ThesisPDF Available

Quantifying Connectivity in Buzzards Bay With Application to the Study of Recruitment Variability in the Bay Scallop (Argopecten irradians)

Authors:

Abstract

The bay scallop has been the foundation of a historically important fishery in Buzzards Bay, MA. However, since the 1980s, the resource has been in decline and the population has been subject to large inter-annual fluctuations. Due to the relatively short life span of bay scallops (2-3 yr), the sensitivity of recruitment to the variability in larval transport to juvenile habitat is high. Long-term Bay-wide restoration strategies will need to take into account the influence of hydrodynamics on connectivity from spawning areas to juvenile- suitable habitat. To examine this connectivity in Buzzards Bay, a coupled biophysical modeling approach is utilized. A high-resolution ocean model (FVCOM) was used to drive an individual-based model of scallop larval transport from spawning to settlement. Connectivity among geographic sub-populations is quantified using Lagrangian probability density functions. The spawning-to-settlement-area connectivity of scallop larvae is principally due to the wind-driven transport. During late spring to summer, the principal time of bay scallop spawning, the southwest sea breeze dominates the wind forcing of Bay currents. The up-Bay flow driven by the surface wind stress in the shallow regions of the Bay results in substantial transport of larvae from mid-bay spawning areas to the upper- Bay. Simulations driven by tides only, without wind forcing, do not result in large larval dispersal distances (<12 km). Our analysis indicates that larvae spawned in the Westport river, Apponagansett Harbor, Nasketucket Bay and Sippican Harbor have a high probability of successful transport to habitats suitable for juveniles. The implication is that these may be promising areas for reseeding efforts aimed at restoring and sustaining the bay scallop population.
University of Massachusetts Dartmouth
Department of Fisheries Oceanography
QUANTIFYING CONNECTIVITY IN BUZZARDS BAY WITH APPLICATION TO
THE STUDY OF RECRUITMENT VARIABILITY IN THE BAY SCALLOP
(ARGOPECTEN IRRADIANS)
A Thesis in
Marine Science and Technology
by
Chang Liu
Submitted in Partial Fulfillment of the
Requirements for the Degree of
Master of Science
January 2014
I grant the University of Massachusetts Dartmouth the non-exclusive right to use the work
for the purpose of making single copies of the work available to the public on a not-for-
profit basis if the University’s circulating copy is lost or destroyed.
Chang Liu
Date
We approve the thesis of Chang Liu
Date of Signature
Geoffrey W. Cowles
Associate Professor of Fisheries Oceanography
Thesis Advisor
Kevin D.E. Stokesbury
Professor and Chairperson, Department of Fisheries Oceanography
Thesis Committee
James H. Churchill
Research Specialist, Woods Hole Oceanographic Institution
Thesis Committee
Louis Goodman
Associate Dean and Graduate Program Director
Steven Lohrenz
Dean, School for Marine Science and Technology
Tesfay Meressi
Associate Provost for Graduate Studies and Research Development
ABSTRACT
Quantifying Connectivity in Buzzards Bay with Application to the Study of Recruitment
Variability in the Bay Scallop (Argopecten irradians)
by Chang Liu
The bay scallop has been the foundation of a historically important fishery in Buzzards
Bay, MA. However, since the 1980s, the resource has been in decline and the population
has been subject to large inter-annual fluctuations. Due to the relatively short life span of
bay scallops (2-3 yr), the sensitivity of recruitment to the variability in larval transport to
juvenile habitat is high. Long-term Bay-wide restoration strategies will need to take into
account the influence of hydrodynamics on connectivity from spawning areas to juvenile-
suitable habitat. To examine this connectivity in Buzzards Bay, a coupled biophysical
modeling approach is utilized. A high-resolution ocean model (FVCOM) was used to
drive an individual-based model of scallop larval transport from spawning to settlement.
Connectivity among geographic sub-populations is quantified using Lagrangian probability
iii
density functions. The spawning-to-settlement-area connectivity of scallop larvae is
principally due to the wind-driven transport. During late spring to summer, the principal
time of bay scallop spawning, the southwest sea breeze dominates the wind forcing of Bay
currents. The up-Bay flow driven by the surface wind stress in the shallow regions of the
Bay results in substantial transport of larvae from mid-bay spawning areas to the upper-
Bay. Simulations driven by tides only, without wind forcing, do not result in large larval
dispersal distances (<12 km). Our analysis indicates that larvae spawned in the Westport
river, Apponagansett Harbor, Nasketucket Bay and Sippican Harbor have a high probability
of successful transport to habitats suitable for juveniles. The implication is that these may
be promising areas for reseeding efforts aimed at restoring and sustaining the bay scallop
population.
iv
ACKNOWLEDGMENTS
First of all, my thankfulness goes to my advisor, Geoff Cowles, for his superb guidance
and support throughout my thesis work. I would also like thank the rest of my thesis
committee: Kevin Stokesbury and Jim Churchill, for their guidance and contribution. My
sincere thanks also go to Val Hall who is very knowledgeable on bay scallop biology,
and Stephen Tettelbach who provided answers to my questions on bay scallop habitat.
My acknowledgement also goes to Brant McAfee (Massachusetts Division of Marine
Fisheries), Joe Costa (Buzzards Bay National Estuary Program), and Rich Signell (U.S.
Geological Survey) who provided data used in this thesis work. My special thanks also
go to many fellow students in SMAST for being good friends of mine and making my
graduate student life more exciting. This work was supported in part by the Woods Hole
Sea Grant program through award NA10OAR4170083. Computations were made on the
UMass Dartmouth GPU cluster which was acquired with support from NSF award CNS-
0959382 and AFOSR DURIP award FA9550-10-1-0354.
v
CONTENTS
List of Figures ix
List of Tables xiii
1 Introduction and Background 1
1.1 Bay Scallop Life History . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Bay Scallop Fishery . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.1 History and Regulations . . . . . . . . . . . . . . . . . . . . . . . 7
1.2.2 Resource Decline . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.2.3 Restoration Efforts . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1.3 Individual-Based Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.4 Setting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
1.5 Thesis Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
1.5.1 Hypotheses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
1.6 Thesis Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2 Approach 22
vi
2.1 Numerical Model Description . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.1 Hydrodynamic Model . . . . . . . . . . . . . . . . . . . . . . . . 22
2.1.2 Individual-Based Model . . . . . . . . . . . . . . . . . . . . . . . 24
2.2 Lagrangian Probability Density Functions . . . . . . . . . . . . . . . . . . 29
3 Numerical Experiments for Larval Dispersal 33
3.1 Delineating Spawning and Settlement Zones . . . . . . . . . . . . . . . . . 33
3.2 IBM Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.3 Connectivity Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.4 Settlement Success Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4 Results 53
4.1 Circulation and Lagrangian Results . . . . . . . . . . . . . . . . . . . . . . 53
4.2 Connectivity for Realistic Cases . . . . . . . . . . . . . . . . . . . . . . . 57
4.3 Impact of Wind on Dispersal and Connectivity . . . . . . . . . . . . . . . . 68
4.4 Settlement Success Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5 Discussion 79
5.1 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
5.2 Uniform Scenario . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
5.3 Wind-driven Transport . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.4 Sensitivity Experiments . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
vii
5.5 Ecological Implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.6 Limitations And Future Extensions . . . . . . . . . . . . . . . . . . . . . . 92
References 93
viii
LIST OF FIGURES
1.1 Distribution of bay scallop subspecies in the northeastern United States . . 3
1.2 Life cycle of the bay scallop . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.3 Total landings of bay scallops in Massachusetts, 1950-2010 . . . . . . . . . 10
1.4 A typical dredge used for commercial harvesting of bay scallops . . . . . . 10
1.5 Landings of bay scallops in Buzzards Bay, MA. . . . . . . . . . . . . . . . 13
1.6 Bathymetry and setting of Buzzards Bay, MA . . . . . . . . . . . . . . . . 17
1.7 The subtidal response of free surface elevation and vertically-averaged
velocities in Buzzards Bay to an idealized southwest sea breeze. . . . . . . 18
2.1 Bathymetry and unstructured grid for the GoM and SEMASS FVCOM . . . 25
2.2 LPDF reconstruction based on particle tracking positions . . . . . . . . . . 31
3.1 Bay scallop suitability areas in southeastern Massachusetts . . . . . . . . . 35
3.2 Spawning zones defined based on bay scallop suitability areas divided by
designated shellfish growing areas. . . . . . . . . . . . . . . . . . . . . . . 36
3.3 Eelgrass in southeastern Massachusetts . . . . . . . . . . . . . . . . . . . . 37
ix
3.4 Settlement zones defined based on extents of water with depth less than
3.5m MSL. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
3.5 Wind roses for early and late spawning simulations for the years 2008–2010 42
3.6 Time series of the idealized periodic variable wind used in Case I3 for a
given day . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.7 Daily-averaged wind speed plot for summer 2009 using July–Aug 2009
BUZM record. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
3.8 Wind rose for the period from 20 December 2009 to 19 January 2010. . . . 47
3.9 Temporal spawning pattern used in the IBM. . . . . . . . . . . . . . . . . . 47
3.10 Temporal settlement distribution weight function D(t). . . . . . . . . . . . 50
3.11 An example of a connectivity matrix element (see text). . . . . . . . . . . . 50
4.1 Time-averaged vertically-averaged velocity field of 2010 early spawning
event . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.2 Subtidal variance ellipses of the depth-averaged velocity in the interior of
the Bay in along- and cross-Bay directions . . . . . . . . . . . . . . . . . . 56
4.3 Locations of individuals released from Zone 14 in 2010 early spawning
case (case R5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4 Lagrangian probability density function of individuals released from Zone
14 during the 2010 early case (R5) . . . . . . . . . . . . . . . . . . . . . . 59
4.5 Connectivity matrices for realistic cases . . . . . . . . . . . . . . . . . . . 60
x
4.6 The spawning-to-settlement-zone-connectivity between grouped regions
for the 2008 early case (R1) . . . . . . . . . . . . . . . . . . . . . . . . . . 64
4.7 Schematic diagram of connectivity pattern among grouped regions repre-
senting the average outcome over the six realistic cases . . . . . . . . . . . 65
4.8 Mean, standard deviation, and coefficient of variance of the connectivity
matrices for 6 realistic cases . . . . . . . . . . . . . . . . . . . . . . . . . 66
4.9 Spawning zones and their corresponding averaged Shannon-Wiener index
(H
0
i
). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.10 Connectivity matrices for the idealized model runs (I1-I4, Table 3.1). For
definitions, see Fig. 4.5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
4.11 Time-averaged vertically-averaged velocity field of 2009 NW-dominant
idealized spawning event (Case I4) . . . . . . . . . . . . . . . . . . . . . . 72
4.12 Zone settlement success (ZSS) for all 25 spawning zones in all numerical
experiments. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.13 Zone settlement success (ZSS) and average zone success (AZS) for all 25
spawning zones in all experimental cases. . . . . . . . . . . . . . . . . . . 77
4.14 The mean (left) and the standard deviation (right) of the time-averaged
streamfunctions over the six realistic cases. . . . . . . . . . . . . . . . . . 78
4.15 Histogram of low-pass filtered larval trajectory lengths of 5000 randomly
sampled individuals during a 14-day larval duration in Case R1. . . . . . . 78
xi
5.1 Connectivity matrix for uniform scenario. . . . . . . . . . . . . . . . . . . 81
5.2 Schematic diagram of the connectivity pattern among grouped regions for
the winter wind idealized case (I4) . . . . . . . . . . . . . . . . . . . . . . 85
5.3 Comparison between the connectivity from the idealized NW-dominant
case and the mean value from the six realistic cases. . . . . . . . . . . . . . 86
5.4 Comparison of ZSS for the eelgrass sensitivity case. . . . . . . . . . . . . . 90
5.5 The AZS of the six realistic cases (a) and commercial bay scallop landings
(Table 1.3) in Buzzards Bay (b). . . . . . . . . . . . . . . . . . . . . . . . 91
xii
LIST OF TABLES
1.1 Shellfishing seasons and bay scallop catch limits in select towns on
Buzzards Bay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
1.2 Bay scallop landings in bushels from four towns or islands in Massachusetts 12
1.3 Commercial bay scallop landings in Buzzards Bay, MA. . . . . . . . . . . 12
1.4 Key geographical data for Buzzards Bay . . . . . . . . . . . . . . . . . . . 17
3.1 IBM model runs for connectivity experiments. . . . . . . . . . . . . . . . . 41
3.2 Wind statistics for realistic cases . . . . . . . . . . . . . . . . . . . . . . . 41
3.3 Summary of the biological assumptions employed in the IBM study . . . . 45
4.1 Mean and standard deviation of the depth-averaged velocity in the interior
of the Bay in along- and cross-Bay directions . . . . . . . . . . . . . . . . 62
4.2 Grouping of the connectivity zones. . . . . . . . . . . . . . . . . . . . . . 62
4.3 Connectivity levels between larger grouped areas for realistic cases. . . . . 74
4.4 Summary of ZSS data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
xiii
xiv
4.5 ANOVA table from the test to determine if zone settlement success (ZSS)
differs between spawning events in different years and spawning zones . . . 76
5.1 Connectivity levels between larger grouped areas for the cross-Bay wind
idealized case (I4). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
CHAPTER 1 INTRODUCTION AND BACKGROUND
Bay scallops constitute a major resource for the fishing economy of southeastern Mas-
sachusetts, USA. However, the value of the resource has been in decline since the 1980s,
and varies significantly from year to year, particularly in local embayments where yearly
harvests may differ by an order of magnitude. Such trends have led to efforts to enhance
scallop populations in specific embayments through seeding, but these have been met with
varying degrees of success. An improved understanding of bay scallop connectivity and the
factors that influence juvenile scallop recruitment variability would be of value in predicting
the impact of management decisions on the local bay scallop populations, and thus be of
benefit to those charged with regulating and safeguarding this important resource.
1.1 BAY SCALLOP LIFE HISTORY
The bay scallop, Argopecten irradians, is a highly fecund and relatively short-lived bivalve
shellfish species (Robert, 1978). Along the eastern seaboard of the United States there
are several subspecies (Fig. 1.1), with the northernmost subspecies, A. irradians irradians
(Lamarck), residing in Massachusetts waters. The life cycle of the bay scallop is divided
1
2
into two distinctive phases: a planktonic larval phase and a benthic phase (Fig. 1.2). The
transport of larvae during the planktonic phase is controlled by physical processes including
tides and wind-driven circulation. The bay scallop has a relatively short life span, normally
12–24 months (Blake and Shumway, 2006). Typically the population is comprised of only
two year classes; thus bay scallops are particularly susceptible to population fluctuations.
Very few of the individuals survive to spawn a second time (Belding, 1910; Barber and
Blake, 1983; Bricelj et al., 1987).
Along the east coast of the United States, bay scallops are found from the Gulf of
Mexico north to Massachusetts Bay (Belding, 1910; Heffernan et al., 1988; Bert et al.,
2011). Bay scallops are principally distributed in enclosed waters such as bays, harbors,
estuaries and sounds (Fig. 1.1; Gutsell, 1930; Blake and Shumway, 2006). The species is
subtidal and ranges from 1.5 to 10.0 m below mean low water, but may be found in depths
as great as 20.0 m (Gutsell, 1930). In the coastal waters of southeastern Massachusetts, bay
scallops occur in the shallow (1.2–2.4 m) waters of the islands of Martha’s Vineyard and
Nantucket, as well as those along the south shore of Cape Cod, the southeastern portion of
Cape Cod Bay, and Buzzards Bay (MacKenzie, 2008).
Bay scallops are hermaphroditic and generally protandrous, which means they spawn
first as males then as females (Loosanoff and Davis, 1963). The species is highly fecund,
and usually spawn at one year of age, producing approximately two million eggs on average
per individual (reviewed by Hall, 1984). Temperature is an essential factor triggering the
3
Figure 1.1: Distribution of bay scallops in the northeastern United States (excerpted from
Bert et al., 2011)
4
spawning. For late spring spawning, bay scallop reach maturation and spawn when the
water temperature increases from 20 to 30
C (Castagna, 1975; Barber and Blake, 1983).
In Massachusetts, most spawning occurs during June and July as the water temperature
increases (Belding, 1910; Taylor and Capuzzo, 1983; Bricelj et al., 1987). A second, later
spawn of bay scallops during their first season also occurs (Taylor and Capuzzo, 1983;
Tettelbach et al., 1999; MacKenzie, 2008). This second spawning event usually happens in
the late summer or fall and may play a significant role in the persistence of the population
from year to year (Tettelbach et al., 1999). These scallops are commonly referred to as fall
spawners.
Sastry (1965) reared bay scallops in the laboratory from fertilized eggs to pre-adult
phase at 24.0±1.0
C and recorded data on their physiology and metamorphosis. The total
time between these two stages, hereafter referred to as the pelagic larval duration (PLD),
is typically 14 days (Loosanoff and Davis, 1963; Sastry, 1965; MacKenzie, 2008), but can
range from 10 to 19 days depending on water temperature and food availability (Castagna
and Duggan, 1971). The average diameter of a bay scallop egg is 62 µm (Sastry, 1965).
After release into the water column, the eggs take on a spherical form and are commonly
located near the bottom following fertilization. The time span of the egg stage ranges from
9 to 24 hours. Trochophore larvae first appear 24 hours following fertilization (Fig. 1.2
2
),
and all surviving eggs will have completed the transition within 48 hours of fertilization.
The average size of trocophore larvae ranges from 78 to 82 µm (Sastry, 1965). Shortly
5
1
2
3
4
5
6
7
Tidal and wind
currents disperse
When water temperatures exceed 18°C
(65°F), adult bay scallops release eggs
and sperm into the water.
Within a few hours,
a trochophore
larva forms.
shell forms as a D-shaped
veliger, which lasts for 2-3
days. These larvae can swim
with an appendage called a
velum. (Size = 0.1mm)
At the pedi-veliger stage, the larva
grows a foot, which it uses to seek
out a place to settle. But it can still
use its velum to swim. This stage
can last from a few days to a week.
(Size = 0.25mm)
When big enough, the juvenile
Adult scallops typically live for 1 to 2
years, spawning once during their
lifetimes. After spawning, scallops
can be harvested in the fall.
Scallops settle as spat on eelgrass
blades and attach for several
weeks with sticky threads they
secrete. This protects them from
(Size = 0.3-5mm)
Jack Cook, WHOI
A scallop’s life
What happens to the larvae of shellsh between birth and settling down remains a mystery. Filling in the knowledge gaps is
essential for nding ways to manage and restore shellsh beds.
Figure 1.2: Life cycle of the bay scallop (reprinted from Oceanus Magazine, Vol. 47, No.
1, 2008)
6
after development into the trochophore stage, they begin to transform into veliger larvae
(Fig. 1.2
3
). The average size of the larvae at this stage is 101 µm and the stage duration
of veligers is about 10 days. Approximately 12-14 days after fertilization, the foot, anterior,
and posterior adductor muscles appear, and the velum is completely reduced. This is the
pediveliger stage in which average size is 184 µm (Fig. 1.2
4
). Metamorphosis to the
juvenile stage also occurs in this stage. The juvenile stage is characterized by settlement
and appearance of the dissonconch shell. After attachment to eelgrass or other suitable
habitat, they will generally remain stationary until they attain a shell size of 20 to 30 mm,
when they descend to the bottom (Robert, 1978; Fig. 1.2
6
).
The preferred habitat of the bay scallop are eelgrass beds (Zostera marina L.) and other
seagrasses growing on sand substrates (Belding, 1910, Fig. 1.2
5
). The substrate of bay
scallop habitat ranges from silt to hard sand. While a certain degree of silt may be tolerated
by adult bay scallops, it is unfavorable to newly settled juveniles.
Bay scallops are essentially non-migratory. In a study carried on in Nantucket Harbor,
MA, when adult scallops were released in close proximity to suitable habitat, they remained
stationary rather than moving to the suitable habitiat (Belding, 1910). The larval stage is
thus likely the most critical in terms of individual movement and population dispersal.
7
1.2 BAY SCALLOP FISHERY
1.2.1 HISTORY AND REGULATIONS
The local bay scallop fishery extends back long before the arrival of the Europeans.
Excavated middens of Native Americans on Martha’s Vineyard, Massachusetts were found
to be full of bay scallops (MacKenzie, 2008). The bay scallop fishery was important
to the coastal communities of Massachusetts from the 1870s to the mid 1980s. Large
landings provided the local communities with considerable economic benefits (Fig. 1.3).
Bay scalloping was a full-fledged industry with shucking houses framing the edges of many
of the local harbors.
The Commonwealth of Massachusetts sets the broad limits of the fishing season and
daily harvests per fisherman and it has overall jurisdiction over the fishery (Belding, 1910).
The legal season usually begins at the beginning of October (November 1st in some towns),
and lasts until the end of March (Table 1.1). The Commonwealth’s Shellfish Act of 1880
entrusted all regulation of the shellfisheries to the town selectmen, who fine-tuned the state
laws to the needs of their communities (MacKenzie, 2008; Table 1.1). Scallop fishermen
have been allowed to harvest bay scallops and other estuarine bivalves only within the
borders of the towns in which they are legal residents. In contrast with shellfish species
such as oysters and quahogs that have several year classes, bay scallops consist principally
of only two year classes, which makes it easier for management agencies to regulate the
8
scallop fishery (MacKenzie, 2008). Scallop fishermen are allowed by law to harvest the
entire population of the older year class. State regulations require that only adult scallops
containing an annual ring on their shells can be harvested. Juvenile scallops, also known
as young-of-the-year (YOY) or seed scallops, do not have an annual growth ring and must
be discarded.
The modern fishery consists of both commercial and recreational components. Recre-
ational fishers use rakes, tongs or dip nets and commercial fishers primarily use dredges.
The bay scallop dredge is 28–36 in (70–90 cm) wide and has a long V-shaped iron bridle.
A typical dredge has metal rings along the bottom with cloth netting, 3-4 in (7.5-10 cm),
along its top (MacKenzie, 2008; Fig. 1.4). The commercial fishing process consists of
hauling up the dredge, culling the juveniles (seed scallops without a growth ring) on deck
and discarding them back to the water. Adult scallops, which presumably will not spawn
again, are retained. The ring size on the dredge is set so that seed scallops are not trapped
in the dredge. However, the year class selectivity of the gear is not 100% effective.
Local regulations may also maintain restrictions on air-sea temperature differences so
that the discards are not subject to increased mortality from cold shock (Table 1.1). In
Massachusetts, a bushel of harvest is equivalent to approximately 350 bay scallops, which
provides roughly 6 pounds of meats (MacKenzie, 2008). In an effort to promote branding,
local names such as “Cape Scallops” (Cape Cod), “Nantucket Scallops, and “Peconic
Scallops” (Peconic Bay, Long Island) have been used to contrast locally-harvested scallops
9
from those derived from extensive aquaculture projects in Eastern Asia. Taylor Seafood,
Inc., located in Fairhaven, MA, runs a successful local bay scallop aquaculture operation
(Zone 11 in Fig. 3.2). Their product, “Taylor Bays”, is widely recognized.
At present there are no targeted surveys of bay scallop abundance in Buzzards Bay.
The Massachusetts Division of Marine Fisheries (MassDMF) conducts an inshore dredge
survey in the fall and spring but the draft limitations of the vessel exclude much of the
known bay scallop habitat. Lacking a good assessment of abundance, landings data are the
best available means to estimate the population structure in the Bay. Given that essentially
all the adult scallops are harvested in a given year, it is reasonable to use catch as a proxy
for population in the case of bay scallops (MacKenzie, 2008).
1.2.2 RESOURCE DECLINE
Bay scallop landings experienced a remarkable decline throughout Massachusetts be-
ginning around 1985 (Fig. 1.3; Fig. 1.5; Table 1.2). Historically, the total catch in
Massachusetts waters was greater than 500,000 lbs and, at times, as high as 2,000,000
lbs until the early 1990s when it fell to 100,000 lbs and at present has not appreciably
recovered from these low numbers. Data also indicates that there are significant interannual
fluctuations in bay scallop populations. From 1950 to 1990, landings varied more than an
order of magnitude on an annual basis. Within Buzzards Bay, recent data suggest that
annual landings ranged from 190 to 189,334 lbs between 2005 to 2010 (Table 1.3).
10
0"
500,000"
1,000,000"
1,500,000"
2,000,000"
2,500,000"
1950"
1952"
1954"
1956"
1958"
1960"
1962"
1964"
1966"
1968"
1970"
1972"
1974"
1976"
1978"
1980"
1982"
1984"
1986"
1988"
1990"
1992"
1994"
1996"
1998"
2000"
2002"
2004"
2006"
2008"
2010"
!"#$%#&'()*+,#$'(+-(./"0'1(
2/"3(
Figure 1.3: Total landings of bay scallops in Massachusetts, 1950-2010. Source of Data:
NMFS Landings Statistics
Figure 1.4: A typical dredge used for commercial harvesting of bay scallops.
11
Table 1.1: Shellfishing seasons and bay scallop catch limits in select towns on Buzzards
Bay. Regulations are gathered from municipal websites.
Town (MA) Season Catch Limit
Westport 1.5 bushel/week (recreational), 7
bushels/day (commercial)
Fairhaven
a
Nov1-Mar31 1 bushel/week (recreational)
Mattapoisett
b
Nov1-Apr1 1 bushel /week
Marion
c
Nov1-Mar31 /
Nov1-May31 (by
area)
No data found
Wareham
d
Oct1-Mar 31(or
start Oct 30) [no
Sunday]
1 bushel/week (recreational), 5 bushels
daily (commercial) or 10/boat (max 2 per-
mits). Temperature Restrictions applies
that shellfish shall not be harvested in
areas containing less than two (2) feet
of water at mean low tide when the
air temperature is less than 32 degrees
Fahrenheit.
Bourne
e
Oct 1 - Mar 31
(no commercial
Sunday)
1 bushel/week (recreational) 5
bushels/day (commercial), 10
bushels/boat (2 permits)
a
http://fairhaven-ma.gov/pages/fairhavenMA_harbor/Regs/
Shellfish%20Regulations%202011.pdf
b
http://www.mattapoisett.net/Pages/MattapoisettMA_natural/
shellfishlimits.pdf
c
http://www.marionma.gov/Pages/MarionMA_Harbormaster/
openclosedshellfishareas.pdf
d
http://www.wareham.ma.us/Public_Documents/WarehamMA_
Harbormaster/ShellfishRegs2011.PDF
e
http://www.townofbourne.com/LinkClick.aspx?fileticket=8%
2bimJDfMUvs%3d&tabid=160&mid=829
12
Table 1.2: Bay scallop landings in bushels (avg./time period) from four towns and islands
in Massachusetts. Westport borders on Buzzards Bay, Chatham is on Cape Cod, while
Martha’s Vineyard and Nantucket are separate islands. The data are binned into six 5-
yr periods from 1976 to 2005 (summarized by MacKenzie, 2008 from local town annual
reports).
Period Westport Chatham Martha’s Vineyard Nantucket
1976-1980 6,800 33,200 23,800 63,300
1981-1985 13,750 25,500 34,300 66,100
1986-1990 200 700 24,900 28,500
1991-1995 100 500 7,900 14,500
1996-2000 780 300 6,200 8,800
2001-2005 70 400 9,600 12,500
Table 1.3: Commercial bay scallop landings in Buzzards Bay, MA.
Area Group 2005 2006 2007 2008 2009 2010 2011 2012
BB0X 1,739 2,850 20,826 4,875 4,725 199 8,073 0
BB1X 0 994 11,650 1,823
BB2X 0 26,847 124,317 13,485
BB3X 0 15,200 27,514 699 2,033
BB4X 3,472 4,272 1,843 17,606 9,155 27,708
BB5X 2,707 0 4,322 11,973 2,572
Total 7,918 7,122 23,663 68,850 189,334 199 8,772 47,621
a) Data source: MA Commercial Catch Reports, compiled by Massachusetts Division of Marine Fisheries.
b) Numbers are in live pounds (whole animal, shell on).
c) Area Groupings represent designated shellfish growing areas (MA DFG, 2009) in intervals of 10, i.e.
BB2X = BB21–BB29.
d) Landings combined into multiple areas to maintain confidentiality.
13
0
10,000
20,000
30,000
40,000
50,000
60,000
1955
1956
1957
1958
1959
1960
1961
1962
1963
1964
1965
1966
1967
1968
1969
1970
1971
1972
1973
1974
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
Bushels
Years
Bay Scallop Landings in Buzzards Bay
Figure 1.5: Landings of bay scallops in Buzzards Bay, Mass. Note the decline following
the late 1970s and early 1980s. Source of data: Massachusetts Division of Marine Fisheries
and Buzzards Bay National Estuary Program.
The recruitment failure of the bay scallops may be caused by algal blooms, deterioration
of habitat (eelgrass decline), changes in sediment, bacterial contamination, urbanization,
increased predation, etc. (MacFarlane, 1999). In addition, due to the relatively short life
span of bay scallops, the sensitivity of recruitment to interannual variability in physical
forcing is very high. Tidal flushing or changes in wind-driven circulation from year to year
and can contribute to recruitment failure.
Loss of eelgrass has been linked to the resource decline of the bay scallop. Costa
et al. (1992) suggest that decline in the bay scallop harvest is associated with eelgrass
declines in Waquoit Bay, Falmouth from 1950s to 1990s. Eelgrass is sensitive to water
pollution, so the distribution of eelgrasses is often used as an indicator for changes in water
quality. The Massachusetts Department of Environmental Protection (MassDEP) initiated
14
a comprehensive, statewide eelgrass mapping program in 1993. Recent MassDEP mapping
efforts indicate that 35% of the eelgrass in Buzzards Bay had been lost between 1996 to
2006 (Costello and Kenworthy, 2011).
1.2.3 RESTORATION EFFORTS
A long-term, intensive bay scallop restoration effort was initiated in Peconic Bay, Long
Island, NY, after the occurrence of the brown tide algal blooms in the late 1980s that almost
eradicated the local population (Tettelbach and Smith, 2009). The method used for early
restoration was “free planting” of scallops in gap areas by dispersing them from a boat
by hand. The planting was performed at low densities to reduce the effects of predation.
As an effort to ensure higher fertilization rate, the later stage of the work employed a
different approach of “jump-starting” by planting artificially reared adults at high densities
(Tettelbach and Smith, 2009).
Local restoration efforts have centered on planting seeds in areas deemed to be suitable
juvenile habitat. In 1995 MassDMF initiated a bay scallop restoration experiment
1
. Bay
scallop larvae were cultured in the MassDMF lobster hatchery until they completed the
free-swimming period and settled. They were subsequently transferred to Taylor Seafood,
Inc. in Fairhaven, MA where they were grown to a size suitable for planting. The sites
chosen for planting were Menemsha Pond, Nasketucket Bay and West Falmouth Harbor.
Larger scallop seeds showed improved survival when compared with smaller scallops.
1
DMF News. Vol. 15 Fourth Quarter. 1995. http://www.mass.gov/dfwele/dmf/
publications/dmfnq495.htm
15
Some local towns raise their own seed or acquire them from Taylor Seafood, Inc. or
from the National Marine Fisheries Service Laboratory in Milford, CT. For example, the
town of Marion purchased and planted 20,000, 50,000, and 35,000 seeds respectively in
2007, 2008 and 2009. Since 2009 the Marion shellfish wardens have abandoned their
efforts, as they did not see any improvement in local catch and could not justify the expense.
They have continued seeding other shellfish such as oysters.
1.3 INDIVIDUAL-BASED MODELING
The individual-based modelling (IBM) approach has been widely used to study coupled
bio-physical processes that influence the transport of larval marine organisms including
both vertebrates and invertebrates. This approach involves coupling physical fields of
currents and ocean properties from hydrodynamic of ocean circulation models with
Lagrangian particle tracking models that represent individuals or groups of individuals.
In the modern approach, the physical fields are derived from three-dimensional ocean
models (e.g. FVCOM, ROMS, POM, HYCOM) which can simulate the spatially and
temporally-varying velocity, diffusivity, temperature, and salinity. The velocity field is used
to drive the advection of individuals using a Lagrangian approach. Diffusion is typically
implemented using a random walk formulation with separate vertical and horizontal
components. Some models may also track additional biological state variables for the
16
individual (e.g. instar, settlement competency) which may rely on hydrographic fields as
well as other ocean properties from the circulation model such as the eddy diffusivity.
The IBM approach can support the investigation of a wide range of phenomena, such
as the connectivity of spawning and nursery areas. Example studies include examinations
of population connectivity between specific regions (e.g. Siegel et al., 2008; Xue et al.,
2008), quantification of larval retention (e.g. Banas et al., 2009; Churchill et al., 2011), and
evaluation of the effects of physical forcing on larval dispersal (e.g. Tian et al., 2009a).
1.4 SETTING
Buzzards Bay is a shallow embayment with a mean depth of 11 m and approximate
dimensions of 45 km in the lengthwise (SW-NE) direction and 12 km in the transverse
direction (NW-SE) (Table 1.4; Fig. 1.6). At its mouth, Buzzards Bay is connected to the
New England Shelf via Rhode Island Sound. Along the southeastern edge it is connected to
Vineyard Sound via several narrow openings (e.g. Quicks Hole, Woods Hole). At its upper
extent the Bay is connected to Cape Cod Bay through the Cape Cod Canal. Salinity in the
Bay ranges from 30-32 and temperature in the Bay ranges from 0-20
C. The density-driven
flow in the Bay is quite weak relative to the wind and tidal driven flow (Signell, 1987). A
key aspect of the circulation is the two-way flow setup in response to along-Bay wind
forcing. When this circulation structure is in place, the depth-averaged flow is downwind
in the shallow regions along the Bay’s edges and is upwind in the deeper interior of the Bay
17
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30
10 20 30 40 50 60 70 80 90 100
Cape Cod Canal
Southeastern
Cape Cod
Buzzards Bay (BB)
Vineyard Sound (VS)
Massachusetts
Figure 1.6: Bathymetry (m) and setting of Buzzards Bay, MA. The black triangle indicates
the location of station BUZM3 of National Data Buoy Center.
Table 1.4: Key geographical data for Buzzards Bay
Approximate length (SW-NE) 45 km
Approximate width (NW-SE) 12 km
Surface area 644 km
2
Max depth 34 m
Mean depth 11 m
Length of coastline 563 km
18
-0.05
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
0.04
0.05
Wind Vector: 0.1 N
Currents: 20 cm/s
Free Surface Relative to DPLY3 with Vert-Avged Vecs
10371.8Transect Transport (m
3
/s):
CMLAB@SMAST/UMass Dartmouth
gcowles@umassd.edu
Figure 1.7: Model results depicting the subtidal response of free surface elevation (color
ramp) and vertically-averaged velocities (black arrows) in Buzzards Bay to an idealized
southwest sea breeze (red arrow). Computations made using the Finite Volume Community
Ocean Model.
19
(Fig. 1.7). This phenomena was originally studied by Csanady (1973) in the context of the
wind-driven circulation of lakes and was later examined in the context of Buzzards Bay by
Signell (1987). The wind forcing in Buzzards Bay is seasonal with Northwest (cross-Bay)
winds being dominant in the late fall to early spring and the Southwest (along-Bay) sea
breeze being dominant between June and September (Signell, 1987). The sea breeze begins
as a low atmospheric pressure forms over Taunton, MA and develops through the day
where it peaks in wind strength in Buzzards Bay during the late afternoon. Due to the the
bi-directionality of the vertically-averaged flow, the wind-driven circulation can contribute
significantly to the dispersal of larvae. The tidal response in the Bay is effectively a standing
wave with tidal currents of weak to moderate strength (<5 cm/s) with the exception of
regions near the holes separating the Bay from Vineyard Sound where differences in the
tidal amplitude and phase between Buzzards Bay and Vineyard Sound drive strong currents
and contribute to a complex residual flow field which can influence larval drift via the
mechanism of tidal eddy dispersion (Signell, 1987).
1.5 THESIS OBJECTIVES
The purpose of this thesis research is to investigate bay scallop larval dispersal and
spawning-to-settlement-zone connectivity in the context of habitat and management. A
coupled biophysical modeling approach is employed to examine the following questions:
20
a) How are the spawning and settlement zones of the bay scallop within discrete manage-
ment zones connected at time scales of the larval duration?
b) How do the physical forcing and its variability regulate this connectivity?
c) What are the important source regions for the existing bay scallop habitats within the
Bay?
1.5.1 HYPOTHESES
The hypotheses that this thesis aims to test are as follows:
H
1
: Connectivity is equal between discrete zones within Buzzards Bay on time scales
relevant to the larval duration of the bay scallop.
H
2
: The mechanisms of wind-driven exchange flow and tidal eddy exchanges provide
equal contributions to the dispersion of scallop larvae within Buzzards Bay.
H
3
: The influence of interannual variability of Bay circulation on the connectivity of
bay scallop is negligible.
1.6 THESIS OVERVIEW
In this thesis, a coupled IBM modeling approach is employed to investigate bay scallop
larval dispersal and connectivity in Buzzards Bay. Connectivity analyses are conducted us-
ing Lagrangian probability density functions (LPDF). Larval transport success is quantified
21
with regards to the larval settlement pattern of the bay scallop. In Chapter 2, the coupled
biophysical modelling approaches, which are the foundation of this work, are described
in detail. The method of computing LPDFs, which forms the basis of the connectivity
study, is presented. In Chapter 3, the detailed setup for the numerical experiments aimed
at determining bay scallop larval connectivity in the context of habitat and management
is described. The results are presented in Chapter 4. A summary of the study, as well as
interpretations and implications of the findings are discussed in Chapter 5.
CHAPTER 2 APPRO ACH
To test our hypotheses, we employed a coupled biophysical modeling strategy for bay scal-
lop dispersal using a high-resolution ocean model and individual based model approach.
Our modeling system consists of the Finite Volume Community Ocean Model (FVCOM),
used to simulate flows within Buzzards Bay and its embayments, and an Individual Based
Model (IBM) of scallop egg/larval transport from spawning areas. Using the results of
the IBM simulations, Lagrangian probability density functions (LPDF) were constructed
to quantify the connectivity among subregions in the domain.
In this chapter, details of the hydrodynamic model and IBM approach are presented.
Then, the construction of the LPDF from the IBM-generated larval tracks is described in
detail.
2.1 NUMERICAL MODEL DESCRIPTION
2.1.1 HYDRODYNAMIC MODEL
To compute the hydrodynamic fields, the Finite-Volume Community Ocean Model (FV-
COM) was employed (Chen et al., 2003, 2006; Cowles, 2008). FVCOM is an open source
22
23
model developed at the University of Massachusetts-Dartmouth with over 2500 registered
users. The kernel of FVCOM computes a solution of the hydrostatic primitive equations
on unstructured grids in the horizontal plane using a second-order accurate finite-volume
formulation for spatial fluxes (Kobayashi et al., 1999). The vertical coordinate is discretized
using a generalized terrain-following approach. FVCOM has been successfully applied to
a wide array of coastal and open ocean studies.
To model the circulation in the area of interest, a one-way nested FVCOM system
was employed. Local modeling of Buzzards Bay was performed using the high-resolution
Southeastern Massachusetts (SEMASS) FVCOM model (Fig. 2.1). The unstructured
mesh of SEMASS FVCOM model consists of 255K elements and 137K nodes and was
generated using the SMS (Surface Water Model System, http://www.aquaveo.
com/sms) software. The vertical coordinate was discretized using 30 evenly spaced σ-
layers. Horizontal resolution in SEMASS FVCOM varies from 5.0 km over the outer
shelf to 500 m along the coast. Resolution was enhanced within Cape Cod Canal
(150 m) and Buzzards Bay (50 m). Bathymetry for SEMASS FVCOM (Fig. 2.1) was
determined using a compilation of relevant bathymetric data products including: the
United States Geological Survey (USGS) 30 arc-second Gulf of Maine database (Roworth
and Signell, 1998), the NOAA 1/3 arc-second Nantucket Inundation Digital Elevation
Model (covering the southern portion of Nantucket Sound; Eakins et al., 2009) and
USGS Buzzards Bay SWATH Bathymetry (http://woodshole.er.usgs.gov/
24
project-pages/coastal_mass/html/buzz_bay.html). The Massachusetts
portion of the coastal boundary was based on a high-resolution coastline developed by
the Massachusetts Office of Coastal Zone Management (http://maps.massgis.
state.ma.us/map_ol/moris.php).
The local model was nested in a larger, regional-scale Gulf of Maine (GoM) FVCOM
model (Fig. 2.1), i.e., the regional model provides open boundary forcing for the local
model. The regional GoM model was based on the Northeast Coastal Ocean Forecast Sys-
tem (NeCOFS; http://fvcom.smast.umassd.edu/research_projects/NECOFS/
index.html) developed in the Marine Ecosystem Dynamics and Modeling Lab at the
University of Massachusetts Dartmouth under the guidance of Changsheng Chen (UMass
Dartmouth). The regional scale mesh consists of 27,571 elements and 14,845 nodes with
30 layers evenly spaced in the vertical σ-coordinate. The regional GoM model is forced
with hydrography and sea surface height at the open boundary, buoyancy flux from the
major regional rivers, and wind stress and heat flux derived from regional hindcasts of the
Weather Research and Forecasting (WRF) model (see Cowles et al., 2008 for details on the
setup and integration).
2.1.2 INDIVIDUAL-BASED MODEL
To simulate the dispersal of the larvae we used an Individual Based Model (IBM) approach.
In this approach, a 3-D Lagrangian particle tracking technique, employing velocity fields
archived hourly from FVCOM, was used to model the advection of the larvae.
25
0
5
10
15
20
25
30
35
40
-71 -70.9 -70.8 -70.7 -70.6 -70.5 -70.4 -70.3 -70.2 -70.1 -70 -69.9
41.2
41.3
41.4
41.5
41.6
41.7
41.8
41.9
42
42.1
42.2
-74 -73 -72 -71 -70 -69 -68 -67 -66 -65 -64 -63 -62 -61 -60
37
38
39
40
41
42
43
44
45
50 100 150 200 250 300 350
Figure 2.1: Bathymetry (in meters) and unstructured grid for the nested GoM (regional)
and SEMASS (local) FVCOM system. Black solid line in the upper panel indicates the
boundary between nested grids. The lower panel shows part of the SEMASS grid.
26
The Lagrangian particle tracking can be run either online or offline. In the online track-
ing approach, particle tracking of individual particles is integrated in the hydrodynamic
model. The calculation of velocity field and particle tracking are performed within the
same time step. For offline tracking, the hydrodynamic model is run first and the velocity
and turbulence fields are archived in hourly increments. The particle tracking model is then
driven by the archived velocity fields. The primary advantage of the offline method is that
results from a single hydrodynamic model run can drive several different particle tracking
experiments.
For this work, we employed the open source IBM package FISCM (FVCOM-based i-
state Configuration Model, https://code.google.com/p/fiscm/). FISCM has
been employed in several IBM studies including the transport of haddock larvae on Georges
Bank (Petrik, 2011) as well as an examination of the physiological factors controlling
the biogeographical boundaries of several Arctic and subarctic copepods species (Ji et al.,
2011).
For each individual, the equation that describes the advection process is
d
dt
~
X(t) =
~
V (
~
X(t), t) (2.1)
where
~
X(t), a three-dimensional variable, is the position of the individual at time t.
~
V (
~
X(t), t), the velocity field evaluated at the position of the individual and time t, is
27
computed by linearly interpolating the flow field output from FVCOM. The 4th-order
explicit 4-stage Runge-Kutta scheme (ERK4) is employed to integrate the particle position
from
~
X(t) to
~
X(t + t).
The IBM time step is a fixed duration for the particle tracking model. Its value should be
carefully determined to maximize the accuracy of the predicted particle trajectory as well
as the computational efficiency. Another constraint of the advection time step is related to
the hourly output interval of FVCOM. There must be an integral number of advection time
steps within the hourly interval. A time step of 240 s was selected for the present work. A
sensitivity case to the IBM time step was conducted and described in Section 5.4.
A vertical random walk formulation (Visser, 1997; North et al., 2006) was implemented
in the IBM to include, statistically, the effects of subgrid-scale hydrodynamic processes. A
modified method was incorporated to include a particle displacement term that is a function
of the gradient of local diffusivity:
z(t + δt) = z(t) + K
0
v
δt + R
r
2
d
K
v
z(t) +
1
2
K
0
v
δt
δt, (2.2)
where R is a random factor with zero mean and standard deviation d =1; K
v
is the vertical
diffusivity provided by FVCOM output; K
0
v
= K
v
/∂z evaluated at z(t); and δt is the
sub-timestep for vertical random walk. Iterating the vertical random walk n times within
28
an advection time step by using a sub-time step δt = ∆t/n was able to ensure the accuracy
of the displacement term related to the gradient of local diffusivity (Visser, 1997).
The horizontal diffusivity from FVCOM uses the Smagorinsky scheme which is not
capable of resolving spatial and temporal variation of the horizontal eddy diffusivity in the
Bay. Therefore a horizontal random walk was not implemented in this study.
Boundary conditions were implemented in the IBM model to regulate the behavior
of individuals when they cross the horizontal coastal boundaries. There are a number of
possible approaches including boundary reflection as well as freezing the individuals in
place following boundary penetration (e.g., Mitarai et al., 2009; Roughan et al., 2011).
In this study, the number of individuals needs to be conserved for the subsequent LPDF
function construction. Therefore boundary conditions were implemented to ensure the
individuals remain within the domain. If an individual moves across a horizontal boundary
during a time step, it is restored to its last known position calculated from the previous
timestep:
~
X(t + t) =
~
X(t). (2.3)
Given that the coastal ocean models do not resolve the very nearshore flow and the
boundary treatment may influence connectivity results, further study of the lateral boundary
condition is needed. A sensitivity study to the horizontal boundary conditions is introduced
in Section 5.4. The vertical boundary condition was straightforward. If an individual
29
moves above or below the water column due to advection or diffusion, it is reflected. This
treatment was realized according to
z(t + δt) =
z(t + δt), : if z(t + δt) < 0
2H z(t + δt), : if z(t + δt) > H
, (2.4)
where H is the depth of the water column in which the individual is located.
2.2 LAGRANGIAN PROBABILITY DENSITY FUNCTIONS
To quantify connectivity, we used the results from the Lagrangian particle tracking
simulation to determine a Lagrangian probability density function (LPDF). An LPDF
f
X
(~x; τ) is defined as the probability density (probability per unit area) of an individual
from a group of individuals X being at a location ~x = (x, y) at time τ (Mitarai et al.,
2009). LPDF approaches have been widely used in processing dispersal patterns driven by
turbulent processes. This approach provides estimation of tracer concentrations, which can
be used to measure coastal connectivity.
To construct the LPDF, we selected the 2-D kernel density estimator (Botev et al., 2010)
due to its inherent properties. The kernel density estimator is a non-parametric technique
for estimation of probability density functions. It is also computationally efficient. Optimal
values for the bandwidth parameters are automatically calculated. As presented by Botev
30
et al. (2010), given a set of particle locations X {
~
X
1
(τ),
~
X
2
(τ), ··· ,
~
X
N
(τ)}, the
corresponding Gaussian kernels at the specific time τ
G(~x,
~
X
i
(τ), c) =
1
2πc
e
~x
~
X
i
(τ)
2
/(2c
2
)
(2.5)
is used to quantify the LPDF
f
X
(~x; τ) =
1
N
N
X
i=1
G(~x,
~
X
i
(τ), c), (2.6)
where c is a spatial scale commonly referred to as the bandwidth and is determined by an
optimized algorithm well-defined by Botev et al. (2010). Note that the dimension of the
LPDF f
X
(~x; τ) is 1/(distance squared).
The procedure of constructing LPDFs numerically was to bin the particle locations into
an n-by-n square grid, calculate the probability density in each cell by dividing the number
of particles that fall in this cell by the size of the cell, and smooth the probability density
field using Gaussian kernels with the optimized bandwidth c. The only required parameter
is n, the number of boxes on one side of the square grid used for binning. A value of
n = 128 was chosen in the current work following a sensitivity study. Fig. 2.2 shows an
example of a reconstruction using such method.
31
Figure 2.2: An example of an LPDF reconstruction (bottom) based on particle tracking
positions (top). Colors indicate the probability density in km
2
.
32
Connectivity is defined as the probability of an individual from a given source zone
being transported to a given destination zone during a specific time period. Connectivity
between the adult and juvenile habitats of bay scallops within Buzzards Bay was computed
using LPDFs. To compute connectivity, model individuals were first released from a
defined source zone and dispersed throughout the domain for a specified time frame
using the Lagrangian tracking model. Using the individual tracks we then integrated the
probability density function over the destination zone to quantify connectivity from source
to destination. The computation of connectivity from source A to destination area, B, at a
specific time τ is defined mathematically as
P
AB
(τ) =
Z
B
f
A
(~x, τ)dB, (2.7)
where f
A
is the LPDF of individuals originating in source zone A and is calculated
according to Eq. 2.6.
CHAPTER 3 NUMERICAL EXPERIMENTS FOR LARVAL DISPERSAL
This chapter describes the detailed design of the IBM experiments including the establish-
ment of discrete spawning and settlement zones as well as the methodology employed for
the quantification of connectivity.
3.1 DELINEATING SPAWNING AND SETTLEMENT ZONES
In the present work, we delineated distinct regions to define the areas of spawning and
settlement. The distinction of spawning and settlement regions was necessary as the
distribution of adult bay scallop and that of habitats to which larvae are able to successfully
recruit do not necessarily coincide (S. Tettelbach, personal communication). In our model
studies, the spawning zones were established using bay scallop suitability areas identified
by MassDMF based on the extents of the adult distribution (MA DFG, 2009; Fig. 3.1).
These suitability areas were then divided into discrete zones using state-designated shellfish
growing areas (Fig. 3.2). Each spawning zone was assigned a name related to its geographic
location (e.g., Sippican Harbor). The total area of spawning zones is 119.05 km
2
,
comprising 18.5% of the total area of the Bay. Settlement zones were defined based
33
34
on the habitat where larvae are able to successfully settle. This habitat includes beds of
eelgrass and other submerged aquatic vegetation (SAV) that are ideal for bay scallop larvae
recruitment (Thayer and Stuart, 1974; Carroll et al., 2009). Regular surveys of eelgrass
extent have been conducted by the Commonwealth of Massachusetts (Fig. 3.3). However,
recent eelgrass surveys were conducted only in select embayments of Buzzards Bay, and
thus are not comprehensive. Based on the observed coverage of eelgrass, it was determined
that eelgrass is generally found in areas with depth not exceeding 3.5 m. Thus, for our
experiments, the areas meeting this depth constraint were used as an estimate of the extent
of SAV beds for our model study. These areas make up 14.7% of the total area of the
Bay and are subsequently divided into individual embayments as distinct settlement zones
that coincide approximately with spawning zones for the IBM experiments (Fig. 3.4). Two
additional settlement zones, with no corresponding spawning zones, are defined along the
Elizabeth Islands.
3.2 IBM CASES
A suite of test cases (Table 3.1), including both realistic and idealized, is used to test our
hypotheses (Section 1.5.1).
Realistic Cases For the realistic cases the coupled biophysical model was used to
simulate both early and late spawning of the bay scallops in Buzzards Bay for the
years 2008–2010 to examine the influence of the seasonal and interannual variability on
35
Figure 3.1: Bay scallop suitability areas in southeastern Massachusetts based on surveys conducted by MassDMF.
The polygons with light blue shading indicates the delineated areas that are believed to be suitable for bay scallops
(map derived from Commonwealth of Massachusetts’ Office of Geographic Information database). The 10-m isobath is
displayed for reference (blue line).
36
Figure 3.2: Spawning zones defined based on bay scallop suitability areas divided by designated shellfish growing areas.
37
Figure 3.3: Composite map of current eelgrass distribution (2001-2010) (from Buzzards Bay National Estuary Program).
38
5 km
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
#
NAME
1
WESTPORT RIVER
2
GOOSEBERRY NECK
3
SLOCUMS
4
APPONAGANSETT HBR
5
CLARKS COVE DARTMOUTH
6
CLARKS COVE NEW BEDFORD
7
NEW BEDFORD HBR
8
WEST ISLAND SOUTH
9
NASKETUCKET BAY
10
WEST ISLAND EAST
11
TAYLOR SHELLFISH PEN
12
BRANDT ISLAND COVE
13
MATT HBR SOUTH
14
MATT HBR
15
MATT HBR EAST
16
SIPPICAN HBR
17
NORTH BUZZARDS BAY
18
NORTH MARION
19
WAREHAM
20
CANAL BAYS
21
PHINNEYS HBR
22
RED BROOK HBRS
23
MEGANSETT HBR
24
WEST FALMOUTH HBR
25
QUISSET HBR
26 ELIZABETH ISLANDS EAST
27 ELIZABETH ISLANDS WEST
Figure 3.4: Settlement zones defined based on extents of water with depth less than 3.5m MSL.
39
hydrodynamic forcing on the dispersal of individuals. The time spans of early (25 June
6 July) and late (25 July 6 August) spawning events are both 10 days. Dates of
these spawning events were chosen to represent typical reproductive events of the summer
spawning season of the bay scallop in Massachusetts (Taylor and Capuzzo, 1983; Bricelj
et al., 1987).
Realistic hydrodynamic hindcasts were conducted from 1 June 1 September for the
years 2008–2010. The seasonal timeframe was selected to comprise the typical time for
bay scallop spawning mentioned in the previous paragraph.
Regional-scale hindcasts with the NeCOFS model were first conducted. These
simulations were carried out in three stages. In the first stage, the regional model was forced
barotropically with constant temperature and salinity using only the tidal elevation forcing
from five major constituents (M
2
, S
2
, N
2
, O
1
, and K
1
) at the open boundary from 1 May
to 15 May. In the second stage, temperature and salinity fields derived from hindcasts of
NeCOFS were added and the regional model integration is continued with inclusion of the
evolving density field to 31 May. In the final stage, the regional model was integrated with
all forcing from 1 June to 1 August, including temperature and salinity profiles along the
open boundary, wind forcing and surface heat flux from hindcasts of the Weather Research
and Forecasting Model (WRF) using the setup described by Chen et al. (2005) , tidal forcing
from five major constituents, discharge data from 43 rivers in the domain and assimilation
40
of daily-averaged sea surface temperature. The model time step for all three stages was
120 s and total wallclock time was approximately five hours using 8 Intel Nehalem cores.
The local SEMASS FVCOM was spun-up with a similar three-stage approach using a
model time step of 5s. The local model was first forced barotropically using only the tidal
elevation at the open boundary from 1 May to 15 May. Temperature and salinity fields
derived from hindcasts of NeCOFS were then added and the model integration is continued
to 31 May. To hindcast the period from 1 June to 1 August, the local model was then
integrated using nesting from the regional model hindcast on the open boundary, surface
heat flux from WRF, and wind forcing from the Buzzards Bay Buoy (BUZM3, Fig. 1.6)
obtained from the National Data Buoy Center. Output from the final hindcast stage was
archived at hourly intervals and used to drive the offline IBM model. The total wallclock
time for all three stages of the local model integration is approximately six days using 96
Intel Nehalem cores.
Statistics of the wind forcing during spawning and throughout the larval duration for six
realistic cases were calculated using the MATLAB toolbox CircStat (Berens, 2009). For
all six realistic cases, both the prevailing wind direction and the mean direction of the wind
fall in the third quadrant (between 180
and 270
), with levels of standard deviation between
50
–65
(Table. 3.2 & Fig. 3.5). Winds for all six periods are thus southwest dominant.
41
Table 3.1: IBM model runs for connectivity experiments. R indicates realistic and I
indicates idealized.
Cases Description
R1 2008 early spawning
R2 2008 late spawning
R3 2009 early spawning
R4 2009 late spawning
R5 2010 early spawning
R6 2010 late spawning
I1 M
2
only, no wind
I2 M
2
+constant SW wind
I3 M
2
+idealized SW sea breeze
I4 Winter wind forcing
Table 3.2: Angular wind statistics for realistic cases. Prevailing wind direction is defined
as the direction with the highest percent of frequency and is determined from the wind
roses (Fig. 3.5). Wind directions are those from which the winds originate.
Case Prevailing
wind direction
(deg N clock-
wise/cardinal
direction)
Mean of wind
direction (deg
N clock-
wise/cardinal
direction)
Angular
standard
deviation (deg)
R1 (2008 early) 220/SW 207.8/SSW 50.4
R2 (2008 late) 230/SW 219.2/SW 60.7
R3 (2009 early) 250/WSW 229.8/SW 64.7
R4 (2009 late) 250/WSW 237.4/WSW 52.2
R5 (2010 early) 230/SW 219.5/SW 57.6
R6 (2010 late) 220/SW 192.4/SSW 64.8
42
5%
10%
15%
2008 Early
WEST EAST
SOUTH
NORTH
2%
4%
6%
8%
2008 Late
WEST EAST
SOUTH
NORTH
5%
10%
15%
2009 Early
WEST EAST
SOUTH
NORTH
5%
10%
15%
2009 Late
WEST EAST
SOUTH
NORTH
0 − 2
2 − 4
4 − 6
6 − 8
8 − 10
10 − 12
12 − 14
14 − 16
16 − 18
18 − 20
5%
10%
15%
2010 Early
WEST EAST
SOUTH
NORTH
5%
10%
15%
2010 Late
WEST EAST
SOUTH
NORTH
Wind Speed
(m/s)
Figure 3.5: Wind roses for early (25 June – 6 July) and late (25 July – 6 August) spawning
simulations for the years 2008–2010. The directions in the wind rose are those from which
the winds originate.
43
Idealized Cases For these cases, idealized forcing such as constant wind was used to
drive the model. By isolating the forcing, these cases help to better understand the nature
of the system response. Comparison of these cases allows us to examine the relative
contributions of wind and tide to the connectivity pattern, as well as how the connectivity
pattern is influenced by the characteristics of the wind forcing.
Case I1 was setup using a barotropic fluid and is forced only by tides using a single
constituent (M
2
). For Case I2, the setup was the same as Case I1 with additional forcing
due to a constant 10 m/s (20 knot) southwest wind. Case I3 was designed to examine the
response to typical summer sea breeze events and is forced by the M
2
tide in combination
with an idealized time-dependent wind representing the sea breeze in Buzzards Bay. This
idealized wind was constructed as a daily repeated SW wind having a time-dependent
magnitude described by the Gaussian function
V (t) =
X
n
V
max
· e
(tt
p
)
2
2·σ
2
, t
p
= 16 + 24n, n = 0, 1, 2, 3, ... (3.1)
where the standard deviation σ =3 h (Fig. 3.6). The peak magnitude V
max
= 12 m/s
occurs daily at 4:00 pm (t
p
=16 h) local time. The purpose of this idealized wind is to
mimic the summer sea breeze events in the Bay, so the parameters of standard deviation
as well as the peak magnitude and time are determined using daily-averaged wind data
from BUZM3 which clearly displays the dominance of this periodic signal (Fig. 3.7). Case
I4 was designed to examine the response of dispersal and connectivity to an alternate wind
44
forcing which was dominantly cross-Bay. For this idealized case, we forced the model with
realistic winds from the period 20 December 2009 to 19 January 2010 (Fig. 3.8) which is
characteristic of this season with dominant NW direction. Although the spawning occurs
only in summer, the purpose of this idealized case is to assess the sensitivity of the response
to wind forcing with different characteristics.
Spawning of the Model Individuals In all of the experiments including realistic and
idealized cases, individuals were released uniformly within the spawning zones. On
average, one individual was released in every 0.4 km
2
of the spawning zones. Temporally,
the individuals were released during the 10-day time span of the spawning event following
a Gaussian distribution, with a frequency of 6 hours (Fig. 3.9). This scheme was based
on the assumption that spawning is distributed over a given period and that peak spawning
occurs towards the middle of this period. Vertically, individuals were released at the bottom
of the water column.
A summary of the biological assumptions made in this IBM study is provided in
Table3.3.
3.3 CONNECTIVITY EXPERIMENTS
To construct the total connectivity matrix for each case, we need to account for the non-
uniform distribution of settlement during the settlement competency period. The pelagic
larval duration (PLD) for the bay scallop is typically 14 days but can range from 10 to
45
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
0
5
10
15
time (hrs)
wind mag (m/s)
Figure 3.6: Time series of the idealized periodic variable wind used in Case I3 with in one
day. Wind direction is SW only. Horizontal axis is hour of the day. Vertical axis is wind
speed in m/s.
Table 3.3: Summary of the biological assumptions employed in the IBM study
Swimming No larv