2019). When sediments remain oxygenated, there is un-
likely to be any additional internal loading of P. However, oxygenated
conditions in the sediment under farms can suppress denitrification
(Carlsson et al., 2012).
Shellfish farming generally has lower environmental impacts than
other forms of aquaculture (Forrest et al., 2009;Kraufvelin and Díaz,
2015). Farmed blue mussels do not require any nutrient external inputs.
This means that unlike other forms of aquaculture, all of the nutrients
removed during harvest make a positive contribution to regional eutro-
phication reduction and a valuable regulative ecosystem service in eu-
trophic waters (Suplicy, 2018;Petersen et al., 2019). However, the
potential for localised nutrient enrichment in the immediate vicinity
of mussel farms does exist in very sheltered areas (e.g., Stadmark and
Conley, 2011;Holmer et al., 2015) and in such areas the possibility of
undesirable local eutrophication must be recognised and addressed.
Furthermore, farms can provide additional habitat for colonization
to supplement natural mussel reefs lost to anthropogenic impacts, espe-
cially human-facilitated invasion impacts of benthic predators. In the
Baltic Sea, the most relevant invasive predator is round goby, which
causes large-scale losses of benthic blue-mussel populations
(Skabeikis et al., 2019), e.g. one case-study location is estimated to
have lost 23% of its 230km
2
pre-invasion mussel reef area due to
round goby predation (Liversage et al., 2019). Suspended mussels will
attract negligible predation from such benthic predators, thus mussel
farming will help restore overall population levels. If a switch does
occur from natural mussel reefsto suspended farm mussels, this may in-
volve a reduced local-scale per-capita impact on eutrophication because
material excreted from suspended mussels will have greater dispersal
and dilution by water movements (Hartstein and Stevens, 2005) rather
than direct benthic retention. In addition, aquaculture activities often
produce benthic shell debris deposits (Sanchez-Jerez et al., 2019)
which increase sediment porosity and oxidised sediment layer depth,
as well as infaunal bioturbation (Zaiko et al., 2010). These benefits
may be expected following extended establishment of mussel farms.
Using mussel farmingas an internal measure to mitigate eutrophica-
tion in the Baltic requires the development of appropriate legislative in-
struments (Ozoliņa, 2017) and resolution of sea-use conflicts along
10 J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
maritime spatial planning process (Kannen, 2014). The model-
predicted locationsof mussel farms for achieving eutrophication reduc-
tion targets do not take multiple sea-use conflicts into account, espe-
cially tourism and fisheries (Lindahl et al., 2005). While maritime
spatial planning tools for optimizing interests of various stakeholders
are well developed, thetools do not yet incorporate the implementation
of mussel farming. Careful planning of large-scale mussel farming could
avoid unacceptable environmental impacts or conflicts with other uses.
Farms should be located in semi-exposed or exposed areas with good
water circulation where negative local effects to benthic habitat quality
are unlikely. Additionally, predation can compromise the production of
bivalves in otherwise suitable areas. Therefore, the risk of losing bio-
mass to, e.g., the eider ducks (Somateria mollissima) in the Outer Baltic
must be assessed before initiating a full-scale mussel production.
Other technical challenges including storms, epiphytes, and in some re-
gions ice, will also need to be considered and lessons learnt from previ-
ous mussel farming programmes need to be applied (National Research
Council, 2010). Furthermore, farm technology adapted to the culturing
of small mussels should be used whenever possible to maximize yields.
Blue mussel farming as a mitigation measure is particularly efficient to
counteract diffuse nutrient emissions as to date there are few other ef-
fective options to remove nutrients that have already reached the sea.
Commercial mussel farming can also contribute to rural sustainability
by providing jobs in economically depressed areas. It may also contrib-
ute to a clean-up of the local marine environment with benefits for local
tourism, recreation and other cultural ecosystem services.
5. Conclusions
Eutrophication is a leading cause of impairment of many aquatic
ecosystems globally. While external measures to control nutrient inputs
must be pursued, there is also a need for internal measures in order to
restore water quality and enable ecosystem recovery in a timely man-
ner. Blue mussel farming is a promising low-impact and native
species-based internal method for eutrophication control in the Baltic
Sea and beyond. Mussels filter the water for phytoplankton and trap nu-
trients which are then removed from the aquatic environment through
harvest, allowing nutrient reuse as part of the circular economy. Blue
mussel farming in the Baltic Sea not only provides a tool for nutrient
mitigation, but also contributesto the social and economic sustainability
of rural areas. These results presented here provide factual data to sup-
port political decisions on internal measures for eutrophication control
and promote the sustainability of the Baltic Sea region through mussel
farming for nutrient management.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Acknowledgements
A vast majority of data was collected and the data analyses were
made under the INTERREG project Baltic Blue Growth. In addition, JK,
AK, KL, MR, KH and HOK were supported by the European Union Sev-
enth Framework Programme for research, technological development
and demonstration though BONUS MARES project. ED and PaK were
funded by the project Baltic EcoMussel (EU Central Baltic Interreg IVA
Programme 2012–2013) and PaK additionally by Svenska Kulturfonden
(Carl Cedercreutz stipendiefound) both in 2013 and in 2014. FRB ac-
knowledges the financial support of the German Academic Exchange
Service (DAAD) - Doctoral Programmes in Germany 2015/16 (grant
57129429). PRJ was supported by “The Sea”research program at the
Åbo Akademi University Endowment. EV acknowledges the SmartSea
project (Grant no. 292985). ML and PB were supported by the BONUS
project OPTIMUS - Optimization of Mussel mitigation culture for fish
feed in the Baltic Sea through the Swedish Agency for Marine and Water
Management contract no 4356-2016.
Author contributions
Conceived the study and wrote the paper: JK, MF,AK, KL, LB, PRJ. Col-
lected data: MR, FRB, PB, IB, ED, SK, PaK, PeK, OL, ML, MML, MM, ANS,
HOK, MO, SK, JR, AŠ, EV, HS. Obtained funding and analysed data: JK,
MF, AK, KH, PRJ, AV. All authors discussed the results and edited the
manuscript.
Data availability
The datasets that were generated and/or analysed during the current
study are freely available from the corresponding author on a request.
Appendix A. Supplementary data
Supplementary data to this article can be found onlineat https://doi.
org/10.1016/j.scitotenv.2019.136144.
References
Aigars, J., et al., 2019. Ecological impacts at the small-scale commercial mussel farms in
the Baltic Sea. Interreg Baltic Sea Region Baltic Blue Growth Project. Final Report.
Andersen,J.H., et al., 2017. Long-term temporal and spatialtrends in eutrophication status
of the Baltic Sea. Biol. Rev. 92, 135–149.
Baltic Sea Hydrographic Commission. Baltic Sea Bathymetry Database Version 0.9.3.
Downloaded from http://data.bshc.pro/ on 13.04.2016 (2013).
Bayne, B.L., 1965. Growth and the delay of metamorphosis of the larvae of Mytilus edulis.
Ophelia 2, 1–47.
Bekkby, T., Isachsen, P.E.,Isaeus, M., Bakkestuen, V., 2008. GIS modeling of wave exposure
at the seabed: a depth-attenuated wave exposure model. Mar. Geod. 31, 117–127.
Breitburg, D., et al.,2018. Declining oxygenin the global ocean and coastalwaters. Science
359, eaam7240.
Cai, W.J., et al., 2011. Acidification of subsurface coastal waters enhanced by eutrophica-
tion. Nat. Geosci. 4, 766.
Carlsson, M.S., Engström, P., Lindahl, O., Ljungqvist, L., Petersen, J.K., Svanberg, L., Holmer,
M., 2012. Effects of mussel farms on the benthic nitrogen cycle on the Swedish west
coast. Aquaculture Environment Interactions 2 (2), 177–191 (2012).
Conley, D.J., 2012. Save the Baltic Sea: geoengineering efforts to bring oxygen into the
deep Baltic should be abandoned. Nature 486, 463–465.
Conley, D.J., et al., 2009a. Controlling eutrophication: nitrogen and phosphorus. Science
323, 1014–1015.
Conley, D.J., et al., 2009b. Hypoxia-related processes in the Baltic Sea. Environ. Sci.
Technol. 43, 3412–3420.
Corell, H., Moksnes, P.O., Engqvist, A., Döös, K., Jonsson, P.R., 2012. Larval depth distribu-
tion critically affects dispersal and the efficiency of marine protectedareas. Mar. Ecol.
Prog. Ser. 467, 29–46.
Dahlbäck, B., Gunnarsson, L., 1981. Sedimentation and sulfate reduction under a mussel
culture. Mar. Biol. 63, 269–275.
De Vries, P., Döös, K., 2001. Calculating Lagrangian trajectories using time-dependent ve-
locity fields. J. Atmos. Sci. 18, 1092–1101.
Diaz, R.J., Rosenberg, R., 1995. Marine benthic hypoxia. A review of its ecological effects
and the behavioral responses of benthic macrofauna. Oceanogr. Mar. Biol. Ann. Rev.
33, 245–303.
EC. European Commission, 2012. Blue Growth Opportunities for Marine and Maritime
Sustainable Growth COM/2012/0494 Final.
Eero, M., Köster, F.W., Vinther, M., 2012. Why is the eastern Baltic cod recovering? Mar.
Pol. 36, 235–240.
FAO, 2018. The State of World Fisheries and Aquaculture. Available at. http://www.fao.
org/3/i9540en/I9540EN.pdf.
Fleming-Lehtinen, V., et al., 2015. Recent developments in assessment methodology re-
veal that the Baltic Sea eutrophication problem is expanding.Ecol. Indic. 48, 380–388.
Forrest,B.M., Keeley, N.B.,Hopkins, G.A., Webb,S.C., Clement, D.M.,2009. Bivalve aquacul-
ture in estuaries: review and synthesis of oystercultivation effects.Aquaculture 298,
1–15.
Gren, M., Lindahl, O., Lindqvist, M., 2009. Values of mussel farming for combating eutro-
phication: an application to the Baltic Sea. Ecol. Eng. 35, 935–945.
Haamer, J.,1996. Improving waterquality in a eutrophied fjordsystem with mussel farm-
ing. Ambio 25, 356–362.
Hamilton, H.A., et al., 2018. Trade and the role of non-food commodities for global eutro-
phication. Nature Sustainability 1, 314.
Hartstein, N.D., Stevens, C.L., 2005. Deposition beneath long-line mussel farms. Aquacul-
ture Engi neering 33, 192–213.
Hedberg, N., Kautsky, N., Kumblad, L. & Wikström, S. A. Limitations of using blue mussel
farms as a nutrient reduction measure in the Baltic Sea. Reportfrom Baltic Sea Centre
2/2018. Stockholm: Stockholm University. Available at: https://www.su.se/polopoly_
11J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
fs/1.406424.1539853328!/menu/standard/file/BSC_Mussels_report_2_2 018_lowres.
pdf (2018).
HELCOM, 2015. Manual for Marine Monitoring in the COMBINE Programme of HELCOM.
HELCOM, 2018.HELCOM thematic assessment of eutrophication 2011-2016. Availableat.
http://www.helcom.fi/baltic-sea-trends/holistic-assessments/state-of-the-baltic-sea-
2018/reports-and-materials/.
Helin, J., 2013. Cost-efficient Nutrient Load Reduction in Agriculture: A Short-term Per-
spective on Reducing Nitrogen and Phosphorus in Finland Doctoral Dissertation. Uni-
versity of Helsinki Available at. http://jukuri.luke.fi/handle/10024/481612.
Holmer, M., Thorsen, S.W., Carlsson, M.S., Kjerulf, P.J., 2015. Pelagic and benthic nutrient
regeneration processes in mussel cultures (Mytilus edulis) in a eutrophic coastal
area (Skive Fjord, Denmark). Estuar. Coasts 38, 1629–1641 (2015).
Hordoir, R., Dieterich, C., Basu, B., Dietze, H., Meier, H.E.M., 2013. Freshwater outflow of
the Baltic Sea and transport in the Norwegian current: a statistical correlation analy-
sis based on a numerical experiment. Cont. Shelf Res. 64, 1–9.
Hordoir, R., Axell, L., Löptien, U., Dietze, H., Kuznetsov, I., 2015. Influence of sea level rise
on the dynamics of salt inflows in the Baltic Sea. J. Geophys. Res. - Oceans 120,
6653–6668.
Hu, S., Wang, Y., Han, H., 2011. Utilization of waste freshwater mussel shell as an eco-
nomic catalyst for biodiesel production. Biomass Bioenergy 35, 3627–3635.
Hurrel, J.W., Deser, C., 2009. North Atlantic climate variability: the role ofthe North Atlan-
tic oscillation. J. Mar. Syst. 78, 28–41.
Kannen, A., 2014. Challenges for marine spatial planning in the context of multiple sea
uses, policy arenas and actors based on experiences from the German North Sea.
Reg. Environ. Chang. 14, 2139–2150.
Kautsky, N., 1982. Quantitative studies on gonad cycle, fecundity, reproductive output
and recruitment in a Baltic Mytilus edulis population. Mar. Biol. 68, 143–160.
Kiessling, A., Futter,M.N., Elofsson,K., Vidakovic, A., 2019. Musselfarming in the BalticSea
as an environmental measure. Availabe at. https://www.slu.se/globalassets/ew/org/
centrb/aquaculture/dokument/policy-brief-new-data-on-musselfarming-as-environ-
mental-service-in-baltic-proper-eng.pdf.
Kotta, J., et al., 2015. Establishing functional relationships between abiotic environment,
macrophyte coverage, resource gradients and the distribution of Mytilus trossulus in
a brackish non-tidal environment. PLoS One 10, e0136949.
Kraufvelin, P., Díaz, E.R., 2015. Sediment macrofauna communities ata small mussel farm
in the northern Baltic proper. Bor. Environ. Res. 20, 378–390.
Larsson, M., Granstedt, A., 2010. Sustainable governance of the agriculture and the Baltic
Sea—agricultural reforms, foodproduction and curbed eutrophication. Ecol. Econ. 69,
1943–1951.
Levitus, S., Boyer, T.P., 1984. World Ocean atlas. Salinity. Volume 5. NOAA atlas.
Lindahl, O., et al., 2005. Improving marine water quality by mussel farming: a profitable
solution for Swedish society. Ambio 34, 131–138.
Liversage, K., Kotta, J., Aps, R., Fetissov, M., Nurkse, K., Orav-Kotta, H., Rätsep, M.,
Forsström, T., Fowler, A., Lehtiniemi, M., Normant-Saremba, M., Puntila-Dodd, R.,
Arula, T., Hubel, K., Ojaveer, H., 2019. Knowledge to decision in dynamic seas:
methods to incorporate non-indigenous species into cumulative impact assessments
for maritime spatial planning. Sci. Total Environ. 658, 1452–1464.
Lurling, M., Mackay, E.B., Reitzel, K., Spears, B.M., 2016. A critical perspective on geo-
engineering for eutrophication management in lakes. Water Res. 97, 1–10.
Maar, M., Saurel, C., Landes, A., Dolmer, P., Petersen, J.K., 2015. Growth potential of blue
mussels (M. edulis) exposed to different salinities evaluated by a dynamic energy
budget model. J. Mar. Syst. 148, 48–55.
MacNally,R., 2000. Regression and model-building in conservationbiology, biogeography
and ecology: the distinction between −and reconciliation of −‘predictive’and ‘ex-
planatory’models. Biodivers. Conserv. 9, 655–671.
Madec, G., 2010. Nemo Ocean Engine, Version 3.3, Tech. Rep. IPSL.
McLaughlan, C., Rose, P., Aldridge, D.C., 2014. Making the best of a pest: the potential for
using invasive zebra mussel (Dreissena polymorpha) biomass as a supplement to
commercial chicken feed. Environ. Manag. 54, 1102–1109.
NationalResearch Council, 2010.Ecosystem Concepts forSustainable BivalveMariculture.
The National Academies Press, Washington, DC https://doi.org/10.17226/12802.
Naylor, R.L., Williams, S.L., Strong, D.R., 2001. Aquaculture –a gateway for exotic species.
Science 294, 1655–1656.
Nielsen, P., Cranford, P.J., Maar, M., Petersen, J.K., 2016.Magnitude, spatial scale and opti-
mization of ecosystem services from a nutrient extraction mussel farm in the eutro-
phic Skive Fjord, Denmark. Aquacult. Env. Interac. 8, 311–329.
Nkemka, V.N., Murto, M., 2013. Two-stage anaerobic dry digestion of blue mussel and
reed. Renew. Energ. 50, 359–364.
Ozoliņa, Z., 2017. Mussel farming and its potential in the Baltic Sea. Econ. Business 30,
40–50.
Petersen, J.K., et al., 2014. Mussels as a tool for mitigation of nutrients in the marine envi-
ronment. Mar. Pollut. Bull. 82, 137–143.
Petersen, J.K., Holmer, M., Termansen, M., Hasler, B., 2019. Nutrient extraction through bi-
valves. In: Smaal, A.C., Ferreira, J.G., Grant, J., Petersen, J.K., Strand, Ø. (Eds.), Goods
and Services of Marine Bivalves. Springer, Cham, pp. 179–208.
Rabalais, N.N., Turner, R.E., Diaz, R.J., Justić, D., 2009. Global change and eutrophication of
coastal waters. ICES J. Mar. Sci. 66, 1528–1537.
Reusch, T.B., et al., 2018. The Baltic Sea as a time machine for the future coastal ocean. Sci.
Adv. 4, eaar8195.
Riedel, B., Diaz, R., Rosenberg, R., Stachowitsch, M., 2016. The ecological consequences of
marine hypoxia: From behavioural to ecosystem responses. In: Solan, M., Whiteley,
N.M. (Eds.), Stressors in the Marine Environment: Physiological Responses and Eco-
logical Implication. Oxford University Press, Oxford, pp. 175–194.
Riisgård,H.U., Larsen, P.S., Turja, R., Lundgreen, K., 2014. Dwarfism of blue mussels in the
low saline Baltic Sea—growth to the lower salinity limit. Mar. Ecol. Prog. Ser. 517,
181–192.
Rydin, E., Kumblad, L., Wulff, F., Larsson, P., 2017. Remediation of a eutrophic bay in the
Baltic Sea. Environ. Sci. Technol. 51, 4559–4566.
Sanchez-Jerez, P., Krüger, L., Casado-Coy, N.,Valle, C., Sanz-Lazaro, C., 2019. Mollusk shell
debris accumulation in the seabed derived from coastal fish farming. Journal of Ma-
rine Science and Engineering 7 (10), 335.
Savchuk, O.P., 2018. Large-scale nutrient dynamics in the Baltic Sea, 1970–2016. Front.
Mar. Sci. 5, 95.
Schröder, T., Stank, J., Schernewski, G., Krost, P., 2014. The impact of a mussel farm on
water transparency in the Kiel Fjord. Ocean Coast. Manage. 101, 42–52.
Sipiä, V.O., Kankaanpää, H.T., Flinkman, J., Lahti, K., Meriluoto, J.A., 2001. Time-dependent
accumulation of cyanobacterial hepatotoxins in flounders (Platichthys flesus) and
mussels (Mytilus edulis) from the northern Baltic Sea. Environ. Toxicol. 16, 330–336.
Skabeikis, A., Morkūnė, R., Bacevičius, E., Lesutienė, J., Morkūnas, J., Poškienė,A.,Šiaulys,
A., 2019. Effect of round goby (Neogobius melanostomus) invasion on blue mussel
(Mytilus edulis trossulus) population and winter diet of the long-tailed duck
(Clangula hyemalis). Biol. Invasions 21 (3), 911–923.
Smith, V.H., 2003. Eutrophication of freshwater and coastal marine ecosystems a global
problem. Environ. Sci. Pollut. R. 10, 126–139.
Spångberg, J., Jönsson, H., Tidåker, P., 2013. Bringing nutrients from sea to land–mussels
as fertiliser from a life cycle perspective. J. Clean. Prod. 51, 234–244.
Stadmark, J., Conley,D.J., 2011. Musselfarming as a nutrient reduction measure in the Bal-
tic Sea: consideration of nutrient biogeochemical cycles. Mar. Pollut. Bull. 62,
1385–1388.
Stigebrandt, A., et al., 2015. An experiment with forced oxygenation of the deepwater of
the anoxic By Fjord, western Sweden. Ambio 44, 42–54.
Stuckas, H., Stoof, K., Quesada, H., Tiedemann, R., 2009. Evolutionary implications of dis-
cordant clines across the Baltic Mytilus hybrid zone (Mytilus edulis,Mytilus trossulus).
Heredity 103, 146–156.
Suplicy, F.M., 2018. A review of the multiple benefits of mussel farming. Rev. Aquacult.
https://doi.org/10.1111/raq.12313.
Uppala, S.M., et al., 2005. The ERA-40 re-analysis. Q. J. Roy. Meteor. Soc. 131, 2961–3012.
Vahtera, E., et al., 2007. Internal ecosystem feedbacks enhance nitrogen-fixing
cyanobacteria blooms and complicate management in the Baltic Sea. Ambio 36,
186–194.
Vidakovic, A., et al., 2015. Evaluation of growth performance and intestinal barrier func-
tion in Arctic Charr (Salvelinus alpinus) fed yeast (Saccharomyces cerevisiae), fungi
(Rhizopus oryzae) and blue mussel (Mytilus edulis). Aquac. Nutr. 22, 1348–1360.
Westerbom, M., 2006. Population Dynamics of Blue Mussels in a Variable Environment at
the Edge ofTheir Range. Doctoral thesis.Faculty of Biosciences, Department of Biolog-
ical and Environmental Sciences, University of Helsinki, p. 62 (2006).
Wijkmark, N., Isæus, M., 2010. Wave exposure calculations for the Baltic Sea. AquaBiota
Water Research AB, Stockholm, Report, 2.
Zaiko, A.,Paskauskas, R., Krevs, A., 2010.Biogeochemicalalteration of the benthic environ-
ment by the zebra mussel Dreissena polymorpha (Pallas). Oceanologia 52 (4),
649–667.
12 J. Kotta et al. / Science of the Total Environment 709 (2020) 136144
