Content uploaded by Simon Weppe
Author content
All content in this area was uploaded by Simon Weppe on Jun 29, 2020
Content may be subject to copyright.
AQUACULTURE ENVIRONMENT INTERACTIONS
Aquacult Environ Interact
Vol. 12: 231–249, 2020
https://doi.org/10.3354/aei00363 Published online June 18
1. INTRODUCTION
At a global scale, coastal shellfish aquaculture is
rapidly expanding in order to help meet the protein
needs of a growing human population (FAO 2018).
As cultured biomass of shellfish increases, it will in -
evitably affect ecosystem dynamics in the local area in
which this culturing is taking place (van der Schatte
Olivier et al. 2020). These aquaculture−environment
interactions may be perceived as either negative or
positive. Effects viewed as negative may include the
organic enrichment of the benthos through biode-
posits (Giles & Pilditch 2006, Cranford et al. 2009),
alteration of the genetic composition of populations
through the escape of larvae or adults (Gausen &
Moen 1991, Apte et al. 2003, Heino et al. 2015), or the
© The authors 2020. Open Access under Creative Commons by
Attribution Licence. Use, distribution and reproduction are un -
restricted. Authors and original publication must be credited.
Publisher: Inter-Research · www.int-res.com
*Corresponding author: craig.norrie@oregonstate.edu
Spill-over from aquaculture may provide a larval
subsidy for the restoration of mussel reefs
Craig Norrie1, 6,*, Brendon Dunphy1, 2, Moninya Roughan3, Simon Weppe4,
Carolyn Lundquist1, 5
1Institute of Marine Science, University of Auckland, Auckland 1010, New Zealand
2School of Biological Sciences, University of Auckland, Auckland 1010, New Zealand
3School of Mathematics and Statistics, UNSW Sydney, NSW 2052, Australia
4Meteorological Service of New Zealand (MetOcean Division), Raglan 3225, New Zealand
5National Institute of Water and Atmospheric Research (NIWA), Hamilton 3216, New Zealand
6Present address: Hatfield Marine Science Center, Cooperative Institute for Marine Resources Studies,
Oregon State University, Newport, OR 97365, USA
ABSTRACT: Worldwide bivalve aquaculture is expanding rapidly. Simultaneously, there has been
a loss of natural bivalve reefs due to anthropogenic activities. As bivalve reefs support several eco-
system functions disproportionate to the area they cover, there is interest in their restoration. The
Firth of Thames (FoT) in northern New Zealand once supported dense populations of green lipped
mussels Perna canaliculus, which were extirpated by a dredge fishery in the mid-20th century.
Efforts to restore these biogenic habitats are underway. The largest standing populations of this
species in the area currently exist in aquaculture. This study aimed to determine if larval spill-over
from aquaculture can provide a larval subsidy to bivalve reef restoration efforts in the FoT. We
used a combination of trace elemental fingerprinting and biophysical modelling techniques to
determine patterns of larval dispersal in the area. Results of both approaches indicated that the
larval pool in the area is well mixed with larvae produced at aquaculture locations capable of set-
tling throughout the study area. Overall this shows, for the first time, that larval spill-over from
aquaculture may provide a subsidy to restoration efforts and assist with establishing sustainable
populations. When determining restoration locations, the potential for aquaculture populations to
act as a larval source should be explicitly considered. Conversely, when considering the location
of new aquaculture sites, the consequences of larval spill-over to surrounding wild populations
should be assessed. We recommend that restoration efforts and aquaculture be carefully inte-
grated in a network approach which could provide both ecological and economic benefits.
KEY WORDS: Larval dispersal · Population connectivity · Bivalve larvae · Biophysical modelling ·
Trace elemental fingerprinting · Restoration · OpenDrift · Shell chemistry
O
PEN
PEN
A
CCESS
CCESS
This authors' personal copy may not be publicly or systematically copied or distributed, or posted on the Open Web,
except with written permission of the copyright holder(s). It may be distributed to interested individuals on request.
Aquacult Environ Interact 12: 231–249, 2020
alteration of flow regimes by the structures on which
aquaculture is taking place (Gibbs et al. 1991, Plew
2011). Effects perceived as positive may involve nutri-
ent remediation in eutrophied systems (Petersen et
al. 2014, Nielsen et al. 2016) or providing habitat for
other species (McLeod et al. 2014, Fariñas-Franco &
Roberts 2018). Although often overlooked in studies
investigating the effects of aquaculture on the local
environment (see reviews by Newell 2004, van der
Schatte Olivier et al. 2020), larval production from
these dense aggregations of bivalves may contribute
to regional larval pools and thereby modify popula-
tion dynamics in the area in which they are growing.
Knowledge of potential larval dispersal pathways
from these cultured populations is essential to gain
an understanding of how larvae from aquaculture
may interact with local wild bivalve populations.
Understanding these interactions is of particular
interest in areas where natural populations of
bivalves have been depleted but cultured popula-
tions exist; e.g. the Firth of Thames (FoT) in northern
New Zealand for green lipped mussel or Chesapeake
Bay for Virginia oysters (Beck et al. 2011, Turner et
al. 2019).
Shellfish beds perform a number of ecosystem
functions disproportionate to their size (Lundquist et
al. 2017). They contribute to water filtration (zu
Ermgassen et al. 2013, Ehrich & Harris 2015), nutri-
ent cycling (Giles & Pilditch 2006, Nielsen et al.
2016), provide habitat structure (Dumbauld et al.
2009), and can influence food web dynamics and bio-
diversity (McLeod et al. 2014, Fariñas-Franco &
Roberts 2018). However, several common anthro-
pogenic stressors are recognised worldwide which
have resulted in degradation or removal of these bio-
genic habitats with a corresponding loss of ecosys-
tem functions (Beck et al. 2011, Alleway & Connell
2015). These stressors include sedimentation, eutro -
phication, overfishing, and dredging. Accordingly,
there is interest in restoring degraded populations
and the ecosystem functions they provide (e.g.
French McCay et al. 2003, Coen et al. 2007, North et
al. 2010). In order for restoration programmes to have
the highest chances of success, an understanding of
larval dispersal and population connectivity is essen-
tial (Lipcius et al. 2008). Understanding larval dis-
persal dynamics allows us to determine the self-
sustainability of restored populations and also select
optimal restoration locations (Elsäßer et al. 2013, Tet-
telbach et al. 2013, Puckett et al. 2018).
Although it is important to track the dispersal of
larvae, in situ tracking of larval movement is complex
due to the small size, high mortality, and large scales
over which larval transport is possible (Gawarkiewicz
et al. 2007, Cook et al. 2014). Thus, several tech-
niques have been developed which allow larval dis-
persal and population connectivity to be indirectly
inferred (reviewed by Kool et al. 2013). Currently,
biophysical models (e.g. Cowen et al. 2000, Werner
et al. 2007, Elsäßer et al. 2013, Thomas et al. 2016)
and trace elemental fingerprinting (e.g. Thorrold et
al. 2002, Becker et al. 2007, Ricardo et al. 2015, Gomes
et al. 2016) are among the most widely applied tech-
niques used to infer the dispersal of larvae. Biophys-
ical modelling couples oceanographic data (e.g.
ocean currents and temperature) with information on
the biology and behaviour of larvae to estimate larval
dispersal over spatial and temporal scales that would
be impossible to measure empirically. Biophysical
models have been used in a number of management
scenarios, such as the design of marine reserves
(Puckett et al. 2014), examining larval export from
marine reserves (Le Port et al. 2014), determining
restoration locations (Tettelbach et al. 2013, Puckett
et al. 2018), predicting the spread of invasive species
(Inglis et al. 2006), and projecting changes under
future climate scenarios (Cetina-Heredia et al. 2015,
Coleman et al. 2017). Trace elemental fingerprinting
is based on the premise that carbonate structures
such as shell or otolith deposited in water masses
with differing physiochemical properties reflect these
differences in the trace elemental composition of
these structures (e.g. Cathey et al. 2012, Kroll et al.
2016, Norrie et al. 2019). Through the sequential
analysis of these structures, the movement of individu-
als can be inferred (e.g. Lazareth et al. 2003, Kroll et al.
2018).
Each method of indirectly estimating larval dis-
persal, however, comes with intrinsic uncertainties
(Ashford et al. 2010, Nolasco et al. 2018). Biophysical
particle tracking models, for example, are limited
by the availability and accuracy of the hydrody-
namic and biological data used in their para -
meterisation (Cetina-Heredia et al. 2019). In con-
trast, trace elemental fingerprinting may be limited
by the characterisation of all potential natal loca-
tions (Werner et al. 2007). Therefore, the incorpora-
tion of larval dispersal estimates obtained though
different methods is likely to provide a more accu-
rate picture of larval dispersal. Here, we incorpo-
rated results from both trace elemental fingerprint-
ing and biophysical models to track the dispersal of
green lipped mussels Perna canaliculus (Gmelin,
1791), hereafter mussels, in the FoT in northern
New Zealand, an area with an active shellfish res-
toration effort (Wilcox et al. 2018).
232
Author copy
Norrie et al.: Larval dispersal from bivalve aquaculture
The FoT once supported dense populations of mus-
sels, which were estimated to cover up to 1300 km2;
these were almost completely extirpated by a dredge
fishery in the 1960s (Paul 2012). Despite the cessation
of fishing 50 yr ago, these populations have not
recovered and only a few small remnant wild mussel
beds remain, which cover an area of approximately
0.64 km2(Morrison 2002, McLeod 2009). Currently,
the largest populations of mussels in the FoT are
those farmed on longline aquaculture, which covers
an area of approximately 29 km2and produced
27 196 t of harvested mussels in 2016 (Hauraki Gulf
Forum 2017). There are currently active restoration
efforts in the area which focus on the translocation of
adults and juveniles, primarily from aquaculture,
with the aim of establishing self-sufficient popula-
tions and recovering lost ecosystem function (Wilcox
et al. 2018). Understanding the potential for larval
spill-over from aquaculture to contribute to restora-
tion efforts will provide important information for the
design of restoration programmes in areas where
aquaculture is present, as it will in form locations for
mussel translocations.
The aim of this study was to examine the potential
for larval spill-over of mussels from aquaculture to
provide a larval subsidy to the restoration efforts in
the FoT. This was achieved using a combination of
elemental fingerprinting techniques and biophysical
modelling approaches. We addressed the following
specific questions: (1) What is the potential for disper-
sal of mussles from aquaculture locations throughout
the FoT? (2) How do larval dispersal patterns gener-
ated by 2 different methods, biophysical modelling
and trace elemental fingerprinting, differ? (3) What
are the overall implications of any larval spill-over for
restoration and aquaculture?
2. MATERIALS AND METHODS
2.1. Study species
Green lipped mussels Perna canaliculus are found
in a variety of habitats throughout New Zealand,
both subtidally and intertidally (Morton & Miller
1973). Mussels are broadcast spawners with peak
spawning occurring from late in the austral spring to
early autumn (Alfaro et al. 2001), have a pelagic lar-
val duration of between 3 and 5 wk (Hayden 1995),
and prefer to settle on filamentous substrates such as
macroalgae and hydroids (Alfaro & Jeffs 2002, Alfaro
et al. 2004). Settlement is complete once byssal
threads have attached, after which mussels are
known as plantigrades or spat. Secondary settlement
may then occur through bysso-pelagic drifting from
primary settlement substrate into adult beds (Jeffs et
al. 1999). The distances over which secondary disper-
sal takes place, however, is likely to be orders of
magnitude lower than primary dispersal in the
plankton (Lane et al. 1985, Le Corre et al. 2013,
Pilditch et al. 2015).
2.2. Study area
The FoT is a large estuarine embayment in north
eastern New Zealand, approximately 30 km long and
20 km wide (Fig. 1). Circulation in the region is
tidally driven and predominantly flows north−south,
with current speeds on the ebb and flood tides gen-
erally well balanced (Supplement 1 at www.int-res.
com/ articles/ suppl/ q012 p231 _ supp.pdf [all supple-
ments], Black et al. 2000). East−west flows also occur
but are typically around 10 times weaker than the
north−south currents (Oldman et al. 2007). The FoT
is generally thought to be well mixed with limited
seasonal stratification in late summer (Black et al.
2000, Zeldis et al. 2005). Primary freshwater inputs
are from the Waihou and Piako Rivers, which drain
into the south eastern FoT. These rivers have a com-
bined mean flow of 89 m3 s−1 (O’Callaghan & Stevens
2017). The area zoned for mussel aquaculture oc -
cupies an area of 29 km2and accounts for 30% of
New Zealand’s GreenshellTM (the trade name of P.
canaliculus) mussel production (Hauraki Gulf Forum
2017). These farms are primarily located on the
northern end of the Coromandel Peninsula centred
on Wilson Bay. There are additional farms located
within and around Coromandel Harbour and Man-
aia Harbour and on the western FoT at Waimang
Point.
2.3. Biophysical particle tracking
2.3.1. Hydrodynamic model
An 11 yr (1995−2005) hindcast hydrodynamic
model configuration of the regional ocean modelling
system (ROMS), developed by the Meteorological
Service of New Zealand (MetOcean Division), was
used in this study. For a full description of this model
and its validation, see Supplement 1. Briefly, the
model has a spatial grid resolution of 250 × 250 m
over the study area, with hourly model output. Both
atmospheric and tidal forcing were included in the
233
Author copy
Aquacult Environ Interact 12: 231–249, 2020
model, which used bathymetry sourced from Land In-
formation New Zealand (https:// data. linz. govt. nz/ layer/
51278-chart-nz-533-firth-of-thames/). Atmospher ic
forcing was applied at a temporal resolution of 60 min
and spatial resolution of 1 km. Due to the limited
stratification in the FoT (Black et al. 2000, Zeldis et al.
2005), a 2-dimensional approach was used which was
followed by a post-process transformation to obtain
estimates of the depth-dependent currents. A logarith-
mic velocity profile was added to depth-averaged
currents to simulate variations in horizontal water
flow (Le Port et al. 2014). This high-resolution model
was nested within a larger regional model of New
Zealand-wide circulation at a grid size of 5 × 5 km.
The high-resolution model was validated using tidal
elevations obtained from a tide gauge at Tiritiri
Matangi, located approximately 50 km northwest of
the study site (36.605° S, 174.888° E) (Supplement 1).
2.3.2. Lagrangian particle tracking
Using the open source Lagrangian particle track-
ing software OpenDrift v.1.0 (Dagestad et al. 2018),
we simulated individual dispersal trajectories for
particles (representing recently spawned larvae)
re leased from 5 selected aquaculture sites through-
out the FoT (COR, MAN, WP, WB1, WB2; Fig. 1).
Sites were selected to represent areas at which
mussel aquaculture occurs in the FoT. Two sites
were within (COR) or close to the entrance (MAN)
of sheltered harbours; 2 sites represented the large
exposed blocks of mussel farms in the eastern FoT
(WB1 and WB2), and one site represented the
small mussel farms in the western FoT (WP). The
OpenDrift particle trajectory model was coupled to
the hydrodynamic model. A new OpenDrift sub-
model based on the already available ‘PelagicEgg’
234
MAN
COR
WB1
WB2
WP
Firth of
Thames
Waiheke Is.
Ponui Is.
100 km0
New Zealand
1
2
1
2
Manaia Harbour
Coromandel
Harbour
N
N
36° 50`
S
36° 50`
36° 40`
37° 10`
37°
36° 59`
S
175° 28` E
175° 20` 175° 30`175° 10`
175° 30` E
37° 20`
36° 30`
S
175° E
C6
C5
C4
C7
Wilson Bay
Thames
C3
C1
C2
Waihou
River
Piako
River W1
Waimango
Point
C4
20 km 10 km
2.5 km
Fig. 1. The Firth of Thames in northern New Zealand, indicating sites used in this study. Blue rectangles: locations from which
particles were released for biophysical modelling; red points: locations at which trace elemental fingerprinting sampling of
Perna canaliculus was undertaken
Author copy
Norrie et al.: Larval dispersal from bivalve aquaculture
module was de veloped specifically for mussel larvae
(‘Bivalvelarvae’ model) and is available at https://
github. com/ simonweppe/ opendrift/ blob/ master/
opendrift/ models/ bivalvelarvae. py.
The model operated with an internal timestep
of 15 min. A horizontal diffusion coefficient of
0.1176 m2s−1 was included based on the equations
of Okubo & Ebbesmeyer (1976) to account for
subgrid- scale diffusion. A vertical diffusion con-
stant of 0.01 m2s−1 over a 15 min timestep was in -
cluded to account for turbulence in the fluid envi-
ronment (based on parameters previously used for
bivalve larval dispersal in Lundquist et al. 2004). A
total of 4.9 million particles were released between
the months of December and June, when peak set-
tlement occurred over the years 1995−2005 (Alfaro
et al. 2001, Smith 2019). Release sites were selected
to represent all general areas in which mussel
aquaculture occurs. Particles were evenly released
within polygons delineated by mussel farming oper-
ations (Fig. 1). As management and harvest prac-
tices are consistent within New Zealand (Jeffs et al.
1999) and therefore across the study area, the num-
ber of particles re leased from each site was scaled
to the surface area occupied by each farm. Particles
were released at a density of 10 particles km−2 of
surface area of mussel farm at 2 h intervals to
ensure particles were seeded across the tidal cycle.
Full model parameters are presented in Table 1.
Particles were seeded at random depths within the
water column between 1 and 10 m to simulate the
depth of the dropper lines on which mussels are culti-
vated (Jeffs et al. 1999). No robust information exists
on whether these mussels exhibit active horizontal or
vertical swimming behaviour, though New Zealand
bivalve larvae have been observed to be well distrib-
uted throughout the water column in a large harbour
(Lundquist & Broekhuizen 2012). Particles were
therefore given a terminal sinking velocity of 0.0025
m s−1, which is at the lower range of settling velocities
of larger bivalve larvae and may take into account
some active settlement behaviour (Chia et al. 1984,
Lundquist et al. 2009). Particles were released every
2 h for the full 7 mo settling period (Dec−June) every
year from 1995−2005 and tracked for up to 5 wk.
Age-dependent settlement was included in the model;
if a particle encountered the seafloor or the coast
(settlement habitat) between 3 and 5 wk after release
(the settlement competency window) (Jeffs et al.
1999), it was deemed to have settled and remained at
this location. Particles which did not en counter set-
tlement habitat within this period were re tired from
the model as this is beyond the known pelagic larval
duration of this species (Hayden 1995). Due to the
unavailability of data regarding the survival of mussel
larvae, we assumed that mortality was spatially and
temporally uniform. Thus, we did not include potential
variations in daily mortality rates, but rather report
proportional larval transport from each site.
2.3.3. Biophysical model analysis
Analysis of the OpenDrift output ‘netcdf’ files was
conducted in MATLAB R2018a (MathWorks). Kernel
density estimations (KDEs) were calculated using the
methods of Botev et al. (2010). KDEs represented the
probability that a released particle would settle in a
given location. A total KDE was calculated for all par-
ticles released from all sites in the model. As the total
KDE was dominated by particles released from areas
with high surface area (WB1 and WB2), individual
KDEs for particles released from each site were also
calculated. The total proportion of particles which
encountered suitable settlement habitat relative to
the total number of particles released from each site
over the 11 yr modelled period was also calculated.
2.4. Trace elemental fingerprinting
2.4.1. Study sites
A total of 8 sites within the FoT were selected for
trace elemental fingerprinting (C1−C7, W1; Fig. 1).
Four of these sites were selected as they were within
235
Variable Value
Horizontal diffusivity 0.1176 m2s−1
Vertical diffusivity 0.01 m2s−1
Turbulent mixing timestep 900 s
Terminal fall velocity 0.0025 m s−1
Minimum settlement age 3 wk
Maximum settlement age 5 wk
Internal timestep 15 min
External timestep 6 h
Density of particles 1 particle per 500 m2of aqua-
released per timestep culture surface area
Particle release depth Randomly between 1 and
10 m depth
Particle release timestep Every 2 h during release
months and years
Release months December−April
Years modelled 1995−2005
Table 1. Parameters used in biophysical modelling of Perna
canaliculus in the Firth of Thames
Author copy
Aquacult Environ Interact 12: 231–249, 2020
236
areas used for longline aquaculture and therefore
potential larval sources. These included 3 of the mod-
elled release sites (C3, C5, and W1). One of these
aquaculture locations was situated in the western
FoT at Waimang Point (W1), 2 were situated within
Coromandel Harbour (C5 and C6), and 1 was situ-
ated within the large mussel farm blocks at Wilson
Bay (C3). The remaining sites were selected as pre-
liminary hydrodynamic modelling indicated they were
within distances over which larvae from these aqua-
culture sites were able to disperse. Additionally, due
to significant and varied anthropogenic modi fication
of the study area over the last 100 yr (Hauraki Gulf
Forum 2017), shell material produced at these sites is
likely to differ in its trace elemental composition. Sites
were generally between 2 and 5 km apart, although
the largest distance between sites was 14.6 km
(W1−C3) and the shortest was 1.5 km (C5−C6).
2.4.2. Field sampling
To determine natural variation in the trace elemen-
tal composition of mussels and establish reference
trace elemental fingerprints within the study area, an
in situ culturing technique was used (e.g. Becker et
al. 2007, Kroll et al. 2018). Whilst previous studies
have employed both moored and drifting in situ cul-
turing methods to develop reference trace elemental
fingerprints with differing results, it is difficult to
establish which is more accurate (Kroll et al. 2018).
We therefore elected to use a moored in situ cultur-
ing method for the development of reference trace
elemental fingerprints. Macroalgae with juvenile
mussels attached were harvested from the low inter-
tidal zone at M ori Bay on the west coast of northern
New Zealand (36.837° S, 174.427° E). This site was
selected due to consistent year-round supply of juve-
niles and spat. Mussels were transported to the
School of Biological Sciences at The University of
Auckland, where individuals between 2 and 8 mm
shell height were removed from the macroalgae and
placed into in situ culturing containers (ICCs).
Although incorporation of elements into shell mate-
rial may vary with age (Strasser et al. 2008), previous
research has shown that these ontogenetic differ-
ences in trace elemental fingerprints are unlikely to
mask spatial differences (Norrie et al. 2016). ICCs
were constructed from 750 ml food-grade, high-
density polyethylene jars with the lids and bases
removed. Each end of the jar was covered with
500 µm Nitex®mesh. This mesh size allowed for flow
of water carrying nutrients but prevented the escape
of animals. At least 20 juveniles were placed into
each ICC, however up to 30 were translocated if
enough animals were collected.
At each site, artificial settlement substrates (SSs)
were deployed in order to collect recent settlers at
each site. Plastic mesh tuffy scrubbing pads (SOS
Tuffy, Clorox) were used as they have been exten-
sively employed in previous settlement studies (re -
viewed by South 2016). Settlement on these substrates
is widely accepted to be a proxy for larval supply
(South 2016). These SSs were deployed with ICCs to
form an experimental unit. To create an experimental
unit, one ICC was attached to a rope at a depth of 7 m
with one SS approximately 10 cm above the ICC and
one 10 cm below. At each site, 3 of these experimental
units were deployed on each sampling event (nSS = 3,
nICC = 3 site−1 per sampling event). Experimental units
were deployed for ap proximately 5 wk, although in-
clement weather events resulted in different deploy-
ment lengths on each sampling occasion (ranging
from 29−55 d) and the loss of all equipment at some
sites for some dates. All sampling occurred over the
austral summer and autumn (Jan−May) in a single
year (2018), as this is a period of high settlement in the
study area (Smith 2019). After retrieval, animals were
removed from each ICC, placed into plastic zip lock
bags, and frozen at −20°C until analysis. Each SS was
placed into a zip lock bag in its entirety, as mussels
are easier to remove from settlement substrate after
freezing and thawing (Smith 2019).
2.4.3. Sample preparation for trace elemental
analysis
Mussels from each ICC were defrosted and the
shells were split open. One valve was randomly
selected from each individual and any adhering par-
ticles were removed using stainless steel forceps. We
inspected shells prior to translocation to ensure that
shell material was a dark green. After recovery of
ICCs, shell material with a distinct lighter green
colour band following dark colourations was deemed
to have been deposited in situ (Fig. 2a). The most
recently formed portion of shell with a light green
colour band along the axis of maximum growth was
broken off and mounted onto microscope slides using
double sided adhesive tape. If no distinct colour
changes were observed, shells were not used for ele-
mental analysis.
Settlers collected from 3 SSs at each site on each
sampling occasion were analysed. In the laboratory,
each SS was defrosted and all material was washed
Author copy
Norrie et al.: Larval dispersal from bivalve aquaculture
with tap water through a 200 µm sieve. All material
washed off each SS was placed into deionised water
(maintained at a resistivity of <1 MΩcm), and mus-
sels were sorted from this material. Mussels from
each SS were selected and fixed to a microscope
slide with double sided adhesive tape.
2.4.4. Trace element analysis
To determine the trace elemental composition of shell
material, laser ablation inductively coupled plasma
mass spectrometry (LA-ICP-MS) was used. Elemental
analyses were performed using a New Wave deep
ultraviolet (193 nm) laser ablation system (Elemental
Scientific Industries) coupled to an Agilent 7700 ICP-
MS (Agilent Technologies). The laser operated with a
spot size of 50 µm, a repetition rate of 5 Hz, and a
dwell time of 30 s. Laser power was 45% and fluence
was between 7 and 7.5 J cm−2. National Institute of
Standards and Tech nology (NIST) 610 and 612 glass
standards were analysed every 20 spots for standardi-
sation, calculation of internal precision, and calibra-
tion purposes. For full laser operating parameters and
detection limits, see Supplement 2 (Tables S2 & S3).
All analyses were performed at the University of
Auckland Plasma Mass Spectrometry Centre. Initially,
13 elements were monitored (Mn, Li, Co, Mg, B, Ca, Sr,
Zn, Cu, Ti, Ni, Ba, Al). However, preliminary experi-
ments revealed potential contamination from ICCs of
Li, B, and Cu in shell cultured in these containers (Sup-
plement 3), therefore these elements were excluded
from further analyses. On reference shell material de -
posited in ICCs, one LA-ICP-MS spot was performed
on the section of broken off shell approximately 200 µm
from the ventral margin of shells (Fig. 2a). This mate-
rial represented the most recently formed shell. On
naturally settled individuals collected on the SS, the
prodissoconch was analysed (Fig. 2b), as this reflects
shell material formed at the individual’s natal loca-
tion (Becker et al. 2007, Carson et al. 2011).
Data was processed using IOLITE trace elemental
reduction software (Paton et al. 2011). Backgrounds
were monitored for 30 s prior to each analysis. Data
was then background-corrected by subtracting back-
ground average counts from the ablation counts. A
pre-ablation procedure where the first 5 s of the
ICP-MS signal was not included in the data was used
to ensure that possible surface contamination or the
periostracum was not included in the data (Marr et
al. 2011, Norrie et al. 2016). To minimise the possibil-
ity of laser burn through to lower layers in the shell
(Strasser et al. 2007), only the next 10 s of the laser
dwell time was included in analyses. Data was then
standardised using the most recent published NIST610
and NIST612 values (Jochum et al. 2005). All data
was standardised to a trace element:calcium ratio
(TE:Ca) in µmol of each TE to mol of Ca.
2.4.5. Statistical analyses
A discriminant function analysis (DFA) was per-
formed on the TE:Ca ratios of shell material de posited
in the ICCs. This allowed the examination of spatial
variation in trace elemental fingerprints and develop-
ment of site-specific reference signals. Due to differing
covariances, a quadratic DFA (QDFA) was performed.
237
a
b
2 mm
1 mm
Fig. 2. Example of individual Perna canaliculus used for
trace elemental fingerprinting. (a) Individual grown in an
in situ culturing container to develop a reference elemen-
tal fingerprint. Dashed red line: separation between shell
formed prior to translocation and that deposited in situ. (b)
Recently settled individual from an artificial settlement
substrate. Red dots: location of LA-ICP-MS analysis on
each individual
Author copy
Aquacult Environ Interact 12: 231–249, 2020
238
Elements were included stepwise into this QDFA to
determine the optimal suite of elements for classifica-
tion of shell material to its formation location (Dunphy
et al. 2015, Norrie et al. 2016). QDFA statistical analy-
ses were performed in JMP v.13 (SAS Institute). Due to
the loss of sampling equipment at some sites through
weather events over the study period, a QDFA was
performed in which all samples collected from each
site were binned together, regardless of the date col-
lected. This allowed the development of mean trace
elemental fingerprints for each site.
To predict natal origins, the discriminant function
trained with the TE:Ca ratios of shell material de po -
sited in ICCs was then used to assign
the shell material at the umbo of each
shell to a predicted formation location.
As the loss of sampling equipment re -
sulted in differing deployment dura-
tions for SSs, the number of settlers
predicted to have originated from each
site was standardised to a per day set-
tlement rate. Receiver operating char-
acteristic (ROC) curves were generated
for each site to estimate the diagnos-
tic ability of the QDFA. These ROC
curves compared the true positive rate
against the false positive rate. Poste-
rior probabilities of each individual
being correctly assigned to its forma-
tion location relative to other sites
were also calculated in order to pro-
vide estimates of the confidence of
the estimates.
3. RESULTS
3.1. Biophysical modelling
3.1.1. Lagrangian KDEs
The total KDE of all particles
released from all sites over the 11 yr
modelled period (Fig. 3) indicates the
area with the highest probability of
receiving settlers was located on the
eastern FoT to the north and south of
the WB1 and WB2 release sites. Par-
ticles are likely to settle on the east
and west coasts of the FoT (but not
the south) with a higher likelihood of
coastal settlement on the eastern FoT
than the western FoT. The KDE indi-
cated that benthic settlement is possible throughout
the FoT, although it is concentrated in the east. The
model also suggested a low probability of larvae
being transported from the FoT to the greater Hau-
raki Gulf, particularly the area north of Coromandel
Harbour and between Ponui Island and the main-
land. This total KDE is dominated by particles
released at areas with a high surface area from which
more particles were released, therefore individual
KDEs were generated for particles released from
each site (Fig. 4).
The site-specific KDEs (Fig. 4) indicated that parti-
cles released from sites which were more sheltered
Fig. 3. Kernel density estimates of all settled particles from all Perna canalicu-
lus model release sites over all release dates in the Firth of Thames. Colour
scale: probability that a particle will settle at a given location relative to the to-
tal number of parties released. Arrows: location of particle release locations
that are difficult to see due to their size
Author copy
Norrie et al.: Larval dispersal from bivalve aquaculture
settled over a smaller area than those released from
more exposed sites. This can be seen by particles
released from COR exhibiting the highest probability
of settlement in the relatively small area close to the
mouth of Coromandel Harbour (Fig. 4a). Particles
released from MAN also exhibited a high likelihood
of settlement over a relatively small area close to
their release site (Fig. 4b). A number of these parti-
cles, however, settled throughout the FoT, particu-
larly in the area north of Coromandel Harbour.
239
Longitude
de
N
N
Probability of settlement (x10–4)
Particle release
locations
abc
Lattitude
N
N
N
175 175.1 175.2 175.3 175.4 175.5 175 175.1 175.2 175.3 175.4 175.5
–36.4
–36.5
–36.6
–36.7
–36.8
–36.9
–37.0
–37.1
–37.2
–37.3
–36.4
–36.5
–36.6
–36.7
–36.8
–36.9
–37.0
–37.1
–37.2
–37.3
175 175.1 175.2 175.3 175.4 175.5
Fig. 4. Site-specific kernel density estimates of settled Perna canaliculus particles released from (a) COR, (b) MAN, (c) WP,
(d) WB1, and (e) WB2. All settled particles released over the December−June release window during the entire 11 yr period
are shown. Colour scale: portion of particles that settled at each location relative to the total number of particles released from
that site. Arrows: locations of particle release locations
Author copy
Aquacult Environ Interact 12: 231–249, 2020
The 3 more exposed sites (WP, WB1, and WB2) had
predicted settlement over wide areas throughout the
FoT. Particles from WP (Fig. 4c) exhibited high settle-
ment probabilities to the south and south east of the
release site, including on the western coast of the
FoT. A number of particles also settled on the eastern
side of the FoT, particularly in the area south of Man-
aia Harbour. Particles released from the WB1 and
WB2 sites (Fig. 4d,e) also settled throughout the FoT.
The settlement patterns of particles released from
these sites was broadly similar, with areas of high
settlement probability located close to their release
site as well as along the eastern coast of the FoT as
far south as the town of Thames. Differences in in the
settlement probabilities between the 2 sites were most
evident in the south and northwest of the FoT. Particles
released from WB1 were likely to settle in the south-
ern FoT close to the Waihou River (Fig. 4d), whilst
particles released from WB2 had a higher probability
of settling in the north western FoT (Fig. 4e).
3.1.2. Spatial differences in the proportion
of settled particles
Only 30% of the 4.9 million particles released be-
tween December and April across the years
1995−2005 encountered suitable settlement habitat
(the coast or seafloor) during the settlement com -
petency window. The remaining 70% of particles
which did not settle were retired from the model. Dif-
ferential settlement rates were observed between
sites (Fig. 5). Highest settlement rates were pre-
dicted for particles released from the relatively shel-
tered sites COR (77% of particles settled) and MAN
(53% settled). Conversely, the more exposed sites
had a much lower proportion of settled particles
which decreased as their distance from the coast in-
creased: 35% of particles released from WP settled,
33% of those released from WB1 settled, and only
23% of particles released from WB2 encountered suit-
able settlement habitat during the settlement compe-
tency period.
3.2. Trace elemental fingerprinting
3.2.1. Spatial variation in trace elemental
fingerprints
Overall, the QDFA on reference shells correctly
classified 72% of individuals to their growth location
using 7 TE:Ca ratios (Mn:Ca, Co:Ca, Ba:Ca, Ti:Ca,
Mg:Ca, Sr:Ca, Zn:Ca) (Table 2). This is highlighted
by the plot of the canonical scores from the QDFA
(Fig. 6) indicating differences between elemental fin-
gerprints of shell grown at different sites. However,
individual site-level classification success was vari-
able. High levels of classification success were
observed at C2 (89%), C3 (100%), C5 (90%), and W1
(92%), indicating unique trace elemental finger-
prints in shells deposited at these sites. Classification
rates of shell material deposited at C1 was moderate,
with 71% of individuals correctly classified. Approx-
imately half of individuals grown at C4 (53%), C6
(55%), and C7 (52%) were correctly classified. The
variability observed in the trace elemental finger-
prints of shell material at C4, C6, and C7 shows that
these sites are more difficult to differentiate based on
their trace elemental fingerprint. The ROC curves
were high for all sites (0.89−0.99), indicating the sen-
sitivity of the QDFA was high (Table 2). Despite the
relatively short distance between C5 and C6 (1.2 km),
there were large differences in the trace elemental
fingerprint of shell material deposited at these sites.
Overall, 90% of individuals from C5 were correctly
classified to their growth site. Classification success at
C6, however, was only 55 % with the majority of mis-
classified individuals attributed to C1, which was
located furthest from C6 at approximately 16 km away.
240
0
25
50
75
100
COR MAN WP WB1 WB2
Release site
Percent of released particles
Fate of
particle
Retired
Settled
Fig. 5. Percent of Perna canaliculus particles released from
each site which encounter suitable settlement habitat (de-
fined as the seafloor or coast) during the settlement compe-
tency window 3−5 wk after release. Retired: the particle did
not encounter suitable settlement substrate within the set-
tlement competency window and was therefore removed
from the model
Author copy
Norrie et al.: Larval dispersal from bivalve aquaculture
3.2.2. Natal locations of recent settlers
The predicted larval origins of recent settlers col-
lected at each of the sites monitored (Fig. 7, Table 3)
suggested that the larval pool in the FoT is well mixed.
Settlers collected at each site were a combination of
larvae produced throughout the study region. A min-
imum of 5 and a maximum of 7 potential natal loca-
tions were predicted for settlers at each site. No set-
tlers at any of the sites collected over the study period
of January−May 2018 were predicted to have origi-
nated at C3. The confidence of the model in assign-
ing individuals to their natal location was high, as
shown by the mean posterior probabilities of correct
assignment to each group ranging from 0.73−0.83
(Table 3) in agreement with the overall classification
success rates of reference shell material (Table 2).
Although overall the larval pool was well mixed,
some sites acted as larval sources and sinks relative to
others. The most important larval source from the per-
spective of the highest number of larvae produced
was C7. Of the individuals which settled at monitored
sites, 30% were predicted to have originated from this
location (1.63 settlers d−1) (Table 3). The high contri-
butions made by C7 were primarily driven by the high
number of settled larvae originating from C7 at W1,
which had the highest settlement rates of all sites
monitored with 2.66 mussels d−1 settling at this loca-
tion (Table 3). A large proportion of larvae which
settled at C1 (48%; 0.28 settlers d−1) and C2 (46%;
0.25 settlers d−1) were also predicted to have origi-
nated from this location. A large proportion of larvae
which settled throughout the study region were also
predicted by the QDFA to have originated at W1 in
the western FoT (0.95 settlers d−1) (Table 3). Self-re-
cruitment was generally low throughout the FoT with
C7 (20% self-recruitment), W1 (14%), and C2 (15%)
exhibiting the highest levels of the sites monitored.
No settlers throughout the FoT were predicted to
have originated at C3 despite the moderate settlement
at this location. Only small numbers of larvae were
predicted to have originated at both C2 and C5; al-
though settlement at C5 was low, a moderate number
of larvae settled at C2. These 3 sites can be considered
larval sinks as few larvae from throughout the FoT
were predicted to have originated at these locations.
4. DISCUSSION
This study set out to answer the following ques-
tions. (1) What is the potential for dispersal of Perna
241
Collection Predicted collection location No. collected No. correct % Correctly classified ROC curve
location C1 C2 C3 C4 C5 C6 C7 W1
C1 89 8 0 2 2 11 6 7 125 89 71.2 0.92
C2 0 25 0 0 0 1 1 1 28 25 89.2 0.97
C3 0 0 13 0 0 0 0 0 13 13 100 0.99
C4 7 5 0 29 2 3 4 4 54 29 53.7 0.93
C5 1 0 0 0 19 1 0 0 21 19 90.4 0.99
C6 3 7 4 2 3 28 1 3 51 28 54.9 0.89
C7 2 0 0 0 4 7 23 8 44 23 52.2 0.94
W1 1 1 0 0 1 2 0 66 71 66 92.9 0.97
No. predicted 103 46 17 33 31 53 35 89 407 292 71.74
Table 2. Classification success rates of Perna canaliculus shell material deposited in the in situ culturing containers between
January and May 2018. All individuals were grouped by site regardless of the dates on which they were collected. Bold values
indicate correct classification to growth site. ROC: receiver operating characteristic
−1
0
1
2
−1 0 1
Canonical score 1
Canonical score 2
SITE
C1
C2
C3
C4
C5
C6
C7
W1
C3
C4
C6
W1
C2
C5
C1
C7
Fig. 6. Mean (±SEM) canonical scores from the quadratic
discriminant function analysis examining trace element to
Ca ratios in the shell material of all Perna canaliculus indi-
viduals collected from monitored sites between January and
May 2018
Author copy
Aquacult Environ Interact 12: 231–249, 2020
242
canaliculus from aquaculture locations throughout
the FoT? (2) How do larval dispersal patterns gener-
ated by biophysical modelling and trace elemental
fingerprinting differ? (3) What are the overall impli-
cations of any larval spill-over for restoration and
aquaculture? The results of both the trace elemental
fingerprinting study and the biophysical modelling
indicated that there is potential for larval spill-over
from aquaculture to areas throughout the FoT. The
larval pool in the FoT is likely to be well mixed, with
larvae from all sites potentially contributing settlers to,
and receiving settlers from, other locations. Al though
the results of both approaches showed similar broad-
scale results, different dispersal pathways were pre-
dicted by each technique. These dif-
ferences high light the importance of
incorporating multiple tracking tech-
niques for a holistic un der standing of
larval dispersal patterns. Together, our
results provide the first evidence, to
our knowledge, that larvae produced
in aquaculture can assist in the resto-
ration of degraded wild bivalve popu-
lations through a larval subsidy).
4.1. What is the potential for
dispersal of P. canaliculus
from aquaculture locations
throughout the FoT?
Both biophysical modelling and trace
elemental fingerprinting indicated the
Waimang Point mussel farms (WP, W1)
were likely to be a source of mussel lar-
vae which settle throughout the study
area. In addition to contributing larvae
to several sites, high settlement rates at
this location also indicate that Wai -
mang Point receives settlers from sev-
eral locations. In a restoration context,
the fact that this location may both re-
ceive and contribute larvae highlights
the potential importance of this site in
regional population dynamics. Restora-
tion of benthic mussel beds at this site
may provide habitat for larvae pro-
duced in other areas of the FoT as well
as contribute larvae to other popula-
tions in the area.
Trace elemental fingerprinting pre-
dicted that the area to the north of
Coromandel Harbour (C7) is an impor-
tant larval source for settlers throughout the study
area despite no known populations existing at this
location. The biophysical model predicted settlement
at this location was likely to be high, particularly for
larvae released from Manaia and Coromandel Har-
bours (MAN and COR). One such explanation for this
result may be that early stage larvae dispersing past
this location incorporate elemental fingerprints from
this site, although this is unlikely as hydrodynamic
modelling indicated larvae in this area are retained
close to their natal location for several days. Alterna-
tively, these results present the possibility that a pop-
ulation exists at this location supported by aquacul-
ture which may then in turn act as a source of larvae
2
N
1
N
1
2
C1
C2
C3
C4
C7
W1
C6
C5
10 km
2.5 km
Coromandel
Harbour
Manaia Harbour
Wilson
Bay
Waimango
Point
N
Firth of
Thames
Fig. 7. Proportional natal locations of Perna canaliculus settling at monitored
locations throughout the Firth of Thames as predicted by the quadratic dis-
criminant function analysis trained using shell material deposited in situ at
these locations. Coloured square adjacent to the site names (W1−C7): colour
codes of settlers originating from these sites. Note the size of circle indicates
the number of settlers at each site standardised per day (Table 3)
Author copy
Norrie et al.: Larval dispersal from bivalve aquaculture
for other sites in the FoT. This demonstrates the
power of integrating the results of the 2 methods. The
possibility of this stepping stone dispersal (Kimura &
Weisss 1964, Palumbi 2003) highlights the need to
consider system-wide connectivity into management
actions based on source−sink dynamics (Kininmonth
et al. 2011). Future surveys of potential mussel popu-
lations in the FoT should investigate this location to
determine the presence of a population at this location
which may supply larvae throughout the study area.
There are several larval sinks in the FoT study
area. The area with the highest probability of settle-
ment as predicted by the biophysical model was
located to the south of Manaia Harbour, with larvae
from each release site capable of settling at this loca-
tion. Trace elemental fingerprinting also demonstrated
that larvae settling in this area (C1, C2) likely origi-
nated from multiple sources. This finding is impor-
tant as the area is known to support one of the few
extant populations of mussels in the FoT (A. Jeffs,
University of Auckland, pers. comm.), suggesting that
aquaculture may already be supporting this popula-
tion. The biophysical model also predicted that the
area to the south of the Wilson Bay mussel farms
would have high rates of settlement. Although this
location currently supports no known populations of
mussels, it is anecdotally known by local mussel
farmers to be an area of high spat settlement (Smith
2019). Moderate settlement was also observed on
artificial SSs in this area (C1, C2).
The similarity between the 2 methods of estimating
larval dispersal shows that decisions on management
in the study area based on this data can be made with
a high degree of confidence and highlights the
advantages of employing multiple techniques. The
restoration of settlement habitat at these sink loca-
tions may allow the recovery of benthic mussel pop-
ulations in this area due to high larval supply. Al -
though adults are able to survive in this area (McLeod
et al. 2012), it is important to consider factors which
may prevent the survival of larvae or juveniles both
during and after dispersal (Brandt et al. 2008, Pineda
et al. 2010, Shima et al. 2015).
4.2. How do the 2 methods of estimating larval
dispersal differ?
Although the overall pattern of a well-mixed larval
pool in the FoT was demonstrated by both methods of
tracking larvae, several differences in the pathways
of dispersal were observed. These differences high-
light the need to integrate multiple techniques of
estimating larval dispersal and population connectiv-
ity to obtain a clearer picture of dispersal patterns
(Ashford et al. 2010, Nolasco et al. 2018). The pri-
mary difference was the lack of settlers attributed to
the large mussel farm blocks off Wilson Bay (WB1,
WB2, C3) by trace elemental fingerprinting despite
the prediction of the biophysical model that settlers
243
Collection Predicted natal location Total No. of days settlement Settlement
location C1 C2 C3 C4 C5 C6 C7 W1 no. substrate deployed rate
settled (no. of deployments) (mussels d−1)
C1 7 3 0 8 0 3 28 9 58 98 (2) 0.59
C2 6 12 0 8 5 2 36 8 77 139 (3) 0.55
C3 8 0 0 10 0 16 9 13 56 84 (2) 0.67
C4 3 2 0 2 0 6 4 5 22 98 (2) 0.22
C5 1 1 0 1 0 3 14 4 24 84 (2) 0.29
C6 8 3 0 0 0 3 2 1 17 139 (3) 0.12
C7 4 0 0 1 0 12 9 19 45 139 (3) 0.32
W1 43 10 0 39 14 34 61 36 237 89 (2) 2.66
Total no. predicted 80 31 0 69 19 79 163 95 536 0.6775
natal origin
Settler contribution 0.82 0.28 0 0.74 0.2 0.82 1.63 0.95
rate (mussels d−1)
Mean posterior prob- 0.8 0.73 − 0.77 0.8 0.77 0.83 0.82
ability (±SEM) (±0.02) (± 0.03) (±0.02) (± 0.05) (±0.02) (± 0.01) (±0.01)
Table 3. Predicted formation locations of Perna canaliculus shell material at the prodissoconch of individuals which settled at
monitored sites in the Firth of Thames between January and May 2018. As weather events resulted in non-standard deploy-
ment times for settlement substrates, standardised settlement per day was calculated. Settler contribution rate was calculated
as the number of settlers per day of sampling which were predicted to have originated at each site
Author copy
Aquacult Environ Interact 12: 231–249, 2020
from this location would dominate settlement patterns
throughout the study area. The biophysical model
also predicted that larvae produced within Coroman-
del Harbour (COR, C5, C6) would be retained in this
area, although this was not shown to be the case in
by trace elemental fingerprinting techniques. Several
possible explanations exist for the differences be -
tween the 2 methods of tracking the dispersal of lar-
vae in the FoT.
It is likely that the variation in temporal scales over
which the 2 studies were conducted was responsible
for the differences observed. While the biophysical
model considered circulation patterns over a 10 yr
period, trace elemental fingerprinting only examined
settlement patterns over a single spawning season. It
is unlikely that dispersal patterns within a single year
will fit with the 10 yr mean, as inter-annual variation
in potential dis persal distances exists (Norrie 2019).
This suggests that an extended sampling period
should be undertaken to determine if long-term
trends ob served through different techniques agree.
In addition to possible variations in circulation, envi-
ronmental conditions in areas of predicted high set-
tlement may not have been suitable for the survival
and settlement of larvae. The study area experienced
heat wave conditions over the period in which field
sampling was conducted (Jan−May 2018) (Herring et
al. 2019), with reports of high mortality of mussels in
aquaculture operations. This may have reduced both
larval production at some locations as well as re -
duced the survival of larvae.
Additionally, there was variation in settlement
rates predicted by the biophysical model between
sites. These differences were driven by the rate at
which larvae encountered potential settlement habi-
tat within the time period in which settlement is pos-
sible. These results show that sites which produce
the highest numbers of larvae may not necessarily
contribute the highest numbers of settlers to popula-
tions. These differences in the rates at which larvae
encounter settlement habitat have the potential to
interact with direct stresses placed on an individual
as it disperses, such as predation, as well as indirect
dispersal stresses which may reduce post-settlement
survival (Nanninga & Berumen 2014). Differences
between the results of biophysical modelling and
trace elemental fingerprinting highlight the need to
consider realised dispersal and include empirical
data into predictions of larval dispersal (Pineda et al.
2007). The similarities in overall patterns between
the 2 methods employed in the current study high-
light the utility of using computer-based simulations
to guide the application of empirical validation
experiments. These empirical experiments may then
provide feedback towards validation and parameter-
isation of future models.
4.3. What are the implications overall for restoration
and aquaculture?
The overall finding of larval spill-over from aqua-
culture has important consequences for the restora-
tion and recovery of depleted bivalve populations. To
our knowledge, this is the first study which demon-
strates that bivalve aquaculture in coastal waters has
the potential to provide a larval subsidy to relict and
restored bivalve reefs and therefore assist with their
recovery. This research shows that if bivalve aqua-
culture is conducted in an area with degraded bi -
valve populations, spill-over from these locations
should be explicitly considered in restoration pro-
grammes. This research also shows that a network
approach should be taken to restoration which in -
cludes connectivity between sites.
If larval spill-over from aquaculture to natural pop-
ulations results in the recovery of degraded popula-
tions or assists with restoration efforts, connectivity to
these locations should be maximised. This highlights
the potential benefits of a network approach to resto-
ration. The need for network approaches in marine
reserve design is well known (e.g. Almany et al.
2007, Gaines et al. 2010, Green et al. 2015, Puckett &
Eggleston 2016, Coleman et al. 2017), and these
approaches should be translated to the design of
restoration programmes. Where larval spill-over is
desirable, additional bivalve aquaculture locations
should be carefully planned in the context of the sys-
tem in which this culturing takes place to ensure that
larval spill-over to natural populations is enhanced.
New restoration efforts should also be considered in
the context of new or existing bivalve aquaculture to
determine locations at which effort should be applied
in order to maximise the chances of success (Arnold
et al. 2017). The inclusion of aquaculture populations
may simplify these network approaches, as many
aquaculture populations are restocked after harvest
resulting in relatively stable populations without the
need to receive settlers.
It is also important to consider the possibility that
larval export from aquaculture may negatively affect
local populations of bivalves. The escape of finfish
from farming operations has been shown to have
potentially deleterious consequences on wild popula-
tions through the reduction of genetic diversity
(Jørstad et al. 2008, Morris et al. 2008, Jensen et al.
244
Author copy
Norrie et al.: Larval dispersal from bivalve aquaculture
2010). In bivalves, genetic introgression to wild stocks
has also been demonstrated which may result in
reduced fitness in the wild population (Apte et al.
2003). In cases where natural populations of a bi -
valve species exist, strategies which minimise poten-
tial larval export from aquaculture should be imple-
mented. Potential strategies may include careful
placement of aquaculture where settlement of larvae
is unlikely, or the implementation of hatchery tech-
niques such as spawning triploid bivalves with lower
reproductive fitness (e.g. Honkoop 2003, Jouaux et
al. 2010). In the case of the FoT, however, no large
natural populations exist, making negative conse-
quences less likely at present.
4.4. Future directions
This study has provided important insight into lar-
val dispersal patterns from aquaculture and their
interactions with restoration. However, to under-
stand the ecological consequences of this dispersal it
is essential that the survival to reproduction is quan-
tified (Pineda et al. 2007, Nanninga & Berumen
2014). Several stressors may prevent the establish-
ment of mussel reefs despite high larval supply
(Burgess et al. 2012). In addition to spatio-temporal
variation in mortality (White et al. 2014, Pineda &
Reyns 2018), physiological stresses placed on an indi-
vidual during dispersal, or post-settlement stresses
may result in decreased fitness or lower reproductive
output (Baker & Rao 2004, Burgess & Marshall 2011,
Shima et al. 2015). Quantifying these stressors is
essential for the selection and further assessment of
viability of restoration sites, as larval supply and set-
tlement will not result in a sustainable population if
settlers do not survive. It is also important to quantify
larval production in order to more accurately para-
meterise biophysical models and reconcile differ-
ences between modelled and empirically determined
dispersal. For example, to better parameterise mod-
els, sites likely to receive high settlement as deter-
mined by modelling and empirical methods should
now be monitored over long time scales in order to
better understand inter- and intra-annual variation in
larval production over the spatial extent of this study.
Additionally, some locations are likely to result in
lower larval output due to differences in the size
structure of the population and age-dependent fecun-
dity (Ren & Ross 2005). This is particularly important
in an aquaculture setting where harvests occur at a
relatively small size, which may reduce larval pro-
duction relative to wild populations.
Understanding the consequences of varying dis-
persal distances is also important in the context of cli-
mate change. Changing ocean conditions such as
salinity and temperature may result in changes in
hydrodynamic flows on a number of scales (Wu et al.
2012, Sen Gupta et al. 2015). As these changes in
hydrodynamic flows will directly affect the potential
dispersal of larvae, it is important that changes in
realised connectivity are considered (Coleman et al.
2011, 2017). Modelling exercises, for example, have
shown that while oceanic circulation changes may
affect larval transport, faster development and there-
fore earlier settlement (related to temperature) could
override the effects of these changed transport path-
ways (Cetina-Heredia et al. 2015) or vice versa. It is
also important to consider that faster development of
larvae in warmer waters will only occur if the supply
of nutrients is high enough to support this develop-
ment; however, the distribution of plankton is also
likely to change, which may decrease larval growth
rates and even possibly increase mortality rates
(Munday et al. 2009, Andrello et al. 2015). Finally, the
possibility of employing methods such as those used
by Nolasco et al. (2018), which explicitly incorporate
the results of biophysical models and trace elemental
fingerprinting into a single statistical framework,
could prove very powerful.
Acknowledgements. C.N. was supported by a UoA doc-
toral scholarship. Funding for fieldwork and laser analyses
were provided by NIWA Coasts and Oceans core funding
(COME1903). The hydrodynamic model was developed
by MetOcean Solutions. M.R. was partially supported by
the Moana Project (www.moanaproject.org), funded by
the New Zealand Ministry of Business Innovation and
Employment, contract number METO1801. Stuart Morrow
from the UoA mass spectrometry centre aided with ele-
mental analyses. Alex Vincent aided with figures. Jessica
Bailey assisted in the laboratory. Fieldwork assistance
was provided by Scott Edhouse, David Bremner, Peter
Schlegel, Esther Stuck, Peter Browne, Errol Murray, and
Jenny Hillman. Alan Bartrom of Gulf Mussels provided
logistical support. Thanks to the Miller lab at OSU for
their input on a draft of the manuscript. We also thank 2
anonymous reviewers and the editor for their constructive
feedback which greatly improved previous versions of the
manuscript.
LITERATURE CITED
Alfaro AC, Jeffs AG (2002) Small-scale mussel settlement
patterns within morphologically distinct substrata at
Ninety Mile Beach, northern New Zealand. Malacologia
44: 1−15
Alfaro AC, Jeffs AG, Hooker SH (2001) Reproductive behav-
ior of the green-lipped mussel, Perna canaliculus, in
northern New Zealand. Bull Mar Sci 69: 1095−1108
245
Author copy
Aquacult Environ Interact 12: 231–249, 2020
Alfaro AC, Jeffs AG, Creese RG (2004) Bottom-drifting algal/
mussel spat associations along a sandy coastal region in
northern New Zealand. Aquaculture 241: 269−290
Alleway HK, Connell SD (2015) Loss of an ecological baseline
through the eradication of oyster reefs from coastal eco-
systems and human memory. Conserv Biol 29: 795−804
Almany GR, Berumen ML, Thorrold SR, Planes S, Jones GP
(2007) Local replenishment of coral reef fish populations
in a marine reserve. Science 316: 742−744
Andrello M, Mouillot D, Somot S, Thuiller W, Manel S (2015)
Additive effects of climate change on connectivity
between marine protected areas and larval supply to
fished areas. Divers Distrib 21: 139−150
Apte S, Star B, Gardner JPA (2003) A comparison of genetic
diversity between cultured and wild populations, and a
test for genetic introgression in the New Zealand green-
shell mussel Perna canaliculus (Gmelin 1791). Aquacul-
ture 219: 193−220
Arnold WS, Meyers SD, Geiger SP, Luther ME, Narváez D,
Frischer ME, Hofmann E (2017) Applying a coupled bio-
physical model to predict larval dispersal and source/sink
relationships in a depleted metapopulation of the eastern
oyster Crassostrea virginica. J Shellfish Res 36: 101−118
Ashford J, La Mesa M, Fach BA, Jones C, Everson I (2010)
Testing early life connectivity using otolith chemistry
and particle-tracking simulations. Can J Fish Aquat Sci
67: 1303−1315
Baker MB, Rao S (2004) Incremental costs and benefits
shape natal dispersal: theory and example with Hemi -
lepistus reaumuri. Ecology 85: 1039−1051
Beck MW, Brumbaugh RD, Airoldi L, Carranza A and others
(2011) Oyster reefs at risk and recommendations for con-
servation, restoration, and management. Bioscience 61:
107−116
Becker BJ, Levin LA, Fodrie FJ, McMillan PA (2007) Complex
larval connectivity patterns among marine invertebrate
populations. Proc Natl Acad Sci USA 104: 3267−3272
Black KP, Bell RG, Oldman JW, Carter GS, Hume TM (2000)
Features of 3-dimensional barotropic and baroclinic cir-
culation in the Hauraki Gulf, New Zealand. N Z J Mar
Freshw Res 34: 1−28
Botev ZI, Grotowski JF, Kroese DP (2010) Kernel density
estimation via diffusion. Ann Stat 38: 2916−2957
Brandt G, Wehrmann A, Wirtz KW (2008) Rapid invasion of
Crassostrea gigas into the German Wadden Sea domi-
nated by larval supply. J Sea Res 59: 279−296
Burgess SC, Marshall DJ (2011) Are numbers enough? Col-
onizer phenotype and abundance interact to affect popu-
lation dynamics. J Anim Ecol 80: 681−687
Burgess SC, Treml EA, Marshall DJ (2012) How do dispersal
cost and habitat selection influence realized population
connectivity? Ecology 93: 1378−1387
Carson HS, Cook GS, López-Duarte PC, Levin LA (2011)
Evaluating the importance of demographic connectivity
in a marine metapopulation. Ecology 92: 1972−1984
Cathey AM, Miller NR, Kimmel DG (2012) Microchemistry
of juvenile Mercenaria mercenaria shell: implications for
modeling larval dispersal. Mar Ecol Prog Ser 465:
155−168
Cetina-Heredia P, Roughan M, Van Sebille E, Feng M, Cole-
man MA (2015) Strengthened currents override the effect
of warming on lobster larval dispersal and survival. Glob
Change Biol 21: 4377−4386
Cetina-Heredia P, Roughan M, Liggins G, Coleman MA,
Jeffs A (2019) Mesoscale circulation determines broad
spatio-temporal settlement patterns of lobster. PLOS
ONE 14: e0211722
Chia FS, Buckland-Nicks J, Young CM (1984) Locomotion of
marine invertebrate larvae: a review. Can J Zool 62:
1205−1222
Coen LD, Brumbaugh RD, Bushek D, Grizzle R and others
(2007) Ecosystem services related to oyster restoration.
Mar Ecol Prog Ser 341: 303−307
Coleman MA, Roughan M, Macdonald HS, Connell SD,
Gillanders BM, Kelaher BP, Steinberg PD (2011) Varia-
tion in the strength of continental boundary currents
determines continent-wide connectivity in kelp. J Ecol
99: 1026−1032
Coleman MA, Cetina-Heredia P, Roughan M, Feng M, Van
Sebille E, Kelaher BP (2017) Anticipating changes to
future connectivity within a network of marine protected
areas. Glob Change Biol 23: 3533−3542
Cook GS, Parnell PE, Levin LA (2014) Population connectiv-
ity shifts at high frequency within an open-coast marine
protected area network. PLOS ONE 9: e103654
Cowen RK, Lwiza KMM, Sponaugle S, Paris CB, Olson DB
(2000) Connectivity of marine populations: Open or
closed? Science 287: 857−859
Cranford PJ, Hargrave BT, Doucette LI (2009) Benthic
organic enrichment from suspended mussel (Mytilus
edulis) culture in Prince Edward Island, Canada. Aqua-
culture 292:
189−196
Dagestad KF, Röhrs J, Breivik Ø, Ådlandsvik B (2018) Open-
Drift v1.0: a generic framework for trajectory modeling.
Geosci Model Dev 11: 1405−1420
Dumbauld BR, Ruesink JL, Rumrill SS (2009) The ecological
role of bivalve shellfish aquaculture in the estuarine
environment: a review with application to oyster and
clam culture in West Coast (USA) estuaries. Aquaculture
290: 196−223
Dunphy BJ, Silva C, Gardner JPA (2015) Testing techniques
for tracing the provenance of green-lipped mussel spat
washed up on Ninety Mile Beach. New Zealand Aquatic
Environment and Biodiversity Report No. 164. Ministry
for Primary Industries, Wellington
Ehrich MK, Harris LA (2015) A review of existing eastern
oyster filtration rate models. Ecol Modell 297: 201−212
Elsäßer B, Fariñas-Franco JM, Wilson CD, Kregting L,
Roberts D (2013) Identifying optimal sites for natural
recovery and restoration of impacted biogenic habitats in
a special area of conservation using hydrodynamic and
habitat suitability modelling. J Sea Res 77: 11−21
FAO (2018) The state of world fisheries and aquaculture
2018—Meeting the sustainable development goals.
FAO, Rome
Fariñas-Franco JM, Roberts D (2018) The relevance of
reproduction and recruitment to the conservation and
restoration of keystone marine invertebrates: a case
study of sublittoral Modiolus modiolus reefs impacted by
demersal fishing. Aquat Conserv 28: 672−689
French McCay DP, Peterson CH, DeAlteris JT, Catena J
(2003) Restoration that targets function as opposed to
structure: replacing lost bivalve production and filtra-
tion. Mar Ecol Prog Ser 264: 197−212
Gaines SD, White C, Carr MH, Palumbi SR (2010) Designing
marine reserve networks for both conservation and
fisheries management. Proc Natl Acad Sci USA 107:
18286−18293
Gausen D, Moen V (1991) Large-scale escapes of farmed
Atlantic salmon (Salmo salar) into Norwegian rivers
246
Author copy
Norrie et al.: Larval dispersal from bivalve aquaculture
threaten natural populations. Can J Fish Aquat Sci 48:
426−428
Gawarkiewicz G, Monismith S, Largier J (2007) Observing
larval transport processes affecting population connec-
tivity. Oceanography (Wash DC) 20: 40−53
Gibbs MM, James MR, Pickmere SE, Woods PH, Shake-
speare BS, Hickman RW, Illingworth J (1991) Hydrody-
namic and water column properties at six stations associ-
ated with mussel farming in Pelorus sound, 1984−85. N Z
J Mar Freshw Res 25: 239−254
Giles H, Pilditch CA (2006) Effects of mussel (Perna
canaliculus) biodeposit decomposition on benthic respi-
ration and nutrient fluxes. Mar Biol 150: 261−271
Gomes I, Peteiro LG, Albuquerque R, Nolasco R, Dubert J,
Swearer SE, Queiroga H (2016) Wandering mussels:
using natural tags to identify connectivity patterns among
Marine Protected Areas. Mar Ecol Prog Ser 552: 159−176
Green AL, Maypa AP, Almany GR, Rhodes KL and others
(2015) Larval dispersal and movement patterns of coral
reef fishes, and implications for marine reserve network
design. Biol Rev Camb Philos Soc 90: 1215−1247
Hauraki Gulf Forum (2017) State of our Gulf 2014. Hauraki
Gulf - Tıkapa Moana/ Te Moananui a Toi. State of the
environment report 2014. Hauraki Gulf Forum, Auckland
Hayden B (1995) Factors affecting the recruitment of farmed
greenshell mussels Perna canaliculus, in Marlborough
Sounds. MSc thesis, University of Otago, Dunedin
Heino M, Svåsand T, Wennevik V, Glover KA (2015) Genetic
introgression of farmed salmon in native populations:
quantifying the relative influence of population size and
frequency of escapees. Aquacult Environ Interact 6:
185−190
Herring SC, Christidis N, Hoell A, Hoerling MP, Stott PA
(2019) Explaining extreme events of 2017 from a climate
perspective. Bull Am Meteorol Soc 100: S1−S117
Honkoop PJC (2003) Physiological costs of reproduction in
the Sydney rock oyster Saccostrea glomerata. Oecologia
135: 176−183
Inglis GJ, Hurren H, Oldman J, Haskew R (2006) Using habitat
suitability index and particle dispersion models for early
detection of marine invaders. Ecol Appl 16: 1377−1390
Jeffs AG, Holland RC, Hooker S, Hayden B (1999) Overview
and bibliography of research on the greenshell mussel,
Perna canaliculus, from New Zealand waters. J Shellfish
Res 18: 347−360
Jensen Ø, Dempster T, Thorstad EB, Uglem I, Fredheim A
(2010) Escapes of fishes from Norwegian sea-cage aqua-
culture: causes, consequences and prevention. Aquacult
Environ Interact 1: 71−83
Jochum KP, Nohl U, Herwig K, Lammel E, Stoll B, Hofmann
AW (2005) GeoReM: a new geochemical database for
reference materials and isotopic standards. Geostand
Geoanal Res 29: 333−338
Jørstad KE, Van Der Meeren T, Paulsen OI, Thomsen T,
Thorsen A, Svåsand T (2008) ‘Escapes’ of eggs from
farmed cod spawning in net pens: recruitment to wild
stocks. Rev Fish Sci 16: 285−295
Jouaux A, Heude-Berthelin C, Sourdaine P, Mathieu M,
Kellner K (2010) Gametogenic stages in triploid oysters
Crassostrea gigas: irregular locking of gonial prolifera-
tion and subsequent reproductive effort. J Exp Mar Biol
Ecol 395: 162−170
Kimura M, Weisss GH (1964) The stepping stone model of
population structure and the decrease of genetic correla-
tion with distance. Genetics 49: 561−576
Kininmonth S, Beger M, Bode M, Peterson E and others
(2011) Dispersal connectivity and reserve selection for
marine conservation. Ecol Modell 222: 1272−1282
Kool J, Moilanen TA, Treml EA (2013) Population connectiv-
ity: recent advances and new perspectives. Landsc Ecol
28: 165−185
Kroll IR, Poray AK, Puckett BJ, Eggleston DB, Fodrie FJ
(2016) Environmental effects on elemental signatures in
eastern oyster Crassostrea virginica shells: using geo-
chemical tagging to assess population connectivity. Mar
Ecol Prog Ser 543: 173−186
Kroll IR, Poray AK, Puckett BJ, Eggleston DB, Fodrie FJ
(2018) Quantifying estuarine-scale invertebrate larval
connectivity: methodological and ecological insights.
Limnol Oceanogr 63: 1979−1991
Lane DJW, Beaumont AR, Hunter JR (1985) Byssus drifting
and the drifting threads of the young post-larval mussel
Mytilus edulis. Mar Biol 84: 301−308
Lazareth C, Putten EV, André L, Dehairs F (2003) High-res-
olution trace element profiles in shells of the mangrove
bivalve Isognomon ephippium: A record of environmen-
tal spatio-temporal variations? Estuar Coast Shelf Sci 57:
1103−1114
Le Corre N, Martel AL, Guichard F, Johnson LE (2013) Vari-
ation in recruitment: differentiating the roles of primary
and secondary settlement of blue mussels Mytilus spp.
Mar Ecol Prog Ser 481: 133−146
Le Port A, Montgomery JC, Croucher AE (2014) Biophysical
modelling of snapper Pagrus auratus larval dispersal
from a temperate MPA. Mar Ecol Prog Ser 515: 203−215
Lipcius RN, Eggleston DB, Schreiber SJ, Seitz RD and others
(2008) Importance of metapopulation connectivity to
restocking and restoration of marine species. Rev Fish
Sci 16: 101−110
Lundquist C, Broekhuizen N (2012) Predicting suitable
shellfish restoration sites in Whangarei Harbour: larval
dispersal modelling and verification. Prepared for Min-
istry of Science and Innovation Envirolink Fund to
Northland Regional Council. National Institute of Water
& Atmospheric Research, Hamilton
Lundquist CJ, Thrush SF, Oldman JW, Senior AK (2004)
Limited transport and recolonization potential in shallow
tidal estuaries. Limnol Oceanogr 49: 386−395
Lundquist CJ, Oldman JW, Lewis MJ (2009) Predicting suit-
ability of cockle Austrovenus stutchburyi restoration
sites using hydrodynamic models of larval dispersal. N Z
J Mar Freshw Res 43: 735−748
Lundquist CJ, Bulmer R, Clark MR, Hillman JR and others
(2017) Challenges for the conservation of marine small
natural features. Biol Conserv 211: 69−79
Marr JP, Baker JA, Carter L, Allan ASR, Dunbar GB, Bostock
HC (2011) Ecological and temperature controls on
Mg/Ca ratios of Globigerina bulloides from the south-
west Pacific Ocean. Paleoceanography 26: PA2209
McLeod IM (2009) Green-lipped mussels, Perna canaliculus,
in soft-sediment systems in northeastern New Zealand.
MSc Thesis, University of Auckland
McLeod IM, Parsons DM, Morrison MA, Le Port A, Taylor
RB (2012) Factors affecting the recovery of soft-sediment
mussel reefs in the Firth of Thames, New Zealand. Mar
Freshw Res 63: 78−83
McLeod I, Parsons D, Morrison M, Van Dijken S, Taylor R
(2014) Mussel reefs on soft sediments: a severely reduced
but important habitat for macroinvertebrates and fishes in
New Zealand. N Z J Mar Freshw Res 48: 48−59
247
Author copy
Aquacult Environ Interact 12: 231–249, 2020
Morris MRJ, Fraser DJ, Heggelin AJ, Whoriskey FG, Carr
JW, O’Neil SF, Hutchings JA (2008) Prevalence and
recurrence of escaped farmed Atlantic salmon (Salmo
salar) in eastern North American rivers. Can J Fish Aquat
Sci 65: 2807−2826
Morrison M, Drury J, Shanker U, Hill A (2002) A broad scale
seafloor habitat assessment of the Firth of Thames using
acoustic mapping, with associated video and grab sam-
ple ground-truthing. NIWA Client Report AKL-2002-014.
National Institute of Water & Atmospheric Research,
Auckland
Morton JE, Miller M (1973) The New Zealand sea shore.
Collins & Sons, London
Munday PL, Leis JM, Lough JM, Paris CB, Kingsford MJ,
Berumen ML, Lambrechts J (2009) Climate change and
coral reef connectivity. Coral Reefs 28: 379−395
Nanninga GB, Berumen ML (2014) The role of individual
variation in marine larval dispersal. Front Mar Sci 1: 71
Newell RE (2004) Ecosystem influences of natural and culti-
vated populations of suspension-feeding bivalve mol-
luscs: a review. J Shellfish Res 23: 51−61
Nielsen P, Cranford PJ, Maar M, Petersen JK (2016) Magni-
tude, spatial scale and optimization of ecosystem serv-
ices from a nutrient extraction mussel farm in the
eutrophic Skive Fjord, Denmark. Aquacult Environ Inter-
act 8: 311−329
Nolasco R, Gomes I, Peteiro L, Albuquerque R, Luna T,
Dubert J, Swearer SE, Queiroga H (2018) Independent
estimates of marine population connectivity are more
concordant when accounting for uncertainties in larval
origins. Sci Rep 8: 2641
Norrie CR (2019) Quantifying population connectivity of
marine larvae: hydrodynamic modelling and shell micro-
chemistry methods to determine larval dispersal of Perna
canaliculus. PhD thesis, The University of Auckland
Norrie CR, Dunphy BJ, Baker JA, Lundquist CJ (2016)
Local-scale variation in trace elemental fingerprints of
the estuarine bivalve Austrovenus stutchburyi within
and between estuaries. Mar Ecol Prog Ser 559: 89−102
Norrie CR, Dunphy BJ, Ragg NLC, Lundquist CJ (2019)
Comparative influence of genetics, ontogeny and the
environment on elemental fingerprints in the shell of
Perna canaliculus. Sci Rep 9: 8533
North EW, King DM, Xu J, Hood RR and others (2010) Link-
ing optimization and ecological models in a decision sup-
port tool for oyster restoration and management. Ecol
Appl 20: 851−866
O’Callaghan JM, Stevens CL (2017) Evaluating the surface
response of discharge events in a New Zealand Gulf-
ROFI. Front Mar Sci 4: 232
Okubo A, Ebbesmeyer CC (1976) Determination of vorticity,
divergence, and deformation rates from analysis of
drogue observations. Deep-Sea Res 23: 349−352
Oldman J, Hong J, Stephens S, Broekhuizen N (2007) Verifica-
tion of Firth of Thames hydrodynamic model. Prepared for
Auckland Regional Council. NIWA Client Report, National
Institute of Water & Atmospheric Research, Auckland
Palumbi SR (2003) Population genetics, demographic con-
nectivity, and the design of marine reserves. Ecol Appl
13: 146−158
Paton C, Hellstrom J, Paul B, Woodhead J, Hergt J (2011)
Iolite: freeware for the visualisation and processing of mass
spectrometric data. J Anal At Spectrom 26: 2508−2518
Paul LJ (2012) A history of the Firth of Thames dredge fish-
ery for mussels: use and abuse of a coastal resource. New
Zealand Aquatic Environment and Biodiversity Report
No. 94. Ministry for Primary Industries, Wellington
Petersen JK, Hasler B, Timmermann K, Nielsen P, Tørring
DB, Larsen MM, Holmer M (2014) Mussels as a tool for
mitigation of nutrients in the marine environment. Mar
Pollut Bull 82: 137−143
Pilditch CA, Valanko S, Norkko J, Norkko A (2015) Post-set-
tlement dispersal: the neglected link in maintenance of
soft-sediment biodiversity. Biol Lett 11: 20140795
Pineda J, Reyns N (2018) Larval transport in the coastal
zone: biological and physical processes In: Carrier TJ,
Reitzel AM, Heyland A (eds) Evolutionary ecology of
marine invertebrate larvae. Oxford University Press,
Oxford, p 145−163
Pineda J, Hare J, Sponaugle S (2007) Larval transport and
dispersal in the coastal ocean and consequences for pop-
ulation connectivity. Oceanography (Wash DC) 20: 22−39
Pineda J, Porri F, Starczak V, Blythe J (2010) Causes of
decoupling between larval supply and settlement and
consequences for understanding recruitment and popu-
lation connectivity. J Exp Mar Biol Ecol 392: 9−21
Plew DR (2011) Shellfish farm-induced changes to tidal cir-
culation in an embayment, and implications for seston
depletion. Aquacult Environ Interact 1: 201−214
Puckett BJ, Eggleston DB (2016) Metapopulation dynamics
guide marine reserve design: importance of connectivity,
demographics, and stock enhancement. Ecosphere 7:
e01322
Puckett BJ, Eggleston DB, Kerr PC, Luettich R (2014) Larval
dispersal and population connectivity among a network
of marine reserves. Fish Oceanogr 23: 342−361
Puckett BJ, Theuerkauf SJ, Eggleston DB Guajardo R and
others (2018) Integrating larval dispersal, permitting,
and logistical factors within a validated habitat suitability
index for oyster restoration. Front Mar Sci 5: 76
Ren J S, Ross AH (2005) Environmental influence on mussel
growth:
a dynamic energy budget model and its appli-
cation to the greenshell mussel Perna canaliculus. Ecol
Modell 189: 347−362
Ricardo F, Génio L, Leal MC, Albuquerque R, Queiroga H,
Rosa R, Calado R (2015) Trace element fingerprinting of
cockle (Cerastoderma edule) shells can reveal harvest-
ing location in adjacent areas. Sci Rep 5: 11932
Sen Gupta A, Brown JN, Jourdain NC, van Sebille E,
Ganachaud A, Vergés A (2015) Episodic and non-uni-
form shifts of thermal habitats in a warming ocean. Deep
Sea Res II 113: 59−72
Shima JS, Noonburg EG, Swearer SE (2015) Consequences
of variable larval dispersal pathways and resulting phe-
notypic mixtures to the dynamics of marine metapopula-
tions. Biol Lett 11: 20140778
Smith RJ (2019) Settlement, retention, growth, and condition
in Greenshell™ mussels (Perna canaliculus) in the Hau-
raki Gulf. PhD thesis, The University of Auckland
South PM (2016) An experimental assessment of measures
of mussel settlement: effects of temporal, procedural and
spatial variations. J Exp Mar Biol Ecol 482: 64−74
Strasser CAA, Thorrold SRR, Starczak VRR, Mullineaux LS
(2007) Laser ablation ICP-MS analysis of larval shell in
softshell clams (Mya arenaria) poses challenges for natu-
ral tag studies. Limnol Oceanogr Methods 5: 241−249
Strasser CA, Mullineaux LS, Walther BD (2008) Growth rate
and age effects on Mya arenaria shell chemistry: implica-
tions for biogeochemical studies. J Exp Mar Biol Ecol
355: 153−163
248
Author copy
Norrie et al.: Larval dispersal from bivalve aquaculture
Tettelbach S, Peterson B, Carroll J, Hughes SWT and others
(2013) Priming the larval pump: resurgence of bay scal-
lop recruitment following initiation of intensive restora-
tion efforts. Mar Ecol Prog Ser 478: 153−172
Thomas Y, Dumas F, Andréfouet S (2016) Larval connectiv-
ity of pearl oyster through biophysical modelling; evi-
dence of food limitation and broodstock effect. Estuar
Coast Shelf Sci 182: 283−293
Thorrold SR, Jones GP, Hellberg ME, Burton RS and others
(2002) Quantifying larval retention and connectivity in
marine populations with artificial and natural markers.
Bull Mar Sci 70(Suppl): 291−308
Turner JS, Kellogg ML, Massey GM, Friedrichs CT (2019)
Minimal effects of oyster aquaculture on local water
quality: examples from southern Chesapeake Bay. PLOS
ONE 14: e0224768
van der Schatte Olivier A, Jones L, Le Vay L, Christie M,
Wilson J, Malham SK (2020) A global review of the eco-
system services provided by bivalve aquaculture. Rev
Aquacult 12: 3−25
Werner FE, Cowen RK, Paris CB (2007) Coupled biological
and physical models. Oceanography (Wash DC) 20:
54−69
White JW, Morgan SG, Fisher JL (2014) Planktonic larval
mortality rates are lower than widely expected. Ecology
95: 3344−3353
Wilcox M, Kelly S, Jeffs A (2018) Ecological restoration of
mussel beds onto soft-sediment using transplanted
adults. Restor Ecol 26: 581−590
Wu L, Cai W, Zhang L, Nakamura H and others (2012) En -
hanced warming over the global subtropical western
boundary currents. Nat Clim Chang 2: 161−166
Zeldis J, Oldman RJ, Ballara SL, Richards LA (2005) Physical
fluxes, pelagic ecosystem structure, and larval fish sur-
vival in Hauraki Gulf, New Zealand. Can J Fish Aquat
Sci 62: 593−610
zu Ermgassen PSE, Spalding MD, Grizzle RE, Brumbaugh
RD (2013) Quantifying the loss of a marine ecosystem
service: filtration by the eastern oyster in US Estuaries.
Estuaries Coasts 36: 36−43
249
Editorial responsibility: Megan La Peyre,
Baton Rouge, Louisiana, USA
Submitted: February 27, 2020; Accepted: May 7, 2020
Proofs received from author(s): June 10, 2020
Author copy