o
Thi
M a r i N e P o P u l at i o N C o N N e C t i V i t y
r c
s a
o
ll
r
e
t
c
i
c
t
l
i
e h
v
e r
a
e
s b
di
s
e
t
e
ir
n p
b
u
ub
t
i
o
li
n o
s
h
e
f a
d in
Coupled Biological
n
y p
O
o
c
r
ea
t
i
n
o
o
n o
g
r
a
f t
p
h
h
i
y
s a
, V

r
o
and Physical Models ti lumcle be 20, Ny photumbocopey mr 3, a qachu
Present Capabilities and Necessary Developments
in
a
r
e
t
, r
e
r
e
l
p
y j
o
o
s
t
ur
for Future Studies of Population Connectivity
in
n
g
a
, o
l o
r o
f The
t
h
e
r m
o
c
e
e
a
a
n
n
o
s i
g
s p
r
a
p
e
h
r
m
y S
i
t
o
B y F r a N C i S C o e . W e r N e r ,
t
c
e
i
d o
e
t
y
nl
. C
r o B e r t K . C o W e N , a N D
y w
o
p
i
y
t
r
h t
i
g

C l a i r e B . Pa r i S
h
h
t 2007 b
e a
p
p
r
o
v
y The
a
The combination of wide-ranging spatial and temporal
l o
f The
o
scales associated with the oceanic environment, together
c
e
a
o
n
c
o
e
with processes intrinsic to the biology of marine organ-
g
a
r
n
a
o
p
g
h
r
y S
isms, makes the quantitative study of population con-
a
p
h
o
y S
c
i
e
nectivity a formidable challenge. Sampling over all
o
t
y
c
.
i
a
e
t
ll r
scales, except for targeted field efforts that focus on
y
. S
i
g
e
h
n
t
d a
s r
selected processes of life stages in limited domains,
e
ll c
s
e
o
r
is presently not possible (Gawarkiewicz et al., this
v
r
e
r
e
d. P
s
p
issue). As such, modeling approaches that simultane-
o
e
n
r
m
d
e
i
s
n
si
ously include key physical dynamics and biological traits
c
o
e t
n i
o: in
s g
provide a way forward to investigate general ecological
r
f
a
o@t
n
t
e
questions as well as provide qualitative assessments regard-
d t
o
s
.o
o c
r
ing connectivity of specific regions and populations (Cowen
g o
o
p
r Th e
y t
h
et al., 2000, 2006; Werner et al., 2001a). In some instances,
i
s a
o
r
c
t
i
model results have provided information of relevance to deci-
e
c
a
l
e f
n
o
o
g
r u
sion-makers in determining marine protected areas and other
r
a
p
s
h
e in t
y S
management strategies (Fogarty and Botsford, this issue). At a
o
e
c
a
i
e
c
t
h
minimum, models can be used to generate hypotheses for empiri-
y
in
, P
g a
o
B
n
cal studies. Overall, coupled biological-physical models are criti-
d r
o
x 1931,
e
s
e
cal tools for addressing the complex processes driving population
a
r
c
h
r
.
o
connectivity in marine systems.
r
c
e
k
p
v
ub
ill
e
li
, MD 20849-1931,
c
a
t
i
o
n
, s
y
s
t
e
mm
a
u
t
i
c r
S
a
e
.
p
r
o
du
54
Oceanogr
O
ap
ceanogr hy
ap
c
Vo
V l.
l. 20, No. 3
o
t
i
o
n
,

---

Marine population connectivity via
provide a mechanistic understanding of
may be relatively short (on the order of
larval dispersal is inherently a coupled
linkages. In this paper, we review some
days and kilometers). Second, the rela-
bio-physical problem. Among the rel-
of the present capabilities in models
tive contributions of key processes will
evant physical processes on continen-
used to study population connectivity,
likely be site specific and depend on
tal shelves and nearshore regions are
identify key challenges for the next-
coastal geometry, proximity to estuar-
wind- and buoyancy-driven currents,
generation models, and consider areas of
ies or deep-ocean boundary currents,
fronts and associated jets, tides (includ-
application in light of new technologies
seasonal stratification, and wind forc-
ing residual currents, internal tides and
and management needs.
ing (e.g., Sponaugle et al., 2002). Third,
bores), and surface and bottom bound-
individual physical processes contain
ary layers (Werner et al., 1997; Scotti and
What Do We KNoW aND
variable length and time scales. Thus,
Pineda, 2007); additional wave dynam-
What are our CaPaBilitieS?
hydrodynamic transport, and disper-
ics, including Stokes drift and radia-
Modeling physical and biological sys-
sion modulated further by biological
tion stress, become important within
tems has rapidly progressed in terms
properties, is fundamentally a multi-
shallow and nearshore reef structures
of the spatial and temporal resolution
scale process. It is recognized that highly
(Monismith, 2007). In turn, these pro-
achieved and the complexity of pro-
resolved flow fields are needed in order
cesses are affected via onshore/offshore
cesses involved (Werner et al., 2001b;
to embed behavioral models within
forcing by eddies, large-scale current
Hofmann and Friedrichs, 2002; Kinlan
hydrodynamic models to examine pro-
meanders, island wakes, and lateral
et al., 2005). Computational capabilities
cesses involving biophysical interactions
intrusions (Caldeira et al., 2005). In the
and the development of novel algorithms
(North et al., in press b). While presently
open ocean and at basin scales, western
have enabled the implementation of
it is difficult to simultaneously resolve
and eastern boundary currents, together
sophisticated ocean circulation models
mesoscale and small-to-intermediate
with large-scale gyre circulation, provide
(Chassignet et al., 2006) and inclusion
scale flow fields, particularly for simula-
oceanographic connectivity between
of detailed biological and geochemical
tions over long time periods, promising
locations separated over hundreds to
processes within them (Rothstein et al.,
results have been obtained through the
thousands of kilometers (e.g., Cowen,
2006a). Methods developed for nesting
use of nested and unstructured grids
1985; Hare and Cowen, 1996; Tang et al.,
models and their extensions into vari-
(e.g., Fringer et al., 2006).
2006; Kettle and Haines, 2006). However,
ous shallow-water environments provide
Using the spatial scales of popula-
physical processes alone do not deter-
opportunities for resolving more realistic
tions' dispersal and connectivity as a
mine the scales of population connectiv-
larval transport pathways (Hermann et
framework for our discussion, we con-
ity. Time scales of larval development
al., 2002). Additionally, developments
sider modeling capabilities at basin
and behavioral characteristics, including
in spatially explicit Individual Based
scales (of order 1,000 km), shelf scales
vertical migration and spatially explicit
Modeling (IBM) approaches allow for
(of order 100 km), and reef scales (of
environmental differences, play impor-
the inclusion of detailed parameteriza-
order 10 km). In each case, we pres-
tant roles (Boehlert and Mundy, 1988;
tions of biological variables required for
ent current capabilities for capturing
Tremblay et al., 1994; Hare et al., 1999;
quantitative estimates of larval dispersal
the relevant processes and illustrate
Bode et al., 2006; North et al., in press a).
(Werner et al., 2001a; Siegel et al., 2003).
with specific examples.
Modeling must work hand in hand
General points can be extracted from the
with field and laboratory studies to test
large number of physical and biologi-
Basin Scales
model predictions and assumptions, bet-
cal processes that relate to the dispersal
Representation of large-scale oce-
ter parameterize and initialize the mod-
and recruitment of marine organisms,
anic response to atmospheric forcing
els, and iteratively strengthen the model
to help define the connectivity prob-
is at a point where relatively success-
capabilities. Yet, the value of coupled
lem. First, the temporal and spatial cor-
ful hindcasts of basin-scale El Niño-
biophysical models is unique as they
relation scales over continental shelves
Southern Oscillation/North Atlantic
Oceanography September 2007
55

---

Days720 trajectories of particles/model organisms
given hydrodynamics and behaviors)
60°N
630
for the transport and migration of the
leptocephali (larvae) of the European
540
eel (Anguilla anguilla) across the North
Atlantic Ocean from the spawning area
450
in the Sargasso Sea to the adult range off
40°N
Europe and North Africa. The success of
360
larvae in reaching particular locations
270
on the eastern side of the North Atlantic
was found to depend on starting loca-
20°N
180
tion in the Sargasso Sea, time of year of
the spawning, and the depth in the water
90
column at which the larvae were trans-
ported. Model results found the fast-
0°
0
est cross-basin larval migration to take
100°W
80°W
60°W
40°W
20°W
0°
about two years, with the route from the
Figure 1. trajectories of eel larvae successfully crossing 25°W within two years of release within
the spawning polygon. Black "+" symbols show locations of larval captures. trajectory colors indi-
Sargasso Sea to Europe taking many of
cate age after release in the Sargasso Sea polygon. From Kettle and Haines (2006) reproduced with
the larvae past the North American East
permission from the Canadian National Research Council Press
Coast during the first year (Figure 1).
The model results are consistent with
the hypothesis that the European eel
and the American eel (Anguilla rostrata)
Oscillation/Pacific Decadal Oscillation
ics. Similarly, useful data-assimilation
could separate themselves on differ-
and related climate phenomena have
and nowcast/forecast systems are pres-
ent sides of the North Atlantic basin on
been achieved, attesting to the quality of
ently being developed, and applications
the basis of the different durations of
the data collected and the models' abil-
from various communities worldwide
their larval stages.
ity to capture relevant internal dynam-
are resulting in successful pilot efforts
Increasing the biological informa-
(e.g., GODAE, HYCOM, Mercator) that
tion included in models can further the
FraNCiSCo ("CiSCo") e. WerNer
now routinely provide hindcasts, fore-
resolution of the drivers of population
(cisco@unc.edu) is George and Alice Welsh
casts, and nowcasts of global ocean state
connectivity. For example, the additional
Professor, Department of Marine Sciences,
estimates (see Lermusiaux et al., 2006).
effects of mortality and settlement sub-
University of North Carolina, Chapel Hil ,
The realism achieved by the physical
strate on possible dispersal outcomes
NC, USA. roBert K. CoWeN is Maytag
models has resulted in several success-
show that physical dispersal alone can
Professor of Ichthyology and Chair, Division
ful biological/ecological studies, such as
yield overestimates of population dis-
of Marine Biology and Fisheries, Rosenstiel
those on the distribution of the copepod
tribution (e.g., see Cowen et al., 2000,
School of Marine and Atmospheric Science,
Calanus finmarchicus in the subpolar
2006; Paris et al., 2005, for case studies
University of Miami, Miami, FL, USA.
North Atlantic by Speirs et al. (2006)
in the Gulf of Mexico and the Caribbean
Claire B. PariS is Research Assistant
and on the dispersal of American and
Basin). A clear illustration of this effect
Scientist, Division of Marine Biology and
European eels in the North Atlantic by
is provided in Figure 2, showing that
Fisheries, Rosenstiel School of Marine and
Kettle and Haines (2006).
while current trajectories might indicate
Atmospheric Science, University of Miami,
The latter study presents a Lagrangian
the potential outcome of larval trans-
Miami, FL, USA.
model (i.e., a model that computes the
port (Figure 2A), they fail to account for
56
Oceanography Vol. 20, No. 3

---

the true probability of successful down-
A
stream transport because larval concen-
20
trations are reduced by several orders
of magnitude by along-trajectory diffu-
sion and mortality of larvae (Figure 2B).
Within 30 days, model larvae released
15
Latitude
from a 1-km2 location near Barbados
spread over 106 km2, representing a
reduction of the original concentration
of larvae by six orders of magnitude
10
-75
-70
-65
-60
-55
(Figure 2A). If mortality is included,
Longitude
there are not enough larvae occurring
within any coastal region to sustain
B
downstream populations from a source
20
population, even when all larvae pro-
duced at the source leave the source area.
Shelf Scales
15
Latitude
Over the last two decades, the advent
and establishment of sophisticated
and realistic coastal circulation models
(e.g., Haidvogel and Beckmann, 1999),
10
-75
-70
-65
-60
-55
including unstructured and nested
Longitude
grids (Lynch et al., 1996; Greenberg
Figure 2. Simulation of the role of dispersal and natural larval mortality on the prob-
et al., 2007), nonhydrostatic models
ability of successful recruitment. (a) Model results of a 30-day dispersal period
(Fringer et al., 2006; Scotti and Pineda,
initiated at the island of Barbados (*). larvae are spread over a 106 km2 area result-
ing in a six-order-of-magnitude dilution of the initial concentration of larvae
2007), and large-eddy simulations (LES;
(i.e., 10,000 larvae km2). (B) Same simulation with the added effect of natural mortality
Lewis, 2005), have enabled the quan-
during the 30-day larval phase (18% d-1). Mortality further reduces the concentration
titative study of key physical processes
of larvae by another three orders of magnitude. about 10% of these larvae (blue dots)
find suitable habitat. Modified from Cowen et al. (2000) with permission from the
in varying degrees of approximation.
American Association for the Advancement of Science.
Similar to the developments in basin-
scale modeling, public-domain "com-
munity" models, such as the Regional
Ocean Model System (ROMS) and the
sustained analyses of certain processes in
fields. IBMs keep track of individuals
Princeton Ocean Model (POM), render
limited area domains (see Robinson and
within a population, and have become a
well-established approaches/protocols to
Lermusiaux, 2002).
de facto modeling approach in efforts to
be considered and tailored (with relative
Taking advantage of the wide range of
study the interactions of marine organ-
ease) to site-specific applications with
robust and advanced circulation models,
isms with their environments and to
known attributes and limitations. Useful
spatially explicit IBMs have been used
understand factors impacting dispersal
data-assimilation and nowcast/forecast
to determine trajectories, or Lagrangian
and population connectivity (Werner et
systems are presently being developed
pathways, of planktonic stages of marine
al., 2001a). The simplest of these stud-
and applied with some having reached
organisms in realistic (i.e., spatially het-
ies ignores biotic factors such as feeding
quasi-operational status, allowing for
erogeneous and time-dependent) flow
and predation, but includes imposed
Oceanography September 2007
57

---

spawning times and locations, swim-
individuals encounter favorable feed-
of the organisms' feeding environments
ming behaviors, and larval-competency
ing environments. Some of these stud-
explicitly include specific prey types
periods. Among the questions success-
ies have also been used to explore other
rather than general functional groups.
fully investigated by such studies are
spatially dependent interactions between
Is it clear from the discussion above
the space-time pathways of larval fish
predators and their prey. For example,
that connectivity requires quantitative
from spawning grounds to nursery areas
the perception of prey by fish larvae
understanding of physical and biological
(Miller et al., 2006), larval retention on
can be effectively increased or reduced
processes integrated in space and time.
submarine banks and islands (Page et al.,
as a consequence of local variation in
One example of an area that illustrates
1999; Paris and Cowen, 2004), effects of
turbulence levels, which alters the vol-
the combined effects of variability in
interannual variability of physical forc-
ume searched (Dower et al., 1997). This
the hydrodynamic and feeding environ-
ing on dispersal of larval fish popula-
requires models to capture not only the
ments is the shelf region off South Africa
tions (Rice et al., 1999), identification
spatial distribution of biotic components
in the Benguela Current system. Using
of spawning locations (Stegmann et al.,
but also their modulation by certain abi-
model-based approaches, Mullon et
1999), dispersal barrier mechanisms
otic environmental factors.
al. (2003) consider evolutionary-based
(Baums et al., 2006), and long-term
Intersections of large- and small-
reproductive strategies and processes
dispersal by tidal currents (Hill, 1994).
scale physics affecting recruitment are
affecting survival and recruitment of
Similar approaches focusing on inverte-
discussed by Werner et al. (2001a), who
pelagic fishes (sardines and anchovies)
brates include the seeding of scallop beds
consider modifications of the perceived
in the context of "Bakun's triad" (Bakun,
on Georges Bank (Tremblay et al.,1994;
feeding environment by turbulence at
1996)--enrichment, enhancement, and
Fogarty and Botsford, this issue), the
the smallest scales, including its effect on
retention (Lett et al., 2006). Of immedi-
behaviorally mediated connectivity of
particle trajectories and on larval growth
ate relevance to population connectivity
copepods between shelves and deep
and survival. The authors found that
is the ability to use models to determine
basins (Speirs et al., 2005) and their on-
regions of enhanced larval growth and
the selection of the source/spawning
shelf retention (Batchelder et al., 2002),
survival--resulting from the enhance-
regions by the populations, and the
the impact of environmental quality and
ment of contact rates and effective prey
subsequent spawning-to-nursery area
larval supply on recruitment of spiny
concentrations by turbulence within the
transport of the fish larvae and the biotic
lobsters in the Florida Keys (Butler,
tidal bottom boundary layer--coincided
processes that determine along-transit
2003), and the onshore transport of bar-
with hydrodynamically retentive subsur-
survival success (see Figure 3).
nacles (Pineda et al., this issue).
face regions of Georges Bank defined in
The inclusion of the effect of feed-
earlier studies (e.g., Werner et al., 1993).
reef Scales
ing environment on successful dispersal
While the above studies have focused
For smaller scales encountered at reefs,
and recruitment has been achieved by
on higher-latitude environments, it is
from 1­10 km, the details of the inter-
using temperature as a proxy for feeding
a greater challenge to model trophic
actions among topography, wind, tid-
environment or through more explicit,
interactions in subtropical and tropical
ally driven currents, and wave motions
but still idealized, representations of
regions that are typically characterized
become increasingly important (Legrand
spatially dependent (but temporally
by high species diversity and oligotro-
et al., 2006; Monismith 2007; Tamura et
fixed) prey fields (see review in Werner
phic waters. In effect, billfish, scombrids,
al., 2007). The flow near shallow coral
et al., 2001a). Using these approaches,
and many coral reef fish larvae appear to
reefs is particularly complex and encom-
Lagrangian trajectories considered
have evolved high prey selectivity, per-
passes a multiplicity of scales: at the larg-
favorable for dispersal and retention
haps to minimize competition (Llopiz
est scales, flows include eddies produced
or appropriate for transport into nurs-
and Cowen, in press, and recent work of
by island wakes, while at the smallest
ery areas are more narrowly defined to
author Cowen and Joel Llopiz, RSMAS).
scales of flow, we need to consider flows
include only those trajectories where the
Such dietary niches require that models
around single coral colonies. Circulation
58
Oceanography Vol. 20, No. 3

---

Figure 3. (a) Schematic of anchovy egg and larval dynamics in the south-
a
ern Benguela Current. The eggs and larvae are transported from west-
ern agulhas Bank spawning sites to nursery areas off the west coast of
africa. (B) Modeled particle tracks (white dots) in the Benguela Current
system of transport over a six-week period (courtesy of Christian Mullon);
the land mass is in red and the depth contours are the remaining shaded
areas. (C) enrichment intensity obtained through simulated particle
upwelling; regions 1­4 are the most enriched off the west coast (com-
pare to the schematic in 3a). (D) Map of simulated pattern of retention.
Values correspond to the proportion of particles retained averaged
over the period 1992­1999 and depth. Note that the retention areas are
located near the enrichment areas. recruitment for both anchovy and
sardine is considered to occur predominantly off the west coast; the high
retention predicted for region 3 may not result in observed successful
recruitment. Figures adapted from Mullon et al. (2002) with permission
from the Canadian National Research Council Press and Lett et al. (2006)
with permission from Blackwell Publishing
B
C
D
models have proven to be effective in
bathymetric mapping of both inner and
Mapping; Andrefouët et al., 2005). The
describing hydrodynamic features and
outer reef areas is essential (Wolanski et
coupling of geographic information
biological or material transport around
al., 2004; Legrand et al., 2006).
system (GIS) and Lagrangian models in
atolls, barriers, and fringing coral reefs
Satellite-derived mapping and spatial
highly fragmented habitats, such as coral
(Kraines et al., 2004). However, to prop-
analyses of the global coral reefs are par-
reefs, permits analyses of metapopula-
erly and quantitatively capture these
ticularly useful for generating the sea-
tion dynamics (i.e., patterns of indi-
features, high-resolution and accurate
scape layer (e.g., Coral Reef Millennium
vidual movement between geographi-
Oceanography September 2007
59

---

cally separated populations) by tracing
(e.g., Bode et al., 2006).
reefs (3­23 km distant). Integration of
the size and arrangement of population
It is also important to integrate the
both sensing and orientation abilities of
patches in transition matrices. Cowen et
topography and habitat attributes along
individuals and attributes of the reef into
al. (2006) used the transition matrix out-
Lagrangian pathways, as they have been
coupled biophysical models should gen-
put to map larval flow (i.e., using graph
shown to affect individual behaviors and
erate better quantitative assessment of
theory, sensu Urban and Keitt, 2001)
movement through time (Paris et al.,
the influence of behavior on connectivity
between 50-km segments of coral reefs
2005). For example, at local reef scales,
at small scales.
in the Caribbean. At the regional scale,
Gerlach et al. (2007) used a multidis-
Finally, a series of Great Barrier Reef
their results matched the realized dis-
ciplinary approach (i.e., hydrographic
(GBR) studies by Wolanski and Spagnol
persal derived from genetic and biogeo-
modeling, population genetics, sensory/
(2000), Wolanski et al. (2003b), and
graphical studies (Taylor and Hellberg,
behavioral experiments) to demonstrate
Legrand et al. (2006) illustrates pres-
2003; Baums et al., 2006; Purcell et al.,
that larvae utilized olfaction to enhance
ent hydrodynamic modeling capabili-
2006). Similar modeling approaches have
return to their natal reef, greatly modify-
ties integrating across scales, from the
been utilized for the Great Barrier Reef
ing passive dispersal between adjacent
shelf down to reefs (see Figures 4 and 5).
Extending over 2500 km along Australia's
northeastern continental shelf, the GBR
comprises almost 3000 individual reefs
ranging in area from 0.01­100 km2.
Processes resulting from the wind- and
tidally driven flow range from meters to
hundreds of kilometers, and from min-
utes to years (Wolanski et al., 2003a),
100 km
requiring that models resolve processes as
small as 10­100 m while at the same time
including the larger-scale background
dynamics. When such models are merged
across a range of spatial and temporal
scales (Wolanski et al., 2003a), two-way
nesting may be needed to allow feed-
back of processes between each grid size
through parameterization and establish-
ment of boundary conditions (e.g., Sheng
et al., 2005; Greenberg et al., 2007).
What MoDeliNg
aDVaNCeS are NeeDeD to
CoNtiNue to iMProVe our
QuaNtitatiVe DeSCriPtioN
oF CoNNeCtiVity?
At each of the scales considered in the
previous sections, it is clear that quan-
Figure 4. Computational finite element mesh of the great Barrier reef (Ne coast of australia) with
titative modeling of population con-
zoomed in views illustrating the range of element sizes. Colors indicate local water-column depth.
Redrawn from Legrand et al. (2006) courtesy of Eric Deleersnijder and Jonathan Lambrechts
nectivity poses unique challenges in
60
Oceanography Vol. 20, No. 3

---

a
terms of the physics and biology needed
to capture the underlying dynamics.
Physical and biological processes occur
at multiple scales and they generally
overlap and interact. For instance, many
source regions are located in the near-
shore environment, where there remain
fundamental issues in resolving near-
shore (e.g., wave-dominated) physics
and its coupling to inner shelf dynam-
ics. In turn, inner-shelf circulation can
be affected by processes occurring along
the outer edge of the continental shelf,
where shelf and oceanic dynamics inter-
act and are often influenced by strong
boundary currents in the presence of
increased levels of mixing and internal
wave fields. At the same time, a criti-
cal aspect of the modeling necessary for
understanding population connectiv-
C
ity is the incorporation of behavior and
other biological processes into models.
B
In the following sections, we briefly dis-
cuss elements of our modeling capabili-
ties that need to be improved.
Physics and hydrodynamics
Advances are needed to better represent
models' internal physical dynamics, par-
ticularly at intermediate/submesoscale
to small scales, such as frontal dynamics
(Gawarkiewicz et al., this issue), wave-
Figure 5. (a) Sample velocity fields in the vicinity of reefs and islands in the great Barrier reef
induced and boundary-layer processes
(courtesy of Jonathan Lambrechts and Eric Deleersnijder). (B) Coral colony (from Monismith, 2007,
reprinted with permission from the annual review of Fluid Mechanics, Volume 39 ©2007 by Annual
(Monismith, 2007), and other nonhy-
Reviews, www.annualreviews.org) (C) laboratory measurements of the streamwise (arrows) and
drostatic flows (e.g., Scotti and Pineda,
vertical flow (color) within the colony measured by magnetic resonance imagery, yet to be modeled
2007). For example, aperiodic, sub-
and merged within larger-scale domains (image from Chang, 2007; courtesy of Sandy Chang).
mesoscale eddies that develop along the
interface between the Florida Current
and the steep Florida shelf edge have
been shown to significantly influence
et al., 2005; D'Alessandro et al., 2007).
There is also a need for better specifi-
(positively and negatively) the delivery of
Additional modeling efforts are to realis-
cation of external forcing surface fluxes
certain coral reef fish larvae to settlement
tically capture these eddies' features (e.g.,
(which continues to challenge all circula-
habitat along the reef track (Sponaugle
Fiechter and Mooers, 2003).
tion models), especially at event scales.
Oceanography September 2007
61

---

For instance, strong winds and the pas-
simple diffusivity. It is well known, how-
retention, fronts for accumulation, and
sage of hurricanes have been correlated
ever, that where the hydrodynamics are
internal bores for two-way transport.
with large recruitment pulses for some
especially complex, such as in the vicin-
The coupling to far-field models and
species of fish and lobster (Shenker et
ity of reefs and within coastal bound-
the two-way coupling across scales con-
al., 1993; Eggleston et al., 1998). More
ary layers in general, particle-tracking
tinue to be a focus of research for several
generally, extreme meteorological events
algorithms need to be adapted to aniso-
groups (Hermann et al., 2002; Sheng
are common for the Caribbean region,
tropic vertical and horizontal diffusivi-
et al., 2005; Chassignet et al., 2006). As
and properly capturing such events is
ties to avoid spurious accumulation of
individuals from a population leave a
important because their spatial environ-
particles (Werner et al., 2001b; Spagnol
local domain, their trajectories need to
mental autocorrelation can have impor-
et al., 2002). Properly accounting for
be determined within a larger domain
tant implications for the source-sink
effects such as "form drag" near the bot-
that encompasses downstream local
dynamics of metapopulations (Schiegg,
tom of or around reefs (e.g., Wolanski,
populations and the metapopulation
2003). As a corollary, a potential increase
1987; Monismith, 2007) and the effect
network. Ideally, a series of two-way
in storm frequency during the warm-
of wind-driven shear stress (or Stokes
nesting grids from basin scale, to shelf
ing of the tropical oceans (Trenbreth,
drift) and gravity waves near the free
scale, and further to local scale should be
2005) could affect the patterns of
surface can have a significant effect on
employed, with higher resolution both
population connectivity.
estimating fluxes across these boundar-
for spawning and recruitment areas.
Because trajectory accuracy is impor-
ies. Future key observations afforded by
This will be particularly important as we
tant for connectivity models, capturing
observing systems and observatories will
consider longer time scales (e.g., climate
the effects of sub-grid-scale represen-
significantly assist in the improvement
scenarios) where biogeographic shifts
tations without the introduction of
of model treatments of the processes at
in the distributions of populations and
artifacts is essential. Yet, sub-grid-scale
sub-grid scales.
their connectivity are likely to be affected
parameterizations (i.e., physical mixing
The interaction of stratified flows with
(Vikebø et al., 2007).
processes) are not well understood and
topography remains unresolved, particu-
thus not well modeled. Mixing, sub- and
larly where topographic variations are
Biology and Behavior
super-diffusive features, and critical
abrupt (e.g., at the shelf break, promon-
A fundamental difference between
shear stress are examples of sub-grid-
tories, oceanic islands), leading to flow
recruitment and connectivity models
scale processes in need of quantitative
regimes with attached or shed eddies
is the focus on temporal and spatial
revisiting. Recent Lagrangian work by
and large-amplitude internal waves, and
scales. In recruitment studies, emphasis
Veneziani et al. (2005) reveals that the
where consequent mixing may co-occur
is on the temporal patterns (i.e., when)
eddy field is characterized by two distinct
(Boyer and Tao, 1987; Pineda, 1994).
and the quantitative aspects (i.e., how
regimes, a background flow associated
Similarly, uncertainties remain concern-
much) of successful dispersal. It is there-
with nonlooping trajectories, and coher-
ing the actual topographic data for a
fore important to identify the physical-
ent vortices generating looping trajecto-
variety of reasons, including inadequate
biological interactions that drive high
ries. The latter flow regime is prominent
sampling and changes of bottom fea-
recruitment versus low recruitment;
at the shelf break and around oceanic
tures over time, particularly in shallow
behaviors related to feeding and growth
islands and atolls and should be param-
nearshore, estuarine, and reef regions.
are key. On the other hand, in connec-
eterized using a Lagrangian stochastic
High-resolution hydrodynamic observa-
tivity studies, the emphasis is on spatial
model (LSM) with a relative vorticity
tions are necessary for the development
patterns related to population link-
factor that correctly simulates the trap-
of models capable of operating in areas
ages. Such models need to be spatially
ping effect of particles (e.g., Paris et al.,
of steep topography, as organisms may
explicit and resolve the scales of source
in press). In contrast, the background
exploit topographic eddies and associ-
and sink populations. Therefore, initial
flow regime can be parameterized by
ated secondary circulation features for
conditions and accuracy of the trajec-
62
Oceanography Vol. 20, No. 3

---

tory become important issues in which
et al., 2007). All behavioral traits are
of the hydrodynamics may be domi-
larval behavior (i.e., swimming and
variable in essence, and thus modeling
nant during spatially and/or temporally
orientation) plays a large role. The sur-
of behavior has to be probabilistic to
restricted phases of the pelagic duration
vival consequences of individual move-
account for these variations. Offline sto-
of organisms (e.g., Koehl and Powell,
ments also emerge as a key component
chastic modeling systems using archived
1994) but should not preclude the influ-
in population connectivity (Cowen et al.,
fields from oceanographic models with
ence of behavior in general. The ques-
2000). While there is no general agree-
proper subsampling (e.g., Hermann et
tion from the modeling standpoint is not
ment at the present time on the best way
al., 2001) are now efficiently used to gen-
whether behaviors should be included,
forward, the use of agent-based (or indi-
erate likelihoods of larval exchange from
but rather how they should be included.
vidual-based) modeling is central to suc-
high-frequency releases of active par-
North et al., (in press a) provide guide-
cessful modeling. A general issue is the
ticles from numerous source locations
lines on how behavioral experiments
disparate range of time and length scales
(Cowen et al., 2006).
in the field and in the laboratory can
for physical processes (e.g., internal
Marine organisms are generally not
provide information to modeling stud-
waves, fronts, eddies, gravity currents)
passive. Among various behaviors, organ-
ies, and in turn how these behaviors
that interact with the biological time
isms swim actively, migrate vertically,
may be imposed in models in response
and length scales, such as the organisms'
and change their buoyancy (e.g., Cowen,
to, for example, external cues, foraging,
vertical position in the water column,
2002; Fuchs et al., 2007). Behavior can
avoidance of predators. The study of
larval duration, and length, timing, and
arise in response to environmental cues,
small-scale interactions of water, larvae,
location of spawning.
or it can be ontogenetically triggered
and their prey on scales of a few kilo-
Larval connectivity pathways condu-
(i.e., during the process of larval develop-
meters to meters is without doubt an
cive to self-recruitment are essential for
ment), with eventual schooling behav-
important research need.
population persistence (Hastings and
ior emerging as the organisms reach
Behavior can be implemented in IBMs
Botsford, 2006; Gaines et al., this issue).
juvenile or adult stages, which, in most
by simple rules (e.g., seeking to maxi-
Life-history traits, driven by selective
cases, signal increasing sensory abilities
mize growth or to minimize mortality
pressure of the species' natural habi-
and survival (e.g., Codling et al., 2007).
risk). However, these simplifications are
tat, may have evolved to exploit hydro-
Many studies have established the impor-
probably not realistic as other trade-offs
graphic regimes that improve the odds of
tance of vertical behaviors in popula-
need to be considered that may vary
larvae returning close to the parent pop-
tion dispersal, retention, settlement, and
with stage and age, and over genera-
ulation (Strathman et al., 2002). Physical
connectivity; larvae located at different
tions when longer time scales are con-
processes can also influence biological
depths will be subjected to different cur-
sidered. Additionally, variability in the
processes, such as the timing of repro-
rents and thus their Lagrangian trajecto-
environment can be large enough that
duction, larval transport and behavior,
ries will be different (e.g., Werner et al.,
it may override these life-history-based
and the timing of settlement (Cowen,
1993; Batchelder et al., 2002; Paris and
assumptions. Alternatives to imposing
2002). Induced behavior (e.g., diel verti-
Cowen, 2004); their trajectories are also
"simple" and likely unrealistic behaviors
cal migration, foraging, predator avoid-
influenced by their pelagic phase dura-
are to consider modifications of behav-
ance, directional horizontal swimming
tion (Tremblay et al., 1994). Similarly,
ior through adaptation (e.g., Giske et al.,
in response to environmental cues) and
directed horizontal swimming rates
2003). Dynamic programming meth-
emerging properties (i.e., larval stage
have been shown to significantly affect
ods allow organisms to "find" optimal
duration, age/stage-dependent vertical
distributions of individual organisms
habitats by balancing risks of predation,
migration, mortality) have been shown
(e.g., Werner et al., 1993). It is safe to
growth, and advective loss. Examples
to influence population connectivity
say that behaviors can be complex and
include the adapted random walk (Huse,
as much as currents do (Paris et al., in
their explicit consideration is essential in
2001), optimization of self-recruitment
press) and can reduce dispersal (Gerlach
many cases. At the same time, the effect
in isolated islands (Irisson et al., 2004),
Oceanography September 2007
63

---

Models should guide in the design of field
experiments, and in turn observations
should improve the model parameters.
and individual-based neural network
cated by recombination, selection, and
strengthen model capabilities. However,
genetic algorithm (ING) (Giske et al.,
behavioral interactions. Many aspects of
model solutions themselves need to be
1998) that allow adaptive behaviors to
classic population genetic theory involve
validated, which is not straightforward,
emerge in populations in complex envi-
assumptions that are incompatible with
given the complexities of the processes
ronments. The ING method provides a
contemporary studies (e.g., the assump-
contributing to the computed realiza-
way to implement behavior in individual-
tion of equal and even migration), are
tions. Formal procedures for skill assess-
based models. One example is provided
too simplified for practical application,
ment of physical models have been dis-
by Strand et al. (2002) wherein an indi-
or else are inscrutable to nonspecialists.
cussed (e.g., Lynch and Davies, 1995),
vidual-based model that uses artificial
Novel and flexible approaches, such as
but quantitative skill assessment of
evolution is used to predict behavior and
object-oriented frameworks (Johnathan
biological models is still relatively unde-
life-history traits on the basis of environ-
Kool, RSMAS, pers. comm., 2007), are
veloped (Arhonditsis and Brett, 2004).
mental data and organism physiology.
required. It is also important that these
Validation of trajectory paths can be
In their approach, evolutionary adapta-
systems be accessible in order to ensure
accomplished using a combination of
tion is based on a genetic algorithm that
that they can be put to use by the people
acoustic and hydrographic (e.g., ADCP,
searches for improved solutions to the
who need them. Despite the difficulties
CTD), Lagrangian (e.g., satellite-tracked
traits of habitat choice, energy allocation,
involved, the potential rewards are worth
floats, fluorescent wax particles), tag-
and spawning strategy. Behaviors emerg-
the effort, as it would provide researchers
ging or mass marking (e.g., otoliths),
ing from model studies can complement
with a means of exploring community
and plankton sampling tools (e.g., trawls,
results from field and laboratory efforts
and ecosystem-scale evolutionary ques-
optical sampling). It may be possible
and allow predictions to be attempted.
tions and validating them with empirical
to validate population-connectivity
Another developing area of research
evidence (e.g., validating connectivity
results by use of geochemical signatures
involves linking population genetic
patterns using genetic marker data).
(Becker et al., 2007; Thorrold et al., this
models with bio-oceanographic models.
issue) or genetic tools (review by Planes,
Modeling the flow of genetic material
Model Validation and
2002; Hedgecock et al., this issue).
presents a significant challenge, both in
Skill assessment
Measurements of postlarval supply at
terms of simulation and analysis because
Modeling has to work hand in hand
multiple sites can also produce a very
of its ability to persist through time. This
with empirical studies to test model
consistent validation of the connectivity
persistence can lead to complex dispersal
predictions and assumptions, better
results without providing explicit knowl-
patterns; dynamics are further compli-
parameterize the models, and iteratively
edge of the source locations.
64
Oceanography Vol. 20, No. 3

---

Future DireCtioNS aND
sive sampling) and temporal variabil-
on the present status of modeling and
aPPliCatioNS
ity over long time scales. However, the
data-assimilation systems, the develop-
As we look forward, models can be
proper deployment and design of such
ment of adequate capabilities for imple-
expected to provide test beds for devel-
field deployments is not trivial as attested
mentation of useful OSSEs in observing
oping field-testable hypotheses, means
by the wide range of processes discussed
systems for use in studies of population
for refining existing ecological and evo-
above. In turn, the effectiveness of the
connectivity may be achievable on the
lutionary theories, guidance for bet-
sampling strategies is determined by the
time scale of a few (two to five) years
ter experimental designs, and possible
accuracy with which the observations
(Rothstein et al., 2006b).
future scenarios in light of expected
can be used to reconstruct the state of the
climate/global change; ultimately, they
natural system being measured. Given
Predictions and Future Scenarios
should be useful to resource managers.
the limited opportunities for evaluation
Over the past several decades we have
We consider these points briefly in our
of sampling strategies against objective
learned about longer-term fluctuations
concluding remarks.
criteria with purely observational means,
that occur within the bounds of "natural"
numerical models offer a framework for
variability (e.g., El Niño, Pacific Decadal
Models assisting the Design
investigation of these issues (Walstad and
and North Atlantic Oscillations). The
of Field Sampling
McGillicuddy, 2000), and observation
ecological impacts of a changing climate
As physical and biological models
system simulation experiments (OSSEs)
are already evident in terrestrial and
progressively mature, they should be
become of central importance in experi-
marine ecosystems, with clear responses
expected to provide hypothesis testing
mental design of questions addressing
of both the flora and fauna, from the spe-
and help guide sampling protocols and
population connectivity.
cies to the community levels (e.g., Harley
observatory/observing system design. For
The aim of OSSEs is to model obser-
et al., 2006). The continued influence of
example, numerical simulations based
vation systems with the intent to quan-
humans on climate will result in further
on high-frequency, quasisynoptic, in situ
tify their sampling properties and opti-
changes in abundance and distribution
measurements of physics and larval dis-
mize their design (i.e., OSSEs attempt
of marine species; thus, there is a need
tributions have proven effective in testing
to simulate unknown ocean properties
to estimate the nature and severity of
larval transport hypotheses (e.g., Helbig
to better measure and discover them).
the possible ecological consequences in
and Pepin, 1998; Paris and Cowen,
OSSEs can be utilized for multiple
order to develop strategies to manage
2004). While observational capabilities
purposes including to: (1) guide the
marine resources. The physical forcing
for high-resolution and rapid measure-
design of an observation system and its
of a changing (e.g., warmer) climate is
ments of physical properties are relatively
components; (2) optimize the use of
hypothesized to have important conse-
advanced, observing systems for deter-
observational resources; (3) assess the
quences that can impact marine popu-
mining larval distributions are less so.
impact of existing or future data streams
lations, for example, changes in wind
Real-time, in situ observing systems will
(e.g., for nowcasting and forecasting of
patterns and intensities that in turn can
be useful in defining the oceanographic
requisite accuracies); (4) understand
impact the strength of ocean currents,
environment, including variability of
the interactions of system components
and increased stratification in the surface
the hydrographic and velocity fields, and
and improve system performance;
layers caused by the warming of surface
in determining the Lagrangian path-
(5) evaluate and validate system perfor-
ocean waters or increased freshwater
ways of identifiable larvae from source
mance using quantitative error estimates;
fluxes. The dispersal of populations (par-
areas. Additionally, relevant technolo-
and (6) compare data-assimilation
ticularly during the early life stages) will
gies continue to emerge (e.g., imaging
methods (GLOBEC.INT, 1994). OSSEs
be directly affected by changes in cur-
systems--see Benfield et al., 2007), and
continue to be used to assess, for exam-
rents and temperature (e.g., O'Connor et
remote observing systems will identify
ple, physical oceanographic array designs
al., 2007). One example is that described
episodic events (e.g., by triggering inten-
in the tropical and coastal oceans. Based
by Vikebø et al. (2007), who explore the
Oceanography September 2007
65

---

effects that a reduction of the thermo-
ing coupled biological-physical models
(MPAs) or the exploration of future sce-
haline circulation (THC) may have on
to climate models capable of examining
narios requires a quantitative descrip-
larval drift and development of Arcto-
climate-change scenarios will be critical
tion of population connectivity. An
Norwegian cod. Using a regional model
in assessing potential direct and indirect
understanding of spatial linkages over
forced by a global climate model, they
impacts of climate change on population
populations will also contribute to the
find a reduction in the THC relative to
connectivity and the ultimate usefulness
explanation of variability in fisheries (see
present-day circulation. The impact of
of the model to resource managers.
Fogarty and Botsford, this issue; Jones
the change in circulation and ocean tem-
et al., this issue). In these applications,
perature on the cod results in a south-
Models as Management tools
realistic descriptions of habitat, hydro-
ward and westward shift in the distribu-
The management of living marine
dynamics, larval transport pathways, and
tion of cod from the Barents Sea onto
resources is inherently spatially depen-
adult growth and survival will provide
shelf regions, a reduction in the predicted
dent. Understanding how marine popu-
a mechanistic understanding of how
individual growth of the pelagic juveniles,
lations are connected in space and time
local populations may be interconnected.
and an increase in the number of larvae
will provide an essential component to
Validated, spatially explicit models will
and pelagic juveniles that advect towards
management of marine resources that
also be useful for designing and assessing
regions where they are unable to survive
is presently not available. For instance,
MPAs in that they will provide the degree
(see Figure 6). Our ability to link exist-
the design of marine protected areas
to which populations are connected and
a
B
79
79
77
77
75
75
73
Latitude
73
Latitude
71
71
300
300
69
69
250
250
67
200
67
200
65
150
65
100
63
100
63
100
0
5
10
15
20
25
30
35
40
45
0
5
10
15
20
25
30
35
40
45
Longitude
Longitude
Figure 6. Simulated distribution of pelagic juvenile cod with (a) the ocean and atmospheric forcing for the present day, and (B) for changed cli-
mate (today + 50 years) scenario run. The color scale indicates wet weight in milligrams. These results indicate that in the +50 year run pelagic
juveniles will have lower weights. The distributions also indicate that a higher number of larvae and pelagic juveniles are advected to the west of
Spitsbergen (to the northwest) in the +50 year run. From Vikebø et al. (2007) with permission from Blackwell Publishing
66
Oceanography Vol. 20, No. 3

---

estimates of the exchange between adja-
aCKNoWleDgeMeNtS
in Transport of Larval Fish and Shellfish Through
Coastal Inlets. M.P. Weinstein, ed., American
cent and distant areas. Models will offer
The authors were supported by the
Fisheries Society Symposium 3.
insights into the mechanisms of how and
National Science Foundation (NSF),
Boyer, D., and L. Tao. 1987. On the motion of linearly
why populations exist in certain systems
National Oceanic and Atmospheric
stratified rotating fluids past capes. Journal of Fluid
Mechanics 180:429­449.
and bring increased understanding of the
Administration (NOAA), U.S. GLOBEC,
Butler, M.J., IV. 2003. Incorporating ecological process
variability in marine populations.
National Ocean Partnership Program
and environmental change into spiny lobster popu-
lation models using a spatially-explicit, individual-
(NOPP), Office of Naval Research
based approach. Fisheries Research 65:63­79.
CoNCluDiNg reMarKS
(ONR), World Bank/GEF Coral Reef
Caldeira, R.M.A., P. Marschesiello, N.P. Nezlin,
P.M. DiGiacomo, and J.C. McWilliams. 2005.
We reviewed present-day capabilities of
Targeted Research Program, and the
Island wakes in the Southern California Bight.
modeling population connectivity in
University of Miami's Maytag Chair in
Journal of Geophysical Research 110:C11012,
marine systems, showing that simultane-
Ichthyology during the preparation of
doi:10.1029/2004JC002675.
Chang, S. 2007. Small-Scale Flow Variability Inside
ous inclusion of physical and biological
this manuscript. We gratefully acknowl-
Branched Coral Colonies: Computations and
processes is essential to properly describe
edge this support.
Experimental Validation. Ph.D. dissertation,
Stanford University, Stanford, CA, 158 pp.
and quantify organisms' dispersal, settle-
Chassignet, E.P., H.E. Hurlburt, O.M. Smedstad, G.R.
ment, and ultimate distribution. We
reFereNCeS
Halliwell, A.J. Wallcraft, E.J. Metzger, B.O. Blanton,
C. Lozano, D.B. Rao, P.J. Hogan, and A. Srinivasan.
also discussed some of the remaining
Andréfouët, S., E.J. Hochberg, C. Chevillon, F.E.
Muller-Karger, J.C. Brock, and C. Hu. 2005. Multi-
2006. Generalized vertical coordinates for eddy-
challenges. Model formulations need
scale remote sensing of coral reefs. Pp. 297­315 in
resolving global and coastal ocean forecasts.
to better represent physical dynamics,
Remote Sensing of Aquatic Coastal Environments.
Oceanography 19(1):20­31.
R.L. Miller, C.E. Del Castillom, and B.A. McKee,
Codling, E.A., J.W. Pitchford, and S.D. Simpson. 2007.
particularly at intermediate and small
eds., Springer.
Group navigation and the "many-wrongs prin-
scales (i.e., at scales of kilometers or less);
Arhonditsis, G.B., and M.T. Brett. 2004. Evaluation of
ciple" in models of animal movement. Ecology
88:1,864­1,870.
rapid advances in the use of unstruc-
the current state of mechanistic aquatic biogeo-
chemical modeling. Marine Ecology Progress Series
Cowen, R.K. 1985. Large-scale pattern of recruitment
tured meshes and the development of
271:13­26.
by the labrid, Semicossyphus pulcher: Causes and
implications. Journal Marine Research 43:719­774.
novel algorithms that allow for multiscale
Bakun, A. 1996. Patterns in the Ocean: Ocean Processes
and Marine Population Dynamics. University of
Cowen, R.K. 2002. Larval dispersal and retention and
problems are both promising and nec-
California Sea Grant, San Diego, California, 323 pp.
consequences for population connectivity. Pp. 149­
essary. Another element of paramount
Batchelder, H.P., C.A. Edwards, and T.M. Powell. 2002.
170 in Ecology of Coral Reef Fishes: Recent Advances.
Individual-based models of zooplankton popula-
P.F. Sale, ed., Academic Press, San Diego, CA.
importance is the better representation
tions in coastal upwelling regions: Implications
Cowen R.K., K.M.M. Lwiza, S. Sponaugle, C.B. Paris,
of larval behavior and environmental
of diel vertical migration on demographic success
and D.B. Olson. 2000. Connectivity of marine
populations: Open or closed? Science 287:857­859.
gradients and cues in coupled models.
and nearshore retention. Progress in Oceanography
53:307­333.
Cowen R.K., C.B. Paris, and A. Srinivasan. 2006.
And, certainly, more quantitative and
Baums I., C.B. Paris, and L.M. Cherubin. 2006. A
Scaling connectivity in marine populations. Science
sustained measurements are required of
bio-oceanographic filter to larval dispersal in a
311:522­527.
reef-building coral. Limnology and Oceanography
D'Alessandro, E., S. Sponaugle, and T. Lee. 2007.
key physical and biological properties. We
51:1,969­1,981.
Patterns and processes of larval fish supply to
suggest that modeling and observational
Becker, B.J., L.A. Levin, F.L. Fodrie, and P.A. McMillan.
the coral reefs of the upper Florida Keys. Marine
2007. Complex larval connectivity patterns among
Ecology Progress Series 331:85­100.
improvements should not occur inde-
marine invertebrate populations. Proceedings of the
Dower, J.F., T.J. Miller, and W.C. Leggett. 1997. The
pendently of each other. Models should
National Academy Sciences of the United States of
role of microscale turbulence in the feeding ecol-
ogy of larval fish. Advances in Marine Biology
guide in the design of field experiments,
America 104:3,267­3,272.
Benfield, M.C., P. Grosjean, P.F. Culverhouse, X.
31:169­220.
and in turn observations should improve
Irigoien, M.E. Sieracki, A. Lopez-Urrutia, H.G.
Eggleston, D.B., L.S. Lipcius, L.S. Marshall Jr., and S.G.
the model parameters. Finally, encour-
Dam, Q. Hu, C.S. Davis, A. Hansen, and others.
Ratchford. 1998. Spatiotemporal variation in post-
2007. RAPID: Research on Automated Plankton
larval recruitment of the Caribbean spiny lobster
aged by the advances achieved in the past
Identification. Oceanography 20(2):172­187.
in the central Bahamas: Lunar and seasonal period-
few decades, models addressing popula-
Bode, M., L. Bode, P.R. Armsworth. 2006. Larval dis-
icity, spatial coherence, and wind forcing. Marine
persal reveals regional sources and sinks in the
Ecology Progress Series 174:33­49.
tion connectivity will be used to pose and
Great Barrier Reef. Marine Ecology Progress Series
Fiechter, J., and C.N.K. Mooers. 2003. Simulation
address ecologically relevant questions,
308:17­25.
of frontal eddies on the east Florida shelf.
Geophysical Research Letters 30(22):2151,
will have forecasting capabilities, and will
Boehlert, G.W., and B.C. Mundy. 1988. Roles of behav-
ioral and physical factors in larval and juvenile fish
doi:10.1029/2003GL018307.
be useful to resource managers.
recruitment to estuarine nursery areas. Pp. 51-67
Fringer, O.B., J.C. McWilliams, and R.L. Street. 2006.
Oceanography September 2007
67

---

A new hybrid model for coastal simulations.
in Oceanography 53:335­367.
face mixed layer. Journal of Theoretical Biology
Oceanography 19(1):46­59.
Hermann, A.J., S. Hinckley, B.A. Megrey, and J.M.
234:565­591.
Fuchs, H.L., M.G. Neubert, and L.S. Mullineaux. 2007.
Napp. 2001. Applied and theoretical considerations
Llopiz, J.K., and R.K. Cowen. In press.
Effect of turbulence-mediated larval behavior
for constructing spatially-explicit individual-based
Trophodynamics of larval billfishes: Precocious
on larval supply and settlement in tidal currents.
models of marine larvae that include multiple
and selective feeding with support for minimal
Limnology and Oceanography 52:1,156­1,165.
trophic levels. ICES Journal of Marine Science
starvation in the low-latitude open ocean. Marine
Gerlach, G., J. Atema, M.J. Kingsford, K.P. Black, and V.
58:1030­1041.
Ecology Progress Series.
Miller-Sims. 2007. Smelling home can prevent dis-
Hill, A.E. 1994. Horizontal zooplankton dispersal by
Lynch, D.R., and A.M. Davies, eds. 1995. Quantitative
persal of reef fish larvae. Proceedings of the National
diel vertical migration in S2 tidal currents on the
Skill Assessment for Coastal Ocean Models. Coastal
Academy Sciences of the United States of America
northwest European continental shelf. Continental
and Estuarine Series Volume 47, 510 pp.
104:858­863.
Shelf Research 14:491-506.
Lynch, D.R., J.T.C. Ip, C.E. Naimie, and F.E. Werner.
Giske, J., G. Huse, and Ø. Fiksen. 1998. Modelling spa-
Hofmann, E.E., and M.A.M. Friedrichs. 2002.
1996. Comprehensive Coastal Circulation Model
tial dynamics of fish. Reviews in Fish Biology and
Predictive modeling for marine ecosystems. Pp.
with Application to the Gulf of Maine. Continental
Fisheries 8:57­91.
537­565 in The Sea. Volume 12: Biological-Physical
Shelf Research 16:875­906.
Giske J., M. Mangel, P. Jakobsen, G. Huse, C. Wilcox,
Interactions in the Ocean. A.R. Robinson, J.J.
Miller, D., C.L. Moloney, C.D. van der Lingen, C.
and E. Strand. 2003. Explicit trade-off rules in
McCarthy, and B.J. Rothschild, eds, John Wiley and
Lett, C. Christian Mullon, and J.G. Field. 2006.
proximate adaptive agents. Evolutionary Ecology
Sons, Inc., NY.
Modeling the effects of physical­biological inter-
Research 5:835­865.
Huse, G. 2001. Modelling habitat choice in fish using
actions and spatial variability in spawning and
GLOBEC.INT, 1994. An Advanced Modeling/
adapted random walk. Sarsia 86:477­483.
nursery areas on transport and retention of sardine
Observation System (AMOS) for physical-biologi-
Irisson, J.O., A. LeVan, M.D. Lara, and S. Planes. 2004.
Sardinops sagax eggs and larvae in the southern
cal-chemical ecosystem research and monitor-
Strategies and trajectories of coral reef fish larvae
Benguela ecosystem. Journal of Marine Systems
ing (concepts and methodology). Globec Special
optimizing self-recruitment. Journal of Theoretical
61:212­229.
Contributions Technical Report, No. 2, 149 pp.
Biology 227:205­218.
Monismith, S.G. 2007. Hydrodynamics of coral reefs.
Greenberg, D.A., F. DuPont, F. Lyard, D.R. Lynch,
Kettle, A.J., and K. Haines. 2006. How does the
Annual Review of Fluid Mechanics 39:37­55.
and F.E. Werner. 2007. Resolution issues in
European eel (Anguilla anguilla) retain its popula-
Mullon, C., P. Fréon, C. Parada, C. Van Der Lingen,
numerical models of oceanic and coastal
tion structure during its larval migration across
and J. Huggett. 2003. From particles to individuals:
circulation. Continental Shelf Research:
the North Atlantic Ocean? Canadian Journal of
Modelling the early stages of anchovy (Engraulis
doi:10.1016/j.csr.2007.01.023.
Fisheries and Aquatic Sciences 63:90­106.
capensis/encrasicolus) in the southern Benguela.
Haidvogel, D.B., and A. Beckmann. 1999. Numerical
Kinlan, B.P., S.D. Gaines, and S.E. Lester. 2005.
Fisheries Oceanography 12:396­406.
Ocean Circulation Modeling. Imperial College
Propagule dispersal and the scales of marine
Mullon C., C. Parada P. Cury, J. Huggett, P. Freon,
Press, London, 318 pp.
community process. Diversity and Distributions
C. van der Lingen, P. Penven, C. Moleney, and
Hare, J.A., and R.K. Cowen. 1996. Transport mecha-
11:139­148.
A. Richardson. 2002. Individual based model-
nisms of larval and pelagic juvenile bluefish
Koehl, M.A.R., and T.M. Powell. 1994. Turbulent
ing (IBM) of the early stages of anchovy in the
(Pomatomus saltatrix) from South Atlantic Bight
transport of larvae near wave-swept shores:
southern Benguela System. Pp. 87­89 in Report of
spawning grounds to Middle Atlantic Bight
Does water motion overwhelm larval sinking?
a GLOBEC-SPACC/IDYLE/ENVIFISH Workshop
nursery habitats. Limnology and Oceanography
Pp. 261­274 in Reproduction and Development of
of spatial approaches to the dynamics of coastal
41:1,264­1,280.
Marine Invertebrates. H. Wilson, G. Shinn, and
pelagic resources and their environment in upwell-
Hare, J.A., J.A. Quinlan, F.E. Werner, B.O. Blanton,
S. Stricker, eds, Johns Hopkins University Press,
ing areas. C.D. van der Lingen, C. Roy, P. Freon, M.
J.J. Govoni, R.B. Forward, L.R. Settle, and D.E.
Baltimore, MD.
Barange, L. Castro, M Gutierrez, L. Nykjaer and F.
Hoss. 1999. Influence of vertical distribution on
Kraines, S.B., I. Masahiko, and H. Komiyama. 2004.
Shillington, eds, GLOBEC Report 16.
the outcome of larval transport during winter
Seasonal variations in the exchange of water
North, E.W., A. Gallego, and P. Petitgas. In press a.
in the Carolina Capes Region: Results of a three-
and water-borne particles at Mjuro Atoll, the
Manual of Recommended Practices for Modelling
dimensional hydrodynamic model. Fisheries
Republic of the Marshall Islands. Coral Reefs
Physical-Biological Interactions in Fish Early-Life
Oceanography 8(Suppl. 2):57­76.
20(4):4,722­4,028.
History. ICES Cooperative Research Reports.
Harley, D.C.G., A.R, Hughes, K.M. Hultgren, B.G.
Legrand, S., E. Deleersnijder, E. Hanert, V. Legat, and
North, E.W., Z. Schlag, R.R. Hood, M. Li, L. Zhong,
Miner, C.J.B. Sorte, C.S. Thornber, L.F. Rodriguez,
E. Wolanski. 2006. High-resolution, unstructured
T. Gross, and V.S. Kennedy. In press b. Vertical
L. Tomanek, and S.L. Williams. 2006. The impacts
meshes for hydrodynamic models of the Great
swimming behavior influences the dispersal of
of climate change in coastal marine systems.
Barrier Reef, Australia, Estuarine Coastal and Shelf
simulated oyster larvae in a coupled particle-track-
Ecology Letters 9:228­241.
Science 68:36­46.
ing and hydrodynamic model of Chesapeake Bay.
Hastings, A., and L.W. Botsford. 2006. Persistence of
Lermusiaux P.F.J., P. Malanotte-Rizzoli, D. Stammer,
Marine Ecology Progress Series.
spatial populations depends on returning home.
J. Carton, J. Cummings, and A.M. Moore. 2006.
O'Connor, M.I., J.F. Bruno, S.D. Gaines, B.S. Halpern,
Proceedings of the National Academy Sciences of the
Progress and prospects of U.S. data assimilation in
S.E. Lester, B.P. Kinlan, and J.M. Weiss. 2007.
United States of America 103:6,067­6,072.
ocean research. Oceanography 19(1):172­183.
Temperature control of larval dispersal and
Helbig, J.A., and P. Pepin. 1998. Partitioning the influ-
Lett, C., C. Roy, A. Levasseur, C. van der Lingen,
the implication for marine ecology, evolution
ence of physical processes on the estimation of
and C. Mullon. 2006. Simulation and quantifi-
and conservation. Proceedings of the National
ichthyoplankton mortality rates. II. Application
cation of enrichment and retention processes
Academy Sciences of the United States of America
to simulated and field data. Canadian Journal of
in the southern Benguela upwelling ecosystem.
104:1,266­1,271.
Fisheries Aquatic Sciences. 55:2,206­2,220.
Fisheries Oceanography:doi:10.1111/j.1365-
Page, F.H., M. Sinclair, C.E. Naimie, J.W. Loder, R.J.
Hermann, A.J., D.B. Haidvogel, E.L. Dobbins, and P.J.
2419.2005.00392.x.
Losier, P. Berrien, and R.G. Lough. 1999. Cod and
Stabeno. 2002. Coupling global and regional circu-
Lewis, D.M. 2005. A simple model of plankton popu-
haddock spawning on Georges Bank in relation
lation models in the coastal Gulf of Alaska. Progress
lation dynamics coupled with a LES of the sur-
to water residence times. Fisheries Oceanography
68
Oceanography Vol. 20, No. 3

---

8:212­226.
10.1007/s10236-005-00005-6,.
Trenberth, K. 2005. Uncertainty in hurricanes and
Paris, C.B., L.M. Cherubin, A. Srinivasan, and R.K.
Shenker, J.M., E.D. Maddox, E. Wishinski, S. Pearl, S.R.
global warming. Science 308:1,753­1,754.
Cowen. In press. Surfing, spinning, or diving from
Thorrold, and N. Smith. 1993. Onshore transport
Urban, D., and T. Keitt, 2001. Landscape connec-
reef to reef: How does it change population con-
of settlement-stage Nassau grouper Epinephelus
tivity: A graph-theoretic perspective. Ecology
nectivity? Marine Ecology Progress Series.
striatus and other fishes in Exuma Sound,
28(5):1205­1218.
Paris, C.B., and R.K. Cowen. 2004. Direct evidence
Bahamas. Marine Ecology Progress Series 98:31­43.
Veneziani, M., A. Griffa, Z.D. Garraffo, and E.
of a biophysical retention mechanism for coral
Siegel, D.A., B.P. Kinlan, B. Gaylord, and S.D. Gaines.
Chassignet. 2005. Lagrangian spin parameter and
reef fish larvae. Limnology and Oceanography
2003. Lagrangian descriptions of marine larval dis-
coherent structures from trajectories released in
49(6):1,964­1,979
persion. Marine Ecology Progress Series 260:83­96.
high-resolution ocean models. Journal of Marine
Paris, C.B., R.K. Cowen, R. Claro, and K.C. Lindeman.
Spagnol, S., E. Wolanski, E. Deleersnijder, R.
Research 63:753­788.
2005. Larval transport pathways from Cuban
Brinkman, F. McAllister, B. Cushman-Roisin, and
Vikebø, F., S. Sundby, B. Adlandsvik, and O. Otterå.
spawning aggregations (Snappers; Lutjanidae)
E. Harnet. 2002. An error frequently made in the
2007. Impacts of a reduced thermohaline circula-
based on biophysical modeling. Marine Ecology
evaluation of advective transport in two-dimen-
tion on transport and growth of larvae and pelagic
Progress Series 296:93­106.
sional Lagrangian models for advection-diffusion
juveniles of Arcto-Norwegian cod (Gadus morhua).
Pineda, J. 1994. Internal tidal bore in the nearshore:
in coral reef waters. Marine Ecology Progress Series
Fisheries Oceanography 16:216­228.
Warm-water fronts, seaward gravity currents and
235:299­302.
Walstad, L.J., and D.J. McGillicuddy. 2000. Data assim-
the onshore transport of neustonic larvae. Journal
Speirs, D., W. Gurney, M. Heath, and S.N. Wood.
ilation for coastal observing systems. Oceanography
of Marine Research 52:427­458.
2005. Modeling the basin-scale demography of
13(1):47­53.
Planes, S. 2002. Biogeography and larval dispersal
Calanus finmarchicus in the N.E. Atlantic. Fisheries
Werner, F.E., F.H. Page, D.R. Lynch, J.W. Loder, R.G.
inferred from population genetic analysis. Pp. 201­
Oceanography 14:333­358.
Lough, R.I. Perry, D.A. Greenberg, and M.M.
220 in Ecology of Coral Reef Fishes: Recent Advances.
Sponaugle, S., R.K. Cowen, A. Shanks, S.G. Morgan,
Sinclair. 1993. Influence of mean 3-D advection
P.F. Sale, ed., Academic Press, San Diego, CA.
J.M. Leis, J. Pineda, G. W. Boehlert, M J. Kingsford,
and simple behavior on the distribution of cod and
Purcell, J., R.K. Cowen, C.R. Hughes, and D.A.
K. Lindeman, C. Grimes, and J.L. Munro. 2002.
haddock early life stages on Georges Bank. Fisheries
Williams. 2006. Weak genetic structure indicates
Predicting self-recruitment in marine populations:
Oceanography 2:43­64.
strong dispersal limits: A tale of two coral reef
Biophysical correlates and mechanisms. Bulletin of
Werner, F.E., J.A. Quinlan, B.O. Blanton, and R.A.
fish. Proceedings of the Royal Society B: Biological
Marine Science 70S:341­375.
Luettich Jr. 1997. The role of hydrodynamics in
Sciences 273:1,483­1,490.
Sponaugle, S., T. Lee, V. Kourafalou, and D. Pinkard.
explaining variability in fish populations. Journal of
Rice, J.A., J.A. Quinlan, S.W. Nixon, W.F. Hettler,
2005. Florida Current frontal eddies and the
Sea Research 37:195­212.
S.M. Warlen, and P.M. Stegmann. 1999. Spawning
settlement of coral reef fishes. Limnology and
Werner, F.E., J.A. Quinlan, R.G. Lough, and D.R.
and transport dynamics of Atlantic menhaden:
Oceanography 50:1,033­1,048.
Lynch. 2001a. Spatially-explicit individual based
Inferences from characteristics of immigrating
Strand, E., G. Huse, and J. Giske. 2002. Artificial
modeling of marine populations: A review of the
larvae and predictions of a hydrodynamic model.
evolution of life history and behavior. American
advances in the 1990's. Sarsia 86:411­421.
Fisheries Oceanography 8(Suppl. 2):93­110.
Naturalist 159:624­644.
Werner, F.E., B. Mackenzie, R. Perry, R. Lough, C.
Robinson, A.R., and P.F.J. Lermusiaux, 2002. Data
Strathmann, R.R., T.P. Hughes, A.M. Kuris, K.C.
Naimie, B. Blanton, and J. Quinlan. 2001b. Larval
assimilation for modeling and predicting cou-
Lindeman, S.G. Morgan, J.M., Pandolfi, and R.R.
trophodynamics, turbulence and drift in Georges
pled physical-biological interactions in the sea.
Warner. 2002. Evolution of local recruitment and
Bank. A sensitivity analysis of cod and haddock.
Pp. 475­536 in The Sea, Volume 12: Biological-
its consequences for marine populations. Bulletin
Scientia Marina 65:99­115.
Physical Interactions in the Ocean. A.R. Robinson,
of Marine Science 70S:377­396.
Wolanksi, E. 1987. Some evidence for boundary mix-
J.R. McCarthy, and B.J. Rothschild, eds, Harvard
Stegmann, P.M., J.A. Quinlan, F.E. Werner, and B.O.
ing near coral reefs. Limnology and Oceanography
University Press, Cambridge, MA.
Blanton. 1999. Projected transport pathways of
32:735­739.
Rothstein, L.M., J.J. Cullen, M. Abbott, E.P. Chassignet,
Atlantic menhaden larvae as determined from
Wolanski, E., and S. Spagnol. 2000. Environmental
K. Denman, S. Doney, H. Ducklow, K. Fennel, M.
satellite imagery and model simulations in the
degradation by mud in tropical estuaries. Regional
Follows, D. Haidvogel, and others. 2006a. Modeling
South Atlantic Bight. Fisheries Oceanography
Environmental Change 1:152­162.
ocean ecosystems--The PARADIGM program.
8(Suppl. 2):111­123.
Wolanski, E., R. Brinkman, S. Spagnol, F. McAllister, C.
Oceanography 19(1):16­45.
Tamura, H., K. Nadaoka, and C. Paringit. 2007.
Steinberg, W. Skirving, and E. Deleersnijder. 2003a.
Rothstein, L., J. Allen, F. Chai, S. Chen, B. Cornuelle,
Hydrodynamic characteristics of fringing coral
Merging scales in models of water circulation:
K. Fennel, P.F.J. Lermusiaux, R. Murtugudde, Y.
reef on the east coast of Ishigaki island, southwest
Perspectives from the Great Barrier Reef. Advances
Spitz, and F. Werner. 2006b. Report of the ORION
Japan. Coral Reefs 26:17­34.
in Coastal Modeling 15:411­429.
Modeling Committee, April 17, 2006, 15 pp.
Tang, L., J. Sheng, B.G. Hatcher, and P. F. Sale, 2006.
Wolanski, E., R.H. Richmond, G. Davis, and V.
Available online at: http://www.orionprogram.
Numerical study of circulation, dispersion, and
Bonito. 2003b. Water and fine sediment dynam-
org/PDFs/ModelingComm_Report.pdf.
hydrodynamic connectivity of surface waters on
ics in transient river plumes in a small, reef-
Schiegg, K. 2003. Environmental autocorrelation:
the Belize shelf. Journal of Geophysical Research
fringed bay, Guam. Estuarine Coastal Shelf Science
Curse or blessing? Trends in Ecology & Evolution
111:C01003, doi:10.1029/2005JC002930.
56:1,029­1,043.
18:212­214.
Taylor, M.S., and M.E. Hellberg. 2003. Genetic evi-
Wolanski, E., R.H. Richmond, and L. McCook. 2004. A
Scotti, A., and J. Pineda. 2007. Plankton accumulation
dence of local retention of pelagic larvae in
model of the effects of land-based, human activi-
and transport in propagating nonlinear internal
Caribbean reef fish. Science 299:107­109.
ties on the health of coral reefs in the Great Barrier
fronts. Journal of Marine Research 65:117­145.
Tremblay, M.J., J.W. Loder, F.E. Werner, C. Naimie, F.H.
Reef and in Fouha Bay, Guam, Micronesia. Journal
Sheng, J., R.J. Greatbatch, X. Zhai, and L. Tang, 2005.
Page, and M.M. Sinclair. 1994. Drift of scallop lar-
of Marine Systems 46:133­144.
A new two-way nesting technique based on the
vae on Georges Bank: A model study of the roles of
smoothed semi-prognostic method for ocean
mean advection, larval behavior and larval origin.
modeling. Ocean Dynamics 55:162­177, doi:
Deep Sea Research Part II 41:7­49.
Oceanography September 2007
69

---