ReproductiveOutputandDurationofthePelagicLarvalStage ... · From the symposium ‘‘Dispersal of Marine Organisms’’ presented at the annual meeting of the Society for Integrative

SYMPOSIUM

Reproductive Output and Duration of the Pelagic Larval StageDetermine Seascape-Wide Connectivity of Marine PopulationsEric A. Treml,1,* Jason J. Roberts,† Yi Chao,‡ Patrick N. Halpin,† Hugh P. Possingham* andCynthia Riginos*

*School of Biological Sciences, The University of Queensland, St. Lucia, Qld, 4072, Australia; †Marine Geospatial Ecology

Laboratory, Nicholas School of the Environment, Duke University, Durham, NC 27708, USA; ‡Remote Sensing Solutions,

Inc., Pasadena, CA 91107, USA

From the symposium ‘‘Dispersal of Marine Organisms’’ presented at the annual meeting of the Society for Integrative and

Comparative Biology, January 3–7, 2012 at Charleston, South Carolina.

1E-mail: [email protected]

Synopsis Connectivity among marine populations is critical for persistence of metapopulations, coping with climate

change, and determining the geographic distribution of species. The influence of pelagic larval duration (PLD) on con-

nectivity has been studied extensively, but relatively little is known about the influence of other biological parameters,

such as the survival and behavior of larvae, and the fecundity of adults, on population connectivity. Furthermore, the

interaction between the seascape (habitat structure and currents) and these biological parameters is unclear. We explore

these interactions using a biophysical model of larval dispersal across the Indo-Pacific. We describe an approach that

quantifies geographic patterns of connectivity from demographically relevant to evolutionarily significant levels across a

range of species. We predict that at least 95% of larval settlement occurs within 155 km of the source population and

within 13 days irrespective of the species’ life history, yet long-distant connections remain likely. Self-recruitment is

primarily driven by the local oceanography, larval mortality, and the larval precompetency period, whereas broad-scale

connectivity is strongly influenced by reproductive output (abundance and fecundity of adults) and the length of PLD.

The networks we have created are geographically explicit models of marine connectivity that define dispersal corridors,

barriers, and the emergent structure of marine populations. These models provide hypotheses for empirical testing.

Introduction

Quantifying the connectivity of populations is fun-

damental to understanding the dynamics of popula-

tions and metapopulations (Possingham and

Roughgarden 1990; Hanski 1998; Botsford et al.

2009b), the way in which species expand their

range (Lester et al. 2007), and how species might

cope with a changing climate (Hughes et al. 2003).

Connectivity also provides valuable insights into the

geographic distribution of species (Lester et al. 2007),

patterns in genetic divergence (Palumbi 1994; Jones

et al. 2009), and endemism (Meyer et al. 2005;

Paulay and Meyer 2006). As a result, understanding

connectivity and its geographic structure is critical

for effective management and conservation of

marine communities (Jones et al. 2007; Almany

et al. 2009; Planes et al. 2009).

Population connectivity explicitly refers to the

movement of individuals between and within sub-

populations (Cowen et al. 2007). The spatiotemporal

scale of individual (or larval) movement can range

from much less than a meter and minutes to thou-

sands of kilometers and many months depending on

the species’ life-history characteristics and the envi-

ronment (Kinlan and Gaines 2003). Characterizing

population connectivity for a given species and

region, therefore, must span much of this spatiotem-

poral continuum (Levin 2006). At one end of the

spectrum, ecological connectivity is concerned only

with those strong connections that consistently

Integrative and Comparative BiologyIntegrative and Comparative Biology, volume 52, number 4, pp. 525–537

doi:10.1093/icb/ics101 Society for Integrative and Comparative Biology

Advanced Access publication July 19, 2012

� The Author 2012. Published by Oxford University Press on behalf of the Society for Integrative and Comparative Biology. All rights reserved.

For permissions please email: [email protected].

at Princeton University on June 13, 2013

http://icb.oxfordjournals.org/D

ownloaded from

http://icb.oxfordjournals.org/

impact local demographics over short time-scales.

This ecological connectivity may be at levels exceed-

ing 10% of total recruitment to a single site (Cowen

and Sponaugle 2009), translating into hundreds to

many thousands of successful settlers recruiting per

year into a population (Cowen et al. 2006). On the

opposite end of this spectrum, evolutionary connec-

tivity (i.e., gene flow) can occur through rare or

weak connections over time-scales of many genera-

tions. Genetic differentiation is expected to arise at

connectivity levels far below a few individual immi-

grants per generation (Slatkin 1993). Therefore, iden-

tifying the magnitude of connectivity that is

meaningful or relevant is essential to understanding

the scaling and geographic structure of population

connectivity (Cowen et al. 2006), as groups of pop-

ulations may be genetically interconnected yet demo-

graphically isolated (Leis 2002; Swearer et al. 2002).

Here, we focus on tropical coral reef systems and

specifically define marine population connectivity as

the spatially explicit recruitment potential of larvae,

i.e., the likelihood that larvae released at a natal site

will survive and settle in a downstream habitat patch.

This is in contrast to realized connectivity, which

results from a complex combination of four pro-

cesses: reproduction at source patches; transport,

survival, and settlement of larvae; postsettlement sur-

vival; and the eventual reproduction of new migrants

at recipient patches (Hedgecock et al. 2007; Cowen

and Sponaugle 2009). Our analysis of potential pop-

ulation connectivity excluded the postsettlement

processes.

Quantifying the importance of biological parame-

ters and the role of the seascape in determining con-

nectivity has been challenging. Much of the recent

literature has focused on the role of the pelagic larval

duration (PLD) in determining realized dispersal dis-

tances and has resulted in mixed conclusions (Shanks

et al. 2003; Kinlan et al. 2005; Lester and Ruttenberg

2005; Bowen et al. 2006; Gaines et al. 2007; Shanks

2009; Weersing and Toonen 2009; Riginos et al.

2011). For example, PLD has been shown to be a

strong predictor of dispersal distances (Shanks

et al. 2003), yet a poor predictor of genetic similarity

(Weersing and Toonen 2009) and size of species’

ranges (Mora et al. 2012). The length of the larval

precompetency period influences local retention

within the source population (Black et al. 1991;

Paris and Cowen 2004) but has been unexplored

with respect to broad-scale connectivity. The

impact of larval mortality on connectivity can be

profound (Cowen et al. 2000), yet the empirical

data are lacking for most species and environmental

conditions (Graham et al. 2008; Connolly and Baird

2010), and the implications for realized connectivity

are unknown. Type of eggs and larvae are significant

predictors of connectivity in fishes but encapsulate

many correlated biological traits (Bradbury et al.

2008; Riginos et al. 2011). Swimming behavior

(Leis 2007) and larval sensory systems are also be-

lieved to play key roles in larval retention, yet the

influence of larval behavior has only been shown for

a few species and in a few locations (Paris and

Cowen 2004; Gerlach et al. 2007), with other species

showing no pattern (Gerlach et al. 2007). Finally, the

role of the seascape and currents in facilitating or

restricting connectivity is largely unknown, except

for several theoretical studies (Gaines et al. 2003;

Largier 2003) and unique seascapes (James et al.

2002; Baums et al. 2006; Banks et al. 2007; White

et al. 2010). Contributing to these mixed results are

several issues in the design of studies, including (1)

the use of simple Euclidean distance as a proxy for

population connectivity, instead of methods incorpo-

rating geography and oceanic currents (Mitarai et al.

2009; White et al. 2010); (2) bias toward certain spe-

cies (Bradbury et al. 2008) that results from a paucity

of biological data available on life-history parameters,

behavioral characteristics, and mortality; (3) a gen-

eral lack of spatially explicit data and reporting; and

(4) ambiguous definitions of connectivity often used

with respect to the ecological or evolutionary context

of the data.

Unfortunately, measuring and predicting connec-

tivity of marine populations is extremely difficult due

to the unknown and variable biology of larvae, the

complex physical environment, and the inherent dif-

ficulties in considering the range in spatial and tem-

poral scales covered by larval dispersal (Werner et al.

2007; Botsford et al. 2009a; Jones et al. 2009).

Although a diversity of approaches and techniques

have been used to provide information on connec-

tivity, ranging from following larvae in situ (Leis

et al. 2006), marking individual larvae (Jones et al.

2005), and otolith microchemistry (Swearer et al.

2003), to biophysical modeling (Cowen et al. 2000;

Mitarai et al. 2009; Kool et al. 2011), population

genetic approaches (Hedgecock et al. 2007), analyti-

cal approaches (Largier 2003; Botsford et al. 2009a),

and data reanalysis (Bradbury et al. 2008; Riginos

et al. 2011), no single methodology has been able

to provide a full quantitative and geographic picture

of realized population connectivity (Levin 2006;

Cowen and Sponaugle 2009; Jones et al. 2009). The

dynamics and spatial structure of this connectivity,

across scales, remains poorly understood (Siegel et al.

2008).

526 E. A. Treml et al.



ownloaded from


Currently, biophysical modeling provides one of

the few methods that can accommodate the temporal

and spatial scales of marine population connectivity.

These approaches have been used at a variety of

scales as a way to understand the spatial patterns

in connectivity by integrating seascape data with a

species’ life-history characteristics (Sammarco and

Andrews 1988; Black and Moran 1991; Cowen

et al. 2000, 2006; Treml et al. 2008; Mitarai et al.

2009; White et al. 2010; Kool et al. 2011). Although

biophysical modeling can be costly, highly technical,

and difficult to verify, the approach has shed new

light on the processes and patterns of connectivity

for some species and geographies. This approach has

the capacity to help disentangle the mixed results

regarding the extent to which life-history parameters

and seascape characteristics determine realized pop-

ulation connectivity.

Here, we have developed a publically available bio-

physical model of marine dispersal that is capable of

creating spatially explicit predictions of population

connectivity across real seascapes in the Indo-

Pacific Ocean. [See Condie et al. (2005) for a similar

system specific to Australia]. This framework allows

one to model and evaluate the influence of life his-

tory and oceanography on marine population con-

nectivity across a region. Specifically, we: (1) produce

multispecies, geographically explicit models of

marine population connectivity for coral reefs

across the entire Indo-Pacific Ocean, (2) determine

the spatial and temporal structure of multispecies

marine population connectivity, and (3) quantify

the role and importance of key physical and biolog-

ical parameters in determining connectivity.

Materials and methods

A biophysical modeling approach was used to simu-

late larval dispersal among all coral reefs (1002 hab-

itat patches) across the Indo-Pacific Ocean revealing

the strength and structure of population connectiv-

ity. This dispersal model has three components: (1) a

gridded map of the seascape, including coral reefs

and coastline features; (2) biological parameters de-

scribing species’ characteristics of adults and larvae;

and (3) data on oceanic currents’ velocity derived

from an independent regional model. Our modeled

seascape included the marine biodiversity hotspot of

the Coral Triangle and the surrounding reefs of

Australia and Micronesia (Fig. 1), from 1008E to

1708E and 308N to 308S. The seascape features

were derived from high-resolution data on shorelines

(Wessel and Smith 1996) and coral reefs (Spalding

et al. 2001). The biological parameters used were as

follows: larval release time and periodicity, reproduc-

tive output per area of habitat (adult density and

fecundity), maximum PLD, precompetency period,

larval swimming and homing behavior during settle-

ment, and larval mortality (Table 1). These parame-

ters allow a wide range of species and of dispersal

strategies to be represented and their influence on

connectivity quantified. For this study, oceanic cur-

rents were obtained from the US Jet Propulsion

Laboratory’s Regional Ocean Modeling System

(ROMS). This circulation model solves the hydro-

static, primitive equations on a coordinate system

following the terrain, with approximately 12.5 km

horizontal resolution and 30 vertical layers for the

entire Pacific Ocean Basin (658N to 458S, 998E to

708W) using realistic coasts and bathymetry (Wang

and Chao 2004; Wang et al. 2005). ROMS was forced

with the NCEP/NCAR reanalysis air–sea fluxes

(Kalnay et al. 1996), including winds, temperature,

and solar radiation. Although validation is ongoing,

the ROMS data have been shown to reproduce sea-

sonal and interannual variability measured across the

equatorial Pacific (Wang and Chao 2004). As a

result, we believe these data adequately represent me-

soscale to broad-scale dynamics of the ocean across

the study region. A 3-year subset was chosen to rep-

resent the decadal-scale and seasonal variability:

1997, a strong El Nino year; 1999, a strong La

Nina year; and 2001, a neutral year. Data on surface

currents were averaged over 3 days and saved and

interpolated to the modeled seascape grid for use in

this analysis. A diffusivity term (50 m2 s�1) was in-

cluded in the dispersal model to represent oceanic

turbulence below the resolution of the ROMS circu-

lation data.

A quantitative assessment of the biological and

physical attributes affecting dispersal of marine

larvae was carried out across species. The simulations

modeled the 2D larval dispersal kernel directly, as a

‘‘cloud’’ of larvae, not from individually based tracks

of particles or larvae. This larval density, or proba-

bility, surface was advected, dispersed, and/or con-

centrated, dependent on the biophysical parameters.

The model moved the dispersal kernel through

the current velocity fields using an advection trans-

port algorithm (Smolarkiewicz 1983; Smolarkiewicz

and Margolin 1998; Smolarkiewicz 2006). This trans-

port scheme minimizes numerical diffusion, mini-

mizes computational requirements, and quantifies

the full dispersal kernel (including the evolutionarily

significant tails) that is often difficult to resolve using

particle-tracking methods.

A simulation of dispersal consisted of releasing a

cloud of larvae over a habitat patch and tracking the

Determinants of marine population connectivity 527



ownloaded from


Fig. 1 The model domain extending from 1008E to 1708E, 308N to 308S. The inset highlights the model’s grid structure and resolution

around a representative coastal area (Cenderawasih Bay, Indonesia). Reef habitat cells are shown in black and land cells in gray.

Table 1 Biological parameters used to represent a species

Parameter Description Acceptable values

Initial larval release time Date and time dispersal starts Month, day, year, and time

Larval release site Initial location of larval release or habitat patch Spatial location represented in gridded habitat

map

Larval output per habitat area Adult density * fecundity Integer value, or relative amount (e.g., 1 unit)

Maximum pelagic larval duration Maximum length of the larval dispersal period Integer value (5, 15, 20, 25, 30, 40, 50, 60 days)

Larval precompetency period Period of time in which larvae cannot settle Gamma probability density function with shape

and scale parameters (e.g., scale¼ 10,

shape¼ .5 results in �5 day precompetency

period)

Daily settlement likelihood Probability that larvae will settle on suitable

habitat with time

Probability value (0.50, 0.70, 0.80, 0.90, 0.94)

Homing behavior Homing behavior used within sensory zone.

If TRUE, all larvae within a habitat cell will

settle, if FALSE, settlement will occur in

proportion to the amount of available habitat

TRUE/FALSE

Daily larval mortality Larval mortality probability over time while

planktonic

A probability function or daily probability

(e.g., 0.5, 0.4, 0.3, 0.2, 0.1, 0.0)




ownloaded from


cloud as it moved through the seascape. As this

cloud came in contact with suitable habitat, the

quantity of larvae settled was recorded. This dispersal

process was repeated for every habitat patch in the

seascape until all larval connections were quantified.

The total quantity of larvae that settled on every

habitat patch was recorded through time and saved

as the dispersal matrix. A number of connectivity

matrices were derived from this dispersal matrix.

The settlement matrix, S, represents the cumulative

number of larvae exchanged between all patches after

considering larval mortality. The connectivity proba-

bility matrix rescales S to the probability of larval

exchange between patches, and its diagonal repre-

sents local retention. The migration matrix quantifies

the proportion of settlers to each patch that came

from each source patch, and the diagonal of this

matrix representing self-recruitment. These matrices,

along with the location of the habitat, were used to

build connectivity networks showing the geographic

structure of the populations’ connectivity (Cowen

et al. 2006; Treml et al. 2008).

The migration rate threshold (MRT) represents a

critical recruitment or connectivity level used in de-

termining what connections are relevant (Cowen

et al. 2006). This limit may be in terms of the pro-

portion of successful settlers or a required number of

larval recruits and provides a method for separating

evolutionarily relevant connectivity from demo-

graphically significant levels (Cowen and Sponaugle

2009). For example, the MRT can be set relative to

the reproductive output of the species of interest (e.g.,

1 recruit out of 100,000 larvae released), or calcu-

lated to match adult mortality or to maintain popu-

lation persistence (Cowen et al. 2006, Supplementary

Table S4). The advection transport algorithm has a

high level of precision (e.g., 1 out of 109 larvae per

model cell) and allows connectivity to be estimated

along this entire gradient from ecologically significant

to evolutionarily significant levels.

A series of 1140 larval dispersal simulations were

completed for the entire Indo-Pacific Ocean to quan-

tify multispecies connectivity across a full range of

parameter values (Table 1) representing unique life

histories. The resultant connectivity matrices and

networks were used to quantify the geographic struc-

ture and scaling in population connectivity, and to

determine the importance of biological and physical

parameters to connectivity.

Marine population connectivity networks were il-

lustrated for three model species: (1) a coral with

high fecundity and abundance, no homing, and

weak swimming capabilities, a 60-day PLD with

a 7-day precompetency period, and an annual

spawning periodicity; (2) a damselfish with moderate

fecundity and abundance, strong swimming, seasonal

spawning periodicity, and a 20-day PLD with a 9-day

precompetency period; (3) an anemonefish with low

fecundity and abundance, strong swimming, homing

behavior, a 10-day pelagic stage with a 2-day pre-

competency period, and seasonal spawning. The con-

nectivity probability matrix was calculated and the

networks mapped to illustrate the geographic struc-

ture emerging from the process of larval dispersal.

These maps and matrices represent geographically

explicit hypotheses of the marine population connec-

tivity for these model species. See Supplementary

Material for technical details of the model.

Scaling in connectivity

The estimates of connectivity for the three model

species were used to illustrate the spatial and tempo-

ral context of marine population connectivity across

the Indo-Pacific by examining graphs of cumulative

larval settlement. Larval settlement was calculated by

summing the connections (for the entire Indo-Pacific

seascape) in the settlement matrix for each species

and across geographic distance and larval durations.

The cumulative larval settlement was plotted against

the downstream distance from the source population

(kilometers) and against time spent in the pelagic

larval stage (days).

Geographic distance is often used as a proxy for

population connectivity under the assumption that

populations that are geographically closer would

have stronger connectivity than would those farther

apart. To explore the validity of this assumption for

marine systems, we explored the relationship be-

tween geographic distance and the strength of con-

nection (probability) across the seascape for the three

model species. A tight inverse relationship (e.g., neg-

ative exponential) could justify the use of geographic

distance as a proxy for population connectivity.

Importance of physical and biological parameters

in local and seascape-wide connectivity

To explore the relative importance of the biological

and physical parameters in determining local to re-

gional population connectivity, we used a model sen-

sitivity approach based on variance decomposition

for multiple linear regression models (Groemping

2007). Using the relaimpo package in R (R

Development Core Team 2011), we calculated the

relative importance of biophysical parameters using

two complementary methodologies: the LMG

(Lindeman, Merenda, and Gold) approach based

on the sequential sums of squares and the




ownloaded from


Proportional Marginal Variance Decomposition

(PMVD) approach. Data were derived from the

series of 1140 matrices of population connectivity,

each representing the regional connectivity resulting

from a unique suite of life-history traits (e.g., PLD,

mortality, behavior) and time period. We first used

this approach to ask what biological and physical

parameters determine local connectivity on a

reef-patch level, in terms of percent self-recruitment

and percent local retention. Percent self-recruitment

was defined as the proportion of total larval settlers

to a habitat patch that originated from that patch.

Local retention was defined as the proportion of

larvae released from a source patch that settled

back to that patch. The biological parameters

include PLD, larval mortality, homing behavior,

precompetency period, likelihood of settlement, and

the MRT (which scales with reproductive output).

The physical parameters include the following: hab-

itat area, patch size, strength of local currents, sur-

rounding suitable habitat, and the number and area

of upstream sources. The patch-level analysis was

restricted to those ecologically significant connec-

tions that were made within 20 days and that con-

tributed more than 0.1% to the total settlement of

individual patches (961,920 observations of 1002 reef

patches over 960 dispersal simulations). We then

asked what biological parameters are most important

in determining seascape-wide connectivity. We used

seven metrics to quantify the seascape connectivity

for each simulation (response variables), all related

to the broad-scale connectedness: total number of

dispersal connections, median distance, upper quar-

tile distance, maximum distance, total settlement,

number of connected clusters, and the size of the

largest connected cluster. The seascape-wide analysis

was based on the full suite of biological parameters,

across all years and seasons. The model’s response

variables, regressors, and values are detailed in

Supplementary Table S1.

Results

Geographic structure of marine population

connectivity

After completing a series of dispersal simulations, the

resultant population connectivity networks were

mapped using the locations of the source and desti-

nation and the estimates of connectivity strength.

The geographic patterns of population connectivity

for the three model species are shown in Fig. 2. Each

dispersal network represents the connectivity poten-

tial for that species, in terms of the probability that

larvae released at a source patch will settle in a

Fig. 2 Population connectivity networks for the three model

species: (a) coral, (b) damselfish, and (c) anemonefish. Red nodes

represent coral reef habitat with the size scaled to the area of

habitat. Dispersal connections between reefs are represented by

gray and white links for strengths above and below the MRT,

respectively. The MRT for each species is as follows: 10�7 (coral),

10�5 (damselfish), and 10�2 (anemonefish).




ownloaded from


particular habitat patch based on the unique charac-

teristics of the seascape and of larval life histories.

We mapped the species’ connectivity networks with

MRTs scaled to the species-specific reproductive

output. Visualizing the connections made above

and below this MRT illustrates the geographic impli-

cations of focusing on evolutionary levels (below

MRT) that include all rare, long-distant, or weak

connections; to stronger ecologically significant

levels (above MRT) where only strong and/or persis-

tent connections are concerned.

Scaling in marine population connectivity

The spatial and temporal scaling of population con-

nectivity for the three model species was quantified

by plotting the cumulative proportional settlement of

all larvae with respect to distance from the source

population and number of days into the pelagic

larval phase. Proportional cumulative settlement

with respect to the time spent in the planktonic

larval phase is shown in Fig. 3a, revealing a temporal

aspect of population connectivity. Proportional set-

tlement of larvae was greatest for these three model

species during the period immediately following the

onset of larval competency. The 95% threshold for

cumulative larval settlement occurred within 13 days

for the coral, 9 days for the damselfish, and 4 days

for the anemonefish. As the duration of the larval

stage increased, larvae continued to settle in suitable

habitat, yet the proportion of successful settlement

later in the pelagic larval period was much lower.

The spatial scaling of marine connectivity across

the same species is shown in Fig. 3b. Most settlement

occurred within 155 km of the source patch for

all species (95% of the total cumulative settlement

occurred within 147, 155, and 28 km for the coral,

damselfish, and anemonefish, respectively). Although

dispersal connections were made at greater distances

(Fig. 2), the proportion of larvae settling at these

greater distances was relatively small.

To explore the relationship between strength of

connectivity and geographic distance, we combined

data across the three model species for all connec-

tions made between habitat patches. We fit the data

to a negative exponential curve to quantify the rela-

tionship between connectivity and distance. Across

these species, over 22% of the variance was explained

(Fig. 4). The relationship and fit were similar for the

two species of fish [damselfish: y¼ 0.947x(�1.091),

R2¼ 0.32; anemonefish: y¼ 1.023x(�1.003), R2

¼

0.32], but somewhat different for the coral species

[y¼ 0.051x(�0.985), R2¼ 0.10]. The fit weakens when

restricted to distances less than 200 km (e.g.,

R2¼ 0.16 for full dataset) with great variability in

connectivity strength and direction across these

shorter distances.

Importance of physical and biological parameters in

local and seascape-wide connectivity

A variance-decomposition approach (Groemping

2007) was used to explore the relative importance

of the biological and physical parameters in deter-

mining population connectivity across all simulations

and time periods. The importance analysis for

Fig. 3 Spatial and temporal scaling of larval settlement. For each of the three model species mapped in Fig. 2, the cumulative

settlement of larvae was calculated with respect to days into the larval period (a) and distance from the source population

(b). For each species, the connectivity probability was used to calculate the proportional settlement of larvae across the entire

seascape. Across species, the vast majority of larval settlement occurs within 155 km and 13 days from larval release, irrespective of

larval life-history characteristics. The 90% and 95% cumulative settlement thresholds per species are as follows: coral larvae settled

within 88 km in 11 days (90%) and 147 km in 13 days (95%), Damselfish larvae settled within 82 km in 8 days (90%) and 155 km in 9

days (95%), and anemonefish larvae settled within 21 km in 3 days (90%) and 28 km in 4 days (95%).




ownloaded from


patch-level self-recruitment and local retention is

shown in Table 2 and analysis of seascape-wide con-

nectivity is shown in Table 3. The results from the

importance analysis were consistent between the two

approaches, so only the LMG values are shown here;

the PMVD results are presented in Supplementary

Tables S2 and S3. Across the seascape and all dis-

persal realizations, self-recruitment was generally

high (mean of 53%). The relationship between

patch-level self-recruitment and the biological and

physical variables was relatively strong (R2¼ 0.48)

and determined primarily by local oceanography,

characteristics of upstream habitats (number and

size of larval sources), the size of the focal habitat

patch, and several larval characteristics (mortality,

length of the precompetency period, and PLD).

Local retention, or the proportion of larvae released

that settle in the natal patch, was consistently low

across all patches and simulations (mean of 5%).

The most important variable determining local reten-

tion (R2¼ 0.39) was the length of the precompetency

period, followed by the presence of homing behavior,

larval mortality, and the strength of local currents.

Across all seven seascape connectivity metrics, the

proportion of variance explained by the biological

parameters was high, ranging between 53 and 75%

(Table 3). Across all seascape-wide connectivity met-

rics, the most important biological parameters deter-

mining connectivity were as follows: (1) duration of

the pelagic larval stage, (2) the MRT, and (3) larval

mortality. Behavior and larval competency consis-

tently played a minor role in regional connectivity.

Discussion

This model provides an approach for generating geo-

graphically explicit larval-connectivity predictions in-

corporating empirical data on the reproductive

output, larval life histories, and the seascape (habitat

configuration and oceanic currents). We discovered

that broad-scale connectivity of marine populations

is determined by the reproductive output of the sub-

populations (MRT), larval mortality, and PLD,

whereas the magnitude of self-recruitment and local

retention is determined primarily by larval behavior,

the length of the precompetency period, local cur-

rents, and the number of populations upstream. This

approach enables the estimation of connectivity from

local and demographically relevant scales to evolu-

tionarily significant levels when rare or weak connec-

tions and persistent barriers may dominate the

spatial patterns of interest. This work moves

beyond identifying a general scale of population con-

nectivity by quantifying the geographic structure of

the full dispersal kernel (Palumbi 2004; Bradbury

et al. 2008), including the strength and direction of

potential connectivity.

Fig. 4 Relationship between geographic distance and the strength

of population connectivity across all model species. There is one

datum point per successful dispersal connection per species.

Values of connectivity strength are in terms of the proportion of

larvae released that survive and settle in downstream habitat.

Distance is in kilometers from the source patch and local re-

tention of larvae was excluded. Data fit to a negative exponential

curve, with best fit of y¼ 0.664x(�1.032), R2¼ 0.22.

Table 2 Relative importance analysis of local-scale population connectivity

Importance analysis for local-scale connectivity

Variable importance (normalized to sum to 100%)

Response R2(%) PLD Mortality Homing Precompetency Settle Habitat

area

Patch

area

Currents Area

in 50km

Sources Upstream

area

Self-recruitment 48.3 2.7 5.4 0.3 3.0 0.2 2.3 5.6 45.3 1.3 25.1 8.8

Local retention 39.2 0.4 14.4 20.8 32.6 8.8 2.9 6.2 13.7 – – –

Note. Results for self-recruitment at an ecologically relevant MRT (MRT40.001), and local retention, showing the proportion of variance

explained by the model (R2), and the normalized importance values for each parameter, based on the LMG approach. The biological parameters

are defined in text and in Table 1.




ownloaded from


Essential criteria for evaluating all estimates of

connectivity include understanding the assumptions

of the model (e.g., parameters, process, resolution)

and aligning the inherent spatial and temporal scale

of the data with the intended uses. For example, al-

though dispersal of a few larvae may be widespread,

it is often undesirable to consider such rare events,

especially when one is concerned with only ecologi-

cally significant processes or management decisions.

As a result, it is critical to scale these connectivity

estimates appropriately to match the species’ charac-

teristics when determining realistic or demographi-

cally significant settlement rates (Cowen et al. 2006).

Although this scaling may vary greatly among spe-

cies, critical annual settlement rates relevant to de-

mographic processes may be on the order of one to

five successful settlers per reproductive adult per year

into the receiving population (Cowen et al. 2006).

These settlement rates translate to MRTs (critical

proportion of larvae released) between 0.01 and

0.00001, depending on the species’ life histories (ex-

amples in Supplementary Table S4). Identifying these

thresholds provides a means of interpreting estimates

of connectivity in the context of the species of inter-

est and of focus of the study (ecological versus

evolutionary).

Importance of physical and biological parameters

in local and seascape-wide connectivity

The predictions presented here are generally consis-

tent with many theoretical studies (Botsford et al.

2001; Siegel et al. 2003; Hastings and Botsford

2006), other broad-scale biophysical models of con-

nectivity (Cowen et al. 2006; Kool et al. 2011), and a

number of empirical studies exploring the impor-

tance of life-history parameters in relation to

potential for dispersal (Bradbury et al. 2008;

Riginos et al. 2011). Our estimates of

self-recruitment, with a mean of 53% across all sim-

ulations and locations, are supported by field studies

across a number of species and locations (e.g., Jones

et al. 1999; Almany et al. 2009). Although highly

variable, the magnitude of self-recruitment is depen-

dent on the characteristics of the local seascape (cur-

rents, habitat topology), and on larval mortality, and

the larval precompetency period. The mean local

larval retention is 5% and predominantly determined

by larval biology (precompetency period, behavior,

mortality) and local oceanography. It is important

to realize that these low values of larval retention

translate to many hundreds of larvae when the pro-

portions are scaled to the species’ reproductive

output (Supplementary Table S4).

Although the length of the pelagic dispersal stage

appears to play a minor role in local-scale connec-

tivity (Table 2), it is much more important in deter-

mining broad-scale connectivity across the seascape

(PLD is of primary importance in 6/7 metrics). In

general, the longer larvae can remain in the plank-

ton, the more opportunities there are for establishing

long-distant connections, leading to a more con-

nected seascape. This is not to imply that PLD de-

termines the connectivity distance per se, but that it

contributes to the development of broad-scale pat-

terns in connectivity. Although the correlation be-

tween PLD and distance has been a recent focus in

the literature (Shanks et al. 2003; Shanks 2009;

Weersing and Toonen 2009), we demonstrate that

it may be a better predictor of evolutionary connec-

tions (Supplementary Fig. S1a) and a rather weak

predictor of ecological connectivity (Supplementary

Fig. S1b). These results, combined with the strong

Table 3 Relative importance analysis of seascape-wide population connectivity

Importance analysis for seascape-wide connectivity

Variable importance (normalized to sum to 100%)

Response R2(%) MRT PLD Mortality Homing Precompetency Settle

Total connections 66.1 37.8 42. 3 13.8 ns ns ns

Median distance (km) 75.0 40.5 39.2 14.8 3.6 ns ns

Upper quartile (km) 73.7 35.5 44.4 14.3 ns ns ns

Maximum distance (km) 66.9 24.1 57.1 11.0 ns ns ns

Total settlement 58.2 ns 13.8 63.3 15.3 2.4 ns

Number of clusters 53.5 66.2 8.5 12.5 8.2 0.9 3.7

Size of largest cluster 70.8 50.9 30.9 11.0 4.8 0.3 ns

Note. Results are shown for all connectivity metrics, showing the proportion of variance explained by the model (R2), and the normalized

importance values for each regressor variable. Nonsignificant (ns) parameters from the regression model were removed from this table.




ownloaded from


geographic structuring, may help explain the mixed

patterns found between PLD and dispersal distances

in the literature (Weersing and Toonen 2009).

Scaling in connectivity

Here, we reveal the full geographic structure of po-

tential connectivity of marine populations across the

Indo-Pacific Ocean (Fig. 2) and find the spatial scale

to be on the order of 150 km. These patterns support

previous estimates and also add a temporal context

to connectivity (Fig. 3a), revealing that most larval

settlement occurs with the first 13 days of the pelagic

larval stage, dependent on the characteristics of larval

competency, irrespective of PLD, larval mortality,

larval behavior, and other life-history characteristics.

These spatial and temporal predictions warrant ad-

ditional work and validation to determine whether

(or where) these patterns hold true across species.

Interestingly, (1) the lack of a consistently strong

correlation between connectivity strength and geo-

graphic distance across species (Fig. 4) and (2) the

strong asymmetries in connectivity, highlight the

need to reconsider using Euclidean distance as a

proxy for population connectivity in related research.

For example, genetic models should be able to incor-

porate isolation by ‘‘dispersal distance’’ as an alter-

native to geographic distance, and marine managers

could consider using these, or similar, ecologically

meaningful measures of distance (and direction) to

replace spatial proximity or adjacency in decision

software (Beger et al. 2010). These data are now

being made available through our efforts, and other

biophysical models.

Geographic structure of connectivity

The geographic structure revealed in the dispersal

networks (Fig. 2) emerged from integrating repro-

ductive output estimates, larval life-history data, hab-

itat characteristics, and oceanic current data. The

connectivity matrices and networks enable unique

insights into the potential connectivity at levels rele-

vant to ecologists, marine managers, biogeographers,

and evolutionary ecologists. Clearly, the geographic

setting and physical oceanography are important in

shaping local to broad-scale patterns in connectivity.

This broad geographic variability reinforces the need

to be cautious when extrapolating connectivity esti-

mates from specific study sites or species to other

geographies as these patterns are highly dependent

on local hydrodynamics and species’ characteristics.

At broad scales, the predicted geographic patterns

in multispecies connectivity (dispersal corridors and

barriers) support those reported in other studies,

ranging from Indo-Pacific biophysical models (Kool

et al. 2011), population genetic studies (Barber et al.

2002), and comparative phylogeographic studies of

the area (Carpenter et al. 2011). Our model repro-

duces dominant dispersal corridors along major oce-

anic currents such as through the Solomon Islands,

along the Great Barrier Reef, through Micronesia,

and along the Indonesian throughflow (Nuryanto

and Kochzius 2009). At the same time, our model

predicts several additional potential dispersal corri-

dors, for example, one connecting Indonesia (and

Timor Leste) with the Kimberley Coast of Australia

and a strong corridor across the South China Sea

between Vietnam and the Spratly Islands. The loca-

tion of dispersal barriers (often semi-permeable) is

also readily apparent in the estimates of connectivity.

Although several of these multispecies barriers are

strongly supported by genetics data such as the

break between Papua New Guinea and reefs of

Papua, Indonesia (Barber et al. 2002; Nuryanto and

Kochzius 2009) and the isolation of reefs in Western

Australia (Underwood et al. 2009), quantitative eval-

uation of the location and strength of these, and

other, barriers requires a more detailed and

species-specific analysis.

Model caveats

There are several shortcomings of this model which

should be considered in future biophysical modeling.

First, increasing the spatial and temporal resolution

(and accuracy) of the modeled seascape and oceanic

currents would be important for improving estimates

of fine-scale population connectivity. The availability

of higher resolution data for oceans and coral reefs is

improving quickly. Second, including more dynamic

and spatially explicit functions of larval mortality

and postsettlement mortality into the model may

lead to more appropriate estimates of connectivity.

Third, vertical swimming needs to be implemented

for a more robust evaluation of its importance to

seascape-wide population connectivity. Fourth, im-

proving our understanding of species-specific larval

life-history characteristics through increased field-

based and laboratory research is critical. Currently,

our empirical data of PLD, larval behavior, and larval

mortality, for example, are from very few species,

individuals, and locations, and virtually unknown

for the vast majority of cases. Finally, for a robust

and quantitative test of predictions of connectivity, a

more detailed analysis is required. This would in-

volve aligning the assumptions of the empirical

data (e.g., genetic, chemical tags, otolith metrics)

with the assumptions and limitations of the




ownloaded from


population connectivity model, enabling the incon-

sistencies between the data and the model to be ex-

posed, thereby highlighting regions where the

connectivity model is wrong, where the empirical

data may be inconsistent, or where other processes

not presently considered may be influencing the ob-

served patterns.

Conclusions

Although perfect information does not exist for any

species or parameter, we are getting closer to identi-

fying the important parameters leading to local and

broad-scale marine population connectivity. These

location-specific predictions can be used to develop

sampling strategies to investigate species-specific dis-

persal patterns and quantify multi-species corridors

and barriers. Once validated, network analysis may

help reveal critical island stepping-stones, quantify

barriers, and highlight the emergent geographic

structure of multispecies connectivity (Treml et al.

2008; Kininmonth et al. 2010).

If scaled properly to the species of interest, these

estimates of connectivity may help inform conserva-

tion priorities from local to regional scales (Beger

et al. 2010). At a minimum, they highlight the role

of local protection in maintaining local to regional

connectivity through decreasing adult mortality and

increasing fecundity, thereby enhancing a local pop-

ulation’s reproductive output. Simply accommodat-

ing the strong geographic structure of marine

connectivity will assist in moving beyond simple

‘‘scales of connectivity’’ to more geographically ex-

plicit ‘‘ocean neighborhoods’’ (Palumbi 2004), pro-

ducing a framework that is more ecologically

meaningful, quantitative and testable, and ultimately

more informative (Treml and Halpin, in press).

The population connectivity code used in this

analysis is freely available through the Marine

Geospatial Ecology Tools software (Roberts et al.

2010). The model may be parameterized for other

regions or spatial scales to help illuminate the com-

plex roles of geography, oceanography, and life his-

tories in conferring population connectivity upon

marine species.

Acknowledgments

We thank Scott Burgess, Libby Liggins, and Jude

Keyse for helpful comments on earlier drafts.

Yarema Reshitnyk and Shaun Coutts provided gen-

erous analysis assistance. Comments from Harold

Heatwole and two anonymous reviewers greatly im-

proved this article.

Funding

Funding for this work was provided by the

Australian Research Council (DP0878306 to CR

and HPP; Federation Fellowship to HPP), the

World Wildlife Fund (Kathryn Fuller Postdoctoral

Research Fellowship to EAT), and the National

Science Foundation (NSF OISE�0730256 to PNH).

Participation in the Dispersal of Marine Organisms

symposium at the Society for Integrative and

Comparative Biology meeting was supported by the

American Microscopical Society, SICB Divisions of

Evolutionary Developmental Biology, Ecology &

Evolution, and Invertebrate Zoology, and by NSF

Grant IOS-1148884 to Sara M. Lindsay.

Supplementary Data

Supplementary Data are available at ICB online.

References

Almany G, Connolly S, Heath D, Hogan J, Jones G,

McCook L, Mills M, Pressey R, Williamson D. 2009.

Connectivity, biodiversity conservation and the design of

marine reserve networks for coral reefs. Coral Reefs

28:339–51.

Banks SC, Piggott MP, Williamson JE, Bove U, Holbrook NJ,

Beheregaray LB. 2007. Oceanic variability and coastal to-

pography shape genetic structure in a long-dispersing sea

urchin. Ecology 88:3055–64.

Barber PH, Palumbi SR, Erdmann MV, Moosa MK. 2002.

Sharp genetic breaks among populations of Haptosquilla

pulchella (Stomatopoda) indicate limits to larval transport:

patterns, causes, and consequences. Mol Ecol 11:659–74.

Baums IB, Paris CB, Cherubin LM. 2006. A bio-

oceanographic filter to larval dispersal in a reef-building

coral. Limnol Oceanogr 51:1969–81.

Beger M, Linke S, Watts M, Game E, Treml E, Ball I,

Possingham HP. 2010. Incorporating asymmetric connec-

tivity into spatial decision making for conservation.

Conserv Lett 3:359–68.

Black KP, Moran PJ. 1991. Influence of hydrodynamics on the

passive dispersal and initial recruitment of larvae of

Acanthaster-Planci (Echinodermata, Asteroidea) on the

great-barrier-reef. Marine Ecol Progr Ser 69:55–65.

Black KP, Moran PJ, Hammond LS. 1991. Numerical-models

show coral reefs can be self-seeding. Marine Ecol Progr Ser

74:1–11.

Botsford LW, Brumbaugh DR, Grimes C, Kellner JB,

Largier J, O’Farrell MR, Ralston S, Soulanille E,

Wespestad V. 2009a. Connectivity, sustainability, and

yield: bridging the gap between conventional fisheries man-

agement and marine protected areas. Rev Fish Biol

Fisheries 19:69–95.

Botsford LW, Hastings A, Gaines SD. 2001. Dependence of

sustainability on the configuration of marine reserves and

larval dispersal distance. Ecol Lett 4:144–50.

Botsford LW, White JW, Coffroth MA, Paris CB, Planes S,

Shearer TL, Thorrold SR, Jones GP. 2009b. Connectivity




ownloaded from


and resilience of coral reef metapopulations in marine pro-

tected areas: matching empirical efforts to predictive needs.

Coral Reefs 28:327–37.

Bowen BW, Bass AL, Muss A, Carlin J, Robertson DR. 2006.

Phylogeography of two Atlantic squirrelfishes (Family

Holocentridae): exploring links between pelagic larval du-

ration and population connectivity. Marine Biol

149:899–913.

Bradbury IR, Laurel B, Snelgrove PVR, Bentzen P,

Campana SE. 2008. Global patterns in marine dispersal es-

timates: the influence of geography, taxonomic category

and life history. Proc Roy Soc B Biol Sci 275:1803–9.

Carpenter KE, Barber PH, Crandall ED, Ablan-Lagman MCA,

Ambariyanto, Mahardika GN, Manjaji-Matsumoto BB,

Juinio-Menez MA, Santos MD, Starger CJ, Toha AHA.

2011. Comparative phylogeography of the coral triangle

and implications for marine management. J Marine Biol

2011:1–14.

Connolly SR, Baird AH. 2010. Estimating dispersal potential

for marine larvae: dynamic models applied to scleractinian

corals. Ecology 91:3572–83.

Cowen RK, Gawarkiewicz G, Pineda J, Thorrold SR,

Werner FE. 2007. Population connectivity in marine sys-

tems: an overview oceanography. Spec Issue Marine

Populat Conn 20:14–21.

Cowen RK, Lwiza KMM, Sponaugle S, Paris CB, Olson DB.

2000. Connectivity of marine populations: open or closed?

Science 287:857–9.

Cowen RK, Paris CB, Srinivasan A. 2006. Scaling of connec-

tivity in marine populations. Science 311:522–7.

Cowen RK, Sponaugle S. 2009. Larval dispersal and

marine population connectivity. Ann Rev Marine Sci

1:443–66.

Gaines S, Gaylord B, Gerber LR, Hastings A, Kinlan BP. 2007.

Connecting places: the ecological consequences of dispersal

in the sea. oceanography. Spec Issue Marine Populat Conn

20:90–9.

Gaines SD, Gaylord B, Largier JL. 2003. Avoiding cur-

rent oversights in marine reserve design. Ecol Appl

13:S32–46.

Gerlach G, Atema J, Kingsford MJ, Black KP, Miller-Sims V.

2007. Smelling home can prevent dispersal of reef fish

larvae. Proc Natl Acad Sci USA 104:858–63.

Graham EM, Baird AH, Connolly SR. 2008. Survival dynam-

ics of scleractinian coral larvae and implications for dis-

persal. Coral Reefs 27:529–39.

Groemping U. 2007. Estimators of relative importance in

linear regression based on variance decomposition. Am

Stat 61:139–47.

Hanski I. 1998. Metapopulation dynamics. Nature 396:41–9.

Hastings A, Botsford LW. 2006. Persistence of spatial popu-

lations depends on returning home. PNAS 103:6067–72.

Hedgecock D, Barber PH, Edmands S. 2007. Genetic

approaches to measuring connectivity. oceanography. Spec

Issue Marine Populat Conn 20:70–9.

Hughes TP, Baird AH, Bellwood DR, Card M, Connolly SR,

Folke C, Grosberg R, Hoegh-Guldberg O, Jackson JBC,

Kleypas J, Lough JM, Marshall P, Nystrom M,

Palumbi SR, Pandolfi JM, Rosen B, Roughgarden J. 2003.

Climate change, human impacts, and the resilience of coral

reefs. Science 301:929–33.

James MK, Armsworth PR, Mason LB, Bode L. 2002. The

structure of reef fish metapopulations: modelling larval dis-

persal and retention patterns. Proc Roy Soc Lond Ser B

Biol Sci 269:2079–86.

Jones G, Almany G, Russ G, Sale P, Steneck R, van Oppen M,

Willis B. 2009. Larval retention and connectivity among

populations of corals and reef fishes: history, advances

and challenges. Coral Reefs 28:307–25.

Jones GP, Milicich MJ, Emslie MJ, Lunow C. 1999.

Self-recruitment in a coral reef fish population. Nature

402:802–4.

Jones GP, Planes S, Thorrold SR. 2005. Coral reef fish larvae

settle close to home. Curr Biol 15:1314–8.

Jones GP, Srinivasan A, Almany GR. 2007. Population con-

nectivity and conservation of marine biodiversity.

Oceanography. Spec Issue Marine Populat Conn 20:100–11.

Kalnay E, Kanamitsu M, Kistler R, Collins W, Deaven D,

Gandin L, Iredell M, Saha S, White G, Woollen J, Zhu Y,

Chelliah M, Ebisuzaki W, Higgins W, Janowiak J, Mo KC,

Ropelewski C, Wang J, Leetmaa A, Reynolds R, Jenne R,

Joseph D. 1996. The NCEP/NCAR 40-year reanalysis proj-

ect. Bull Am Meteorological Soc 77:437–71.

Kininmonth S, van Oppen MJH, Possingham HP. 2010.

Determining the community structure of the coral

Seriatopora hystrix from hydrodynamic and genetic net-

works. Ecol Model 221:2870–80.

Kinlan BP, Gaines SD. 2003. Propagule dispersal in marine

and terrestrial environments: a community perspective.

Ecology 84:2007–20.

Kinlan BP, Gaines SD, Lester SE. 2005. Propagule dispersal

and the scales of marine community process. Diver Distr

11:139–48.

Kool JT, Paris CB, Barber PH, Cowen RK. 2011. Connectivity

and the development of population genetic structure in

Indo-West Pacific coral reef communities. Global Ecol

Biogeogr 20:695–706.

Largier JL. 2003. Considerations in estimating larval dispersal

distances from oceanographic data. Ecol Appl 13:S71–89.

Leis JM. 2002. Pacific coral-reef fishes: the implications of

behaviour and ecology of larvae for biodiversity and con-

servation, and a reassessment of the open population par-

adigm. Environ Biol Fishes 65:199–208.

Leis JM. 2007. Behaviour as input for modelling dispersal of

fish larvae: behaviour, biogeography, hydrodynamics, on-

togeny, physiology and phylogeny meet hydrography.

Marine Ecol Progr Ser 347:185–93.

Leis JM, Hay AC, Trnski T. 2006. In situ ontogeny of behav-

iour in pelagic larvae of three temperate, marine, demersal

fishes. Marine Biol 148:655–69.

Lester SE, Ruttenberg BI. 2005. The relationship between pe-

lagic larval duration and range size in tropical reef fishes: a

synthetic analysis. Proc Roy Soc B Biol Sci 272:585–91.

Lester SE, Ruttenberg BI, Gaines SD, Kinlan BP. 2007. The

relationship between dispersal ability and geographic range

size. Ecol Lett 10:745–58.

Levin LA. 2006. Recent progress in understanding larval dis-

persal: new directions and digressions. Integr Comp Biol

46:282–97.

Meyer CP, Geller JB, Paulay G. 2005. Fine scale endemism on

coral reefs: archipelagic differentiation in turbinid gastro-

pods. Evolution 59:113–25.




ownloaded from


Mitarai S, Siegel DA, Watson JR, Dong C, McWilliams JC.

2009. Quantifying connectivity in the coastal ocean with

application to the Southern California Bight. J Geophys

Res Oceans 114:C10026, doi:10.1029/2008JC005166.

Mora C, Treml EA, Roberts J, Crosby K, Roy D,

Tittensor DP. 2012. High connectivity among habitats pre-

cludes the relationship between dispersal and range size in

tropical reef fishes. Ecography 35:89–96.

Nuryanto A, Kochzius M. 2009. Highly restricted gene flow

and deep evolutionary lineages in the giant clam Tridacna

maxima. Coral Reefs 28:607–19.

Palumbi SR. 1994. Genetic-divergence, reproductive isolation,

and marine speciation. Ann Rev Ecol Syst 25:547–72.

Palumbi SR. 2004. Marine reserves and ocean neighborhoods:

the spatial scale of marine populations and their manage-

ment. Ann Rev Environ Resources 29:31–68.

Paris CB, Cowen RK. 2004. Direct evidence of a biophysical

retention mechanism for coral reef fish larvae. Limnol

Oceanogr 49:1964–79.

Paulay G, Meyer C. 2006. Dispersal and divergence across the

greatest ocean region: do larvae matter? Integr Comp Biol

46:269–81.

Planes S, Jones GP, Thorrold SR. 2009. Larval dispersal con-

nects fish populations in a network of marine protected

areas. Proc Natl Acad Sci USA 106:5693–97.

Possingham HP, Roughgarden J. 1990. Spatial population-

dynamics of a marine organism with a complex life-cycle.

Ecology 71:973–85.

Riginos C, Douglas KE, Jin Y, Shanahan DF, Treml EA. 2011.

Effects of geography and life history traits on genetic dif-

ferentiation in benthic marine fishes. Ecography 34:566–75.

Roberts JJ, Best BD, Dunn DC, Treml EA, Halpin PN. 2010.

Marine geospatial ecology tools: an integrated framework

for ecological geoprocessing with ArcGIS, Python, R,

MATLAB, and C plus. Environ Modelling Software

25:1197–207.

Sammarco PW, Andrews JC. 1988. Localized dispersal and

recruitment in great barrier-reef corals - the helix experi-

ment. Science 239:1422–4.

Shanks AL. 2009. Pelagic larval duration and dispersal dis-

tance revisited. Biol Bull 216:373–85.

Shanks AL, Grantham BA, Carr MH. 2003. Propagule dis-

persal distance and the size and spacing of marine reserves.

Ecol Appl 13:S159–69.

Siegel DA, Kinlan BP, Gaylord B, Gaines SD. 2003.

Lagrangian descriptions of marine larval dispersion.

Marine Ecol Progr Ser 260:83–96.

Siegel DA, Mitarai S, Costello CJ, Gaines SD, Kendall BE,

Warner RR, Winters KB. 2008. The stochastic nature of

larval connectivity among nearshore marine populations.

Proc Natl Acad Sci USA, 0802544105.

Slatkin M. 1993. Isolation by distance in equilibrium and

nonequilibrium populations. Evolution 47:264–79.

Smolarkiewicz PK. 1983. A simple positive definite advection

scheme with small implicit diffusion. Monthly Weather Rev

111:479–86.

Smolarkiewicz PK. 2006. Multidimensional positive definite

advection transport algorithm: an overview. Int J

Numerical Meth Fluids 50:1123–44.

Smolarkiewicz PK, Margolin LG. 1998. MPDATA: a finite-

difference solver for geophysical flows. J ComputatPhys

140:459–80.

Spalding M, Ravilious C, Green EP. 2001. World atlas of

Coral Reefs. Berkely: University of California Press.

Swearer SE, Forrester GE, Steele MA, Brooks AJ, Lea DW.

2003. Spatio-temporal and interspecific variation in otolith

trace-elemental fingerprints in a temperate estuarine fish

assemblage. Estuarine Coastal Shelf Sci 56:1111–23.

Swearer SE, Shima JS, Hellberg ME, Thorrold SR, Jones GP,

Robertson DR, Morgan SG, Selkoe KA, Ruiz GM,

Warner RR. 2002. Evidence of self-recruitment in demersal

marine populations. Bull Marine Sci 70:251–71.

Treml E, Halpin P, Urban D, Pratson L. 2008. Modeling pop-

ulation connectivity by ocean currents, a graph-theoretic

approach for marine conservation. Landscape Ecol

23:19–36.

Treml EA, Halpin PN. In press. Marine population connec-

tivity identifies ecological neighbors for conservation plan-

ning in the Coral Triangle. Conserv Lett. http://online

library.wiley.com/doi/10.1111/j.1755-263X.2012.00260.x/

abstract.

Underwood JN, Smith LD, Oppen MJHv, Gilmour JP. 2009.

Ecologically relevant dispersal of corals on isolated reefs:

implications for managing resilience. Ecol Appl 19:18–29.

Wang P, Song YT, Chao Y, Zhang HC. 2005. Parallel com-

putation of the Regional Ocean Modeling System. Int J

High Perform Comput Appl 19:375–85.

Wang X, Chao Y. 2004. Simulated Sea Surface Salinity vari-

ability in the tropical Pacific. Geophys Res Lett 31:L02302.

Weersing K, Toonen RJ. 2009. Population genetics, larval dis-

persal, and connectivity in marine systems. Marine Ecol

Progr Ser 393:1–12.

Werner FE, Cowen RK, Paris CB. 2007. Coupled biological

and physical models: present capabilities and necessary de-

velopments for future studies of population connectivity.

oceanography. Spec Issue Marine Populat Conn 20:54–69.

Wessel P, Smith WHF. 1996. A global, self-consistent, hierar-

chical, high-resolution shoreline database. J Geophys Res

Solid Earth 101:8741–3.

White C, Selkoe KA, Watson J, Siegel DA, Zacherl DC,

Toonen RJ. 2010. Ocean currents help explain population

genetic structure. Proc Roy Soc B Biol Sci 20:695–706.




ownloaded from


Documents

ReproductiveOutputandDurationofthePelagicLarvalStage ... · From the symposium ‘‘Dispersal of Marine Organisms’’ presented at the annual meeting of the Society for Integrative