Dynamic models applied to giant barnacle culture

ORI GINAL RESEARCH

Dynamic models applied to giant barnacle culture

Lorenzo I. Andrade • Daniel A. Lopez • Boris A. Lopez

Received: 18 March 2010 / Accepted: 30 January 2011 / Published online: 17 February 2011� Springer Science+Business Media B.V. 2011

Abstract In this study, a dynamic model is applied to giant barnacle (Austromegabalanuspsittacus) spat collection from artificial substrates located in the wild. Semi-industrial

culture of the giant barnacle, A. psittacus ‘‘picoroco’’ in southern Chile is an interesting

option for aquaculture diversification. The model establishes relationships between vari-

ables and carries out simulations to determine their effects on spat provision. The dynamic

hypothesis proposes that the number of giant barnacle spat obtained from the wild is

influenced by competent larval abundance (cyprids) over time, substrate availability and

mortality after larval settlement. In the conceptual model, 15 variables were selected and

related, establishing the polarities of each causal relationship and of each feedback loop.

The Stock & Flow diagram was undertaken using STELLA 9.0 simulation software.

Simulation tests were carried out to establish the consistency of the model using the

empirical background obtained from semi-industrial cultures in southern Chile. The model

establishes relative quantity of competent larvae, substrate area and the number of spat, as

key variables. Synchrony between level of cyprid abundance and location of artificial

substrates in the water is critical to achieve maximum collector efficiency. A difference of

less than 1 week out of synchronization produces significant losses (60–70%) in spat

production. When the deployment of collectors and maximum quantity of competent larvae

are synchronized, sensitivity analysis establishes an increase of up to 49.4% in the number

of spat, as a result of the collector area released by spat early mortality. The application of

dynamic models in aquaculture constitutes a useful tool for optimizing the process. The

model proposed enables us to understand the processes associated with obtaining seed from

the environment and can be applied to other similar processes, such as mytilid cultures.

Keywords Acorn barnacles � Aquaculture � Dynamics models � Southern Chile

L. I. Andrade � D. A. Lopez (&) � B. A. LopezDepartment of Aquaculture & Aquatic Resources, Universidad de Los Lagos,Avenida Fuchslocher 1305, Osorno, Chilee-mail: [email protected]

123

Aquacult Int (2011) 19:1047–1060DOI 10.1007/s10499-011-9421-4

Introduction

Although approximately twelve barnacle species can be considered as commercially

important, only the giant barnacle, Austromegabalanus psittacus, ‘‘picoroco’’, is cultured

(Freire and Garcia-Allut 2000; Lopez et al. 2010). Cultures of this species are currently

being carried out in suspended systems on a semi-industrial scale, in southern Chile, using

spat obtained in the wild (Lopez et al. 2005, 2010; Lopez 2008). Spat can be obtained from

the environment in different types of artificial substrates suspended from long lines. Mass

recruitment is concentrated in spring and the beginning of summer, at depths of between 4

and 6 m. Specimens reach an average commercial size of 3.5 cm carino-rostral length in

18 months; their height is around 10 cm and total weight is 150 g (Lopez et al. 2010).

Culture of the giant barnacle is economically feasible (Bedecarratz et al. pers. com.).

This species has traditionally been exploited on a small scale by artisanal fisheries,

reaching landings of between 200 and 600 tons annually over the past few years (SER-

NAPESCA 1998–2008). In acorn barnacle species, the competent cyprid larvae settle on

different types of substrates, after which the process of metamorphosis begins (Anderson

1996). Density-dependent effects produced during early growth, when larval settlement is

intense, are well known in various species of acorn barnacles (Bertness 1989; Hills and

Thomason 2003; Leslie 2005), but in A. psittacus they are subsequently attenuated by

structural base modifications (Lopez et al. 2007). The success of giant barnacle aquaculture

depends on the provision of spat that will, in turn, be determined by the competent larval

supply. Recruitment depends on oceanographic processes affecting the competent larval

supply and the substrate characteristics (Pineda 2007). Interaction that occurs at substrate

level after larval settlement, both between individuals of the same species and with

other species (Connell 1985; Walker 1995; Methratta and Petraitis 2008), also affects

recruitment.

Using dynamics system models, relationships between variables and feedback structure

maps can be established and simulation tools determined. These models have proved to be

an effective means of establishing how complex systems function, and, as such, are being

actively applied (Dacko 2010; Ahmed and Prashar 2010; Hsieh and Yuan 2010).

Dynamic models have been applied to other aquatic organisms in the areas of inver-

tebrate management (Bald and Borja 2002; Bald et al. 2006, 2009), population dynamics

(Marin 1997; Marin et al. 1998), and ecological systems (Costanza and Voinov 2001;

Costanza and Gottlieb 1998; Costanza et al. 1998; Marin et al. 2009). Nevertheless, they

have not been applied to aquaculture processes.

This study aims to formulate a dynamic model applicable to giant barnacle spat col-

lection from artificial substrates in the wild, by establishing relationships between variables

and carrying out simulations to determine their effect on spat provision. Information

obtained through simulations can be used to design strategies that optimize the process of

seed collection from the environment.

Methodologies

The model was designed and developed following stages established by Sterman (2000).

The limits and aims of the model were determined by simulating the dynamic behavior of

Austromegabalanus psittacus cyprid settlement in artificial collectors up to the spat phase,

after metamorphosis. Abundance of competent barnacle larvae in a given area can be

influenced by multiple variables (Pineda 2007; Pfeiffer-Herbert et al. 2007; Metaxas and

1048 Aquacult Int (2011) 19:1047–1060

123

Saunders 2009). For this reason, competent larval abundance was assumed as an exoge-

nous variable, to simulate varying behavior and observe its impact on giant barnacle spat

harvest.

The variables selected (n = 15) were Number of spat, Larval settlement, Spat mortality,Substrate area available, Substrate released, Occupied substrate, Substrate loss, Spatdensity, Fraction of area released by early mortality, Initial area for obtaining spat, Areaoccupied by each spat, Maximum spat density, Fraction of substrate area not used, Rel-ative quantity of competent larvae, and Fraction of spat mortality due to density-dependenteffects.

The 15 variables selected and defined (Table 1) were related in a conceptual model or

dynamic hypothesis, establishing the polarities of each causal relationship and of each

feedback loop.

The dynamic hypothesis proposes that the number of giant barnacle spat obtained is

influenced by the abundance of competent larvae over time, substrate availability, and

mortality after settlement as a density-dependent effect (depending on larval settlement

intensity). The aforementioned was expressed in a cause–effect diagram with the 15

selected variables. The types of causal relationships (? or -) and the feedback loops were

determined according to the standard procedures adopted in the formulation of dynamic

models (Senge 1990; Sterman 2000).

Table 1 Variables used (n = 15) in the model for obtaining giant barnacle spat from the wild and theirrespective descriptions

Variable Description

Number of spat Quantity of settled and metamorphosed giant barnacle larvae presentat a given moment in time

Larval settlement Quantity of competent giant barnacle larvae settled per day

Spat mortality Number of dead spat per day (number of dead spat per month/30 days)

Substrate area available Substrate area available for giant barnacle settlement at a givenmoment in time

Substrate released Substrate freed daily due to early spat mortality, which can bereused by new giant barnacle larval settlements

Occupied substrate Substrate area occupied daily by giant barnacle larval settlement

Substrate loss Substrate area lost daily due to occupation by organisms, other thancompetent giant barnacle larvae

Spat density Spat abundance with respect to initial substrate area

Fraction of area released by earlymortality

Fraction of substrate vacated due to early mortality that can bereused for larval settlement

Initial area for obtaining spat Initial substrate area for obtaining giant barnacle spat

Area occupied by each spat Average fixed value of area occupied by each settled larva

Maximum spat density Maximum spat density according to total competent giant barnaclelarvae that can potentially settle on the substrate.

Fraction of substrate area not used Fraction of substrate area not used by giant barnacle competentlarvae

Relative quantity of competent larvae Relative number of competent larvae that will potentially settle onthe available substrate, in relation to the possible maximum

Fraction of spat mortality due todensity-dependent effects

Spat mortality due to density-dependent effects, according to larvalsettlement density

Aquacult Int (2011) 19:1047–1060 1049

123

The dynamic hypothesis was used as a basis for developing the simulation model using

a Forrester Diagram or ‘‘Stock and Flow’’ Diagram (Forrester 1961; Morecroft 2007).

Formalization of the dynamic hypothesis in the Stock & Flow diagram required definition

and classification of the variables (Table 1). The Stock & Flow diagram was undertaken

using STELLA 9.0 simulation software, establishing initial equations and units of mea-

surement (Table 2).

Once the simulation software was formalized, simulation tests were undertaken to

establish the model’s consistency using, as the principal reference, the empirical back-

ground obtained from semi-industrial cultures in Metri Bay (41�360S; 72�430W), southern

Chile (Table 3). This is a wave-protected bay with average tidal variations of 7 m and an

average depth of 20 m. Three double lines, 100 m in length, were installed, from which

different types of collectors were suspended. An initial deployment was carried out

between July 2007 and December 2009. A second deployment was undertaken between

August 2009 and August 2010. Three types of collectors were used: polyvinyl chloride

(PVC) plates, 2 mm thick, measuring 70 9 20 cm; tubes of the same material, 4 mm thick

and 100 cm length 9 10 cm diameter and ‘‘bidin’’ plates, a felt type treated with tar,

70 cm length 9 30 cm diameter. All the collectors were positioned vertically in groups of

3, joined with polypropylene ropes of 3 mm, with a weight in the lower end to ensure the

vertical position was maintained.

Table 2 Definition of the model variables with their respective equations and units of measurement. Forsome parameters, initial values are indicated

Variable Equation Unit Variable type

Number of spat (ns) ns(t) = ns(t - dt) ?(ls - sm) dt

Individuals Level

dt = time step = 1;

t = time (days)

Larval settlement (ls) ls = md*sa*10,000 Individuals/day Flow

Spat mortality (sm) sm = fm*ns Individuals/day Flow

Substrate area available (sa) sa(t) = sa(t - dt) ?(us - os - sl) dt

m2 Level

dt = time step = 1;

t = time (days)

Substrate released (sr) sr = sm*ao*fr m2/day Flow

Occupied substrate (os) os = ao*ls m2/day Flow

Substrate loss (sl) sl = fs*sa m2/day Flow

Spat density (sd) sd = nc/si Individuals/m2 Auxiliary endogenous

Fraction of area released by earlymortality (fr)

fr = 0.5 Proportion Auxiliary exogenous

Initial area for obtaining spat (is) is = 1 m2 Auxiliary exogenous

Area occupied by each spat (ao) ao = 0.000001 m2/individual Auxiliary exogenous

Maximum spat density (md) md = f(cl) (**) Individuals/cm2 Auxiliary endogenous

Fraction of substrate area not used (fs) fs = f(cl) (**) Proportion Auxiliary endogenous

Relative quantity of competent larvae (cl) cl = f(TIME) Proportion Auxiliary endogenous

Fraction of spat mortality due todensity-dependent effects (fm)

fm = f(sd) (**) Proportion Auxiliary endogenous

(**) md is a direct function of cl; fs is an inverse function of cl; and fm is a direct function of sd

1050 Aquacult Int (2011) 19:1047–1060

123

The collectors were suspended at depths of 0, 4, and 6 m. Recruitment was evaluated

monthly. To this end, 3 collectors of each type were selected at depths of 0, 4, and 6 m, and

the total number of spat was determined, in addition to recording the number of dead

specimens.

Growth was only evaluated in the polyvinyl chloride plate and tube collectors, in

monthly determinations of 3 collectors of each type, at the three depths. Carino rostral

length, a density-independent measurement of body size, was measured monthly (Lopez

et al. 2007). Commercial size was taken as the average size of the specimens on the

national market. In a sample of 200 specimens, the average was 3.5 cm carino-rostral

length. In order to calculate the number of spat required to produce a harvest of 1 gross ton,

an average weight, on reaching the commercial size, of 119 g was used, including the shell.

Recruitment densities of 3 specimens/cm2 were used for the simulations. This value was

obtained in polyvinyl chloride collectors suspended from long lines, at a depth of 4 m.

Using this model, scenarios were defined according to the moment the artificial collectors

were placed in the water, with respect to a competent larval abundance curve. A period of

30 days was assumed for obtaining spat, considering an increasing phase of 15 days to reach

the maximum value and a similar decreasing phase, to reach the minimum reference value

(Table 4). This can be explained as the capacity of competent acorn barnacle larvae to attach

themselves to substrates becomes limited over time (Rittschof et al. 1984; Toonen and

Pawlik 1994) and because larval abundance depends on the period of maximum sexual

maturity. In the giant barnacle, this is concentrated over a period of 1 month and is reflected

in the recruitment period (Lopez 2008). It has been assumed that collectors are in optimum

conditions for larval settlement and have been previously conditioned for acquiring biofilm.

The model assumes that a relative quantity of competent larvae (cl) value equal to 1

represents the maximum possible quantity of competent larvae found in the wild and, when

there are no competent larvae in the water, it assumes a value of 0. Nevertheless, given that

recruitment has been verified almost all year round, a minimum cl value was fixed at 0.05.

The scenarios evaluated consider installation of artificial collectors in the water, to

obtain spat during 11 different days with respect to the competent larval abundance curves

defined in Table 4 (Table 5).

Number of spat (ns) 90 days after installation of artificial collectors in the water was the

indicator chosen to evaluate the best scenario for obtaining wild seed. Once the best result

scenario, in relation to spat obtained, was determined, a sensitivity analysis was applied to

variations in the variable Fraction of area released by early mortality (fr). The fr values

used were: 0 (null area released by any dead spat); 0.25 (25% area released due to

mortality); 0.5 (50% area released due to mortality); 0.75 (75% area released due to

mortality); and 1.0 (100% area unoccupied due to mortality).

Table 3 Reference data for giant barnacle cultures on the model production scale

Average spat density in the artificial collectors (No/cm2) 0.1–3

Size of the artificial collector (cm2) 10,000

Number of artificial collectors/long line 100

Early mortality (%) [90%

Number of spat required to produce 1 gross ton harvest 8,340

Range of time from spat to harvest (months) 18–24

The data were obtained from semi-industrial cultures in Metri Bay (41�; 360S; 72� 430W) developed between2008 and 2010. The artificial collectors are tubular structures of polyvinyl chloride, 100 cm in height and10 cm in diameter, suspended in groups of three units from the long lines

Aquacult Int (2011) 19:1047–1060 1051

123

Results

Giant barnacle spat abundance obtained from artificial collectors in the wild depends on

recruitment and substrate area available for spat collection. Recruitment depends on larval

settlement, that is, on the quantity of cyprids that adhere to the substrate. This, in turn, is

determined by the abundance of competent larvae in the water column. Other factors

affecting spat abundance that must be considered include the followings: larval settlement

of other species that compete for the substrate with the giant barnacle competent larvae,

and early spat mortality, produced, mainly, during metamorphosis (Fig. 1).

The dynamic hypothesis identified four negative and two positive feedback loops

(Fig. 2). The key variables are Relative quantity of competent larvae, Number of spat, and

Substrate area available. The variable that most influences the structure and general

behavior of the system is Relative quantity of competent larvae, given that it originates two

causal relationship pathways: one begins with a positive effect on the Maximum spatdensity, and the other has a negative effect on the Fraction of substrate area not used. The

first alternative assumes that in the absence of competent larvae, the substrate will be

occupied by other organisms, resulting in loss of area available for giant barnacle larval

settlement. Similarly, giant barnacle competent larval abundance prevents the substrate

being occupied by other organisms. Both causal relationship pathways are connected to

feedback loops and share common variables, of which the most important are Number ofspat and Substrate area available.

The negative loop between the variables Substrate loss and Substrate area availablemeans that increasing Substrate area available also increases the possibility of Substrateloss (occupation of area by other organisms), and, in turn, increasing Substrate lossdecreases Substrate area available. The negative loop between three variables, Substratearea available, Larval settlement and Occupied substrate, expresses that increasing Sub-strate area available also increases the probability of Larval settlement that leads to an

increase in the Occupied substrate, generating a decrease in the Substrate area availablefor giant barnacle larval settlement.

The negative loop between Number of spat and Spat mortality establishes that on

increasing the Number of spat, Spat mortality also increases and this decreases the Numberof spat. The negative loop between Number of spat, Spat density, and Fraction of spat

Table 4 Distribution of abun-dance or relative quantity ofcompetent larvae (cl) over aperiod of 30 days, where maxi-mum value is obtained 15 daysafter the initial increase of larvaein the water

After the maximum is reached(cl = 1), a descent, inverse to theascent behavior, is assumed

Difference in days with respect to the maximum cl value (thenegative values are days preceding and the positive values aredays following, maximum cl value)

cl

-15 0.050

-12 0.075

-9 0.100

-6 0.200

-3 0.300

0 1.000

?3 0.300

?6 0.200

?9 0.100

?12 0.075

?15 0.050

1052 Aquacult Int (2011) 19:1047–1060

123

mortality due to density-dependent effects, indicates that when Number of spat increases,

an increase in Spat density generates an increase in Fraction of spat mortality due todensity-dependent effects and subsequently Spat mortality, which determines a decrease in

the Number of spat.The positive loop between the variables Substrate area available, Larval settlement,

Number of spat, Spat mortality and Substrate released establishes that the greater the

Substrate area available, the greater the probability of Larval settlement increasing, which

leads to an increase in the Number of spat; the density-dependent effects on Spat mortalityproduce an increase in Substrate released, thus generating greater availability of Substratearea available.

Table 5 Scenarios for evaluating the location of artificial collectors in the water for settlement of com-petent larvae, according to the abundance curve of competent larvae

Difference in days with respect to the maximum cl value

(- ): days prior to the maximum cl value

(+): days subsequent to the maximum cl value

Scenario

No

-15 -12 -9 -6 -3 0 +3 +6 +9 +12 +15

1 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

2 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

3 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

4 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

5 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

6 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

7 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

8 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

9 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

10 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

11 0.05 0.075 0.1 0.2 0.3 1 0.3 0.2 0.1 0.075 0.05

The black boxes correspond to the moment collectors are placed in the water. It was assumed that early spatmortality liberates 50% of the space occupied (fr = 0.5), making this space available for new giant barnaclelarval settlements

Aquacult Int (2011) 19:1047–1060 1053

123

The positive loop between the variables Substrate area available, Larval settlement,Number of spat, Spat density, Fraction of spat mortality due to density-dependent effects,

Spat mortality, and Substrate released indicates that the greater the Substrate areaavailable, the greater the probability of Larval settlement, producing an increase in the

Number of spat. When Spat density is higher, the Fraction of spat mortality due to density-

dependent effects is greater, and, as a result, Spat mortality increases, generating Substratereleased and consequently more Substrate area available.

Formalization of the model considered two level or stock variables, Substrate areaavailable and Number of spat, that depend directly on their respective flows, both incoming

Fig. 1 Stages of giant barnacle spat collection from the wild and the relationship between them

Fig. 2 Dynamic hypothesis for obtaining giant barnacle spat from the wild, based on the settlement ofcompetent larvae on artificial substrates. The causal relationships between variables are expressed with theircorresponding polarity (? or -); B negative or balanced feedback loop, R positive or reinforced feedbackloop

1054 Aquacult Int (2011) 19:1047–1060

123

and outgoing. These stock variables related to the auxiliary variables or parameters,

determine the behavior of the model. It is assumed that in the absence of competent

barnacle larvae (cyprids), the substrate is lost, due to occupation by competitor organisms.

Number of spat is influenced by two flows, one incoming (Larval settlement) and

another outgoing (Spat mortality). Variations over time in Substrate area available, depend

on three flows, one incoming (Substrate released) and two outgoing (Substrate loss and

Occupied substrate) (Fig. 3).

Simulations of eleven scenarios in terms of the moment in time when artificial col-

lectors are placed in the water, in relation to the period of maximum competent larval

supply, reveal that the best results for obtaining spat occur when the collectors are posi-

tioned in synchronization with the maximum abundance of competent larvae (scenario 6);

while the worst results (\15%) were obtained with a 15-day period out of synchronization

between installation of collectors and maximum larval abundance in the plankton (sce-

narios 1 and 11) (Table 6).

Using the best scenario result as a reference, it can be observed that a difference of

15 days in deployment of the collectors implies a considerable decrease in number of spat

obtained. Specifically, loss of opportunity in terms of obtaining spat would be close to

90%. In the case of a difference of 3 days, loss of opportunity would be between 11.5%

(-3 days) and 22.4% (?3 days).

Sensitivity analysis of the best scenario model shows the effect that substrate released as

a result of early spat mortality would have on Number of spat 90 days after positioning the

collectors (Table 7). A difference of 33.1%, between null substrate released and maximum

substrate released as a result of Spat mortality, was recorded.

Fig. 3 Formalization of the model for obtaining giant barnacle spat from the wild on artificial substrates,based on competent larval settlement, using STELLA 9.0 software (for abbreviations, see Table 2)

Aquacult Int (2011) 19:1047–1060 1055

123

Discussion

Culture of the giant barnacle, Austromegabalanus psittacus, ‘‘picoroco’’, in southern Chile

is an interesting option for Aquaculture diversification in the country (Lopez 2008; Lopez

et al. 2010). Semi-industrial cultures have been developed from spat collected from dif-

ferent types of artificial substrates located in the wild (Lopez 2008; Lopez et al. 2010). This

strategy has been used successfully for many years in commercial cultures of mytilids in

southern Chile (Winter et al. 1984; Navarro and Gutierrez 1990). Aquaculture production

of the ‘‘mussel’’ Mytilus chilensis has fluctuated over the last few years between 16.000

and 187.000 ton/year (Sernapesca Chile, 1998–2008); using the same strategy, commercial

cultures of the ‘‘ribbed mussel’’ Aulacomya ater and the ‘‘giant mussel’’ Choromytiluschorus have also been undertaken on a smaller volume. Sustainability of these cultures

depends on the quantity and predictability of collecting spat from artificial collectors. The

main advantage is the reduced cost compared to larvae and spat production in hatcheries.

To this end, this strategy has also been tested in another barnacle, Megabalanus azoricus,

Table 6 Quantity of giant barnacle spat in artificial collectors

Scenarios (No) Moment in time when collectorsare positioned in the water, withrespect to maximum cl value(-numbers are days prior toand ?numbers are days after)

Maximum initiallevel of spatobtained(maximum clper collector)

Level of spat obtained(90 days afterpositioning collectorsin the water)per collector

%achievementwith respectto the bestscenario

1 -15 6,006 124 10.4

2 -12 10,142 231 19.4

3 -9 18,503 445 37.5

4 -6 45,567 824 69.4

5 -3 71,854 1,051 88.5

6 0 100,001 1,188 100.0

7 ?3 47,627 922 77.6

8 ?6 30,173 671 56.5

9 ?9 13,790 370 31.1

10 ?12 9,213 259 21.8

11 ?15 5,783 170 14.3

Efficiency when placement of artificial collectors in the water coincides with maximum quantity of com-petent larvae (cl = 1)

Table 7 Sensitivity analysis of scenario 6 in Table 5

fr ns

0.00 967

0.25 1,073

0.50 1,188

0.75 1,311

1.00 1,445

Number of spat (ns) obtained in 1 long line (100 m) using 1 m2\ collectors, considering 5 different values forthe parameter Fraction of area released by early mortality (fr)

1056 Aquacult Int (2011) 19:1047–1060

123

in the Azores (Pham et al. 2008). The principle disadvantages are related to dependence on

highly variable environmental factors. Recruitment of acorn barnacle spat has been studied

intensely in fouling communities and in ecological processes of intertidal and subtidal

populations. Knowledge accumulated could be applied to spat collection in acorn barnacle

Aquaculture practices. It has been established that, in addition to substrate characteristics,

spat collection depends on coastal circulation processes and local geomorphological fac-

tors (Pineda 2000; McCulloch and Shanks 2003; Lagos et al. 2008). Competent larval

supplies and availability of adequate substrates are factors that influence larval settlement.

Competent larval supplies depend on various oceanographic processes that operate on

different spatial and temporal scales (Roughgarden et al. 1988; Dudas et al. 2009).

Available substrate varies spatially and also according to physical factors that affect larval

settlement (Anderson and Underwood 1994; Berntsson and Jonsson 2003). Once larval

settlement occurs in barnacles, both interspecific (Connell 1985) and intraspecific (Pineda

et al. 2009) interactions are produced that affect recruitment levels. Occupation of the

substrate by other species reduces substrate available for barnacle spat; nevertheless,

complex, interrelated processes also occur, either facilitating or inhibiting settlement

among these non-barnacle colonizers (Farrell 1991). In view of these antecedents, the

model proposed for obtaining giant barnacle spat from the wild establishes relative

quantity of competent larvae, substrate area and number of spat, as key variables. The latter

variable is critical to giant barnacle aquaculture. Synchrony between maximum level of

competent larvae and location of artificial substrates in the water is a process that is critical

to achieving maximum collector efficiency. Although A. psittacus recruits throughout the

year, highest recruitment is concentrated over a short period of time of approximately

1 month (Lopez 2008). On the other hand, if barnacle cyprid larvae do not feed, they also

have a limited number of days to settle successfully (Tremblay et al. 2007). The model

establishes that differences of 15 days between maximum supply of competent larvae and

deployment of collectors in the water occasioned a reduction of almost 90% in spat

obtained. A period out of synchronization of less than 1 week also produces significant

losses (60–70%). The results obtained indicate the importance of carrying out a daily

census of competent larvae. In intertidal and subtidal acorn barnacles, there is considerable

evidence of high temporal variation in reproduction and larval settlement (Hawkins

and Hartnoll 1982; Miron et al. 1996; Berntsson et al. 2000; Jenkins et al. 2000; Dionisio

et al. 2007; Jacinto and Cruz 2008; Satheesh and Godwin 2008; Kado et al. 2009), ranging

from interannual variations generated by periodic events to differences between tidal

cycles.

In contrast, intense larval settlements in barnacles can generate high early mortality

among recently settled individuals, as a result of density-dependent processes (Bertness

1989; Hills and Thomason 2003). In A. psittacus, suppression and dominance effects

subsequent to larval settlement have been verified, independent of the period of year

(Lopez et al. 2007). Substrate vacated as a result of early spat mortality will eventually

permit an increase in the number of spat through new larval settlement pulses. The model

establishes that this new source of recruits could significantly increase number of spat.

Nevertheless, this depends on post-settlement interactions with other species and the pre-

occupation of substrate, as has been verified in crustacean communities (Dean and Hurd

1980; Anderson and Underwood 1994; Durr and Wahl 2004).

The application of dynamic models in aquaculture constitutes a useful tool for opti-

mizing processes, based on the definition of relevant variables and simulations of the

results of their interactions. The results related to spat collection from the wild in giant

barnacle cultures indicate that planning operations, together with information on competent

Aquacult Int (2011) 19:1047–1060 1057

123

larval supply, are important aspects for ensuring efficiency in production technologies.

Success of these cultures depends, to a large extent, on the efficiency of the spat collection

process. The model can also be applied to seed collection in other species, whose culture

functions similarly, as is the case of the mytilid species.

Acknowledgments We are grateful to Fondef (projects D03I1116 and D07I1042) for financing the giantbarnacle cultures. Similarly, the facilities provided by the Aquaculture and Marine Science Centre of theUniversidad de Los Lagos in Metri Bay are much appreciated. The collaboration of Sergio E. Arriagada,Alexis V. Santibanez, Mauricio O. Pineda, Oscar A. Mora, and Jose M. Uribe in the culture activities and ofSusan Angus in the translation of the manuscript is also gratefully acknowledged. Finally, thanks are due tothe anonymous reviewers for their suggestions.

References

Ahmed S, Prashar D (2010) Evaluating municipal water conservation policies using a dynamic simulationmodel. Water Res Man 24(13):3371–3395

Anderson D (1996) Barnacles—structure, function, development and evolution. Chapman & Hall, LondonAnderson M, Underwood A (1994) Effects of substratum on the recruitment and development of an

intertidal estuarine fouling assemblage. J Exp Mar Biol Ecol 184:217–236Bald J, Borja A (2002) Modelling the management of clam (Ruditapes decussatus) exploitation in the

Plentzia Estuary (Basque Country, Northern Spain). In: Proceedings of the XX international confer-ence of the system dynamics society, Palermo, Italy, July 28–August 1, 2002

Bald J, Borja A, Muxica I (2006) A system dynamics model for the management of the gooseneck barnacle(Pollicipes pollicipes) in the marine reserve of Gaztelugatxe (Northern Spain). Ecol Model194:306–315

Bald J, Sinquin A, Borja A, Caill-Milly N, Duclercq B, Dang C, de Montaudouin X (2009) A systemdynamics model for the management of the Manila clam, Ruditapes philippinarum (Adams and Reeve,1850) in the Bay of Arcachon (France). Ecol Model 220:2828–2837

Berntsson K, Jonsson P (2003) Temporal and spatial patterns in recruitment and succession of a temperatemarine fouling assemblage: a comparison of static panels and boat hulls during the boating season.Biofouling 19(3):187–195

Berntsson KM, Jonsson PR, Lejhall M, Gatenholm P (2000) Analysis of behavioural rejection of micro-textured surfaces and implications for recruitment by the barnacle Balanus improvisus. J Exp Mar BiolEcol 251(1):59–83

Bertness M (1989) Positive and negative density-dependent mortality and the population structure ofSemibalanus balanoides in a sheltered bay habitat. Ecology 70:257–268

Connell J (1985) The consequence of variation in initial settlement versus post-settlement mortality in rockyintertidal communities. J Exp Mar Biol Ecol 93:11–45

Costanza R, Gottlieb S (1998) Modelling ecological and economic systems with STELLA: part II. EcolModel 112(2–3):81–84

Costanza R, Voinov A (2001) Modeling ecological and economic systems with STELLA. Part III. EcolModel 143(1–2):1–7

Costanza R, Duplisea D, Kautsky U (1998) Ecological modelling and economic systems with Stella-Introduction. Ecol Model 110(1):1–4

Dacko M (2010) Systems Dynamics in modelling sustainable management of the environment and itsresources. Polish J Env Stud 19(4):699–705

Dean T, Hurd L (1980) Development in an estuarine fouling community: the influence of early colonists onlater arrivals. Oceanologıa 46(3):295–301

Dionisio M, Rodriguez A, Costa A (2007) Reproductive biology of Megabalanus azoricus (Pilsby), theAzorean barnacle. Inv Rep Dev 50(3):155–162

Dudas S, Rilov G, Tyburczy J, Menge B (2009) Linking larval abundance, onshore supply and settlementusing instantaneous versus integrated methods. Mar Ecol Prog Ser 387:81–95

Durr S, Wahl M (2004) Isolated and combined impacts of blue mussels (Mytilus edulis) and barnacles(Balanus improvisus) on structure and diversity of a fouling community. J Exp Mar Biol Ecol306(2):181–195

Farrell T (1991) Model and mechanisms of succession: an example from a rocky intertidal community. EcolMonogr 61(1):95–113

1058 Aquacult Int (2011) 19:1047–1060

123

Forrester J (1961) Industrial dynamics. MIT Press, CambridgeFreire J, Garcia-Allut A (2000) Socioeconomic and biological causes of management failures in European

artisanal fisheries: the case of Galicia (NW Spain). Mar Policy 24(5):375–384Hawkins S, Hartnoll R (1982) Settlement patterns of Semibalanus balanoides (L) in the Isle of Man

(1977–1981). J Exp Mar Biol Ecol 62:271–283Hills J, Thomason J (2003) Recruitment density can determine adult morphology and fecundity in the

barnacle, Semibalanus balanoides. Biofouling 19(3):205–213Hsieh YH, Yuan ST (2010) Modelling service experience design processes with customer expectation

management. A system dynamics perspective. Kybernetes 39(7):1128–1144Jacinto D, Cruz T (2008) Tidal settlement of the intertidal barnacles Chthamalus spp., in SW Portugal:

interaction between diel and semi-lunar cycles. Mar Ecol Prog Ser 366:29–135Jenkins S, Aberg P, Cervin G, Coleman R, Delany J, Della Santina P, Hawkins S, LaCroix E, Myers A,

Lindegarth M, Power A, Roberts M, Hartnoll R (2000) Spatial and temporal variation in settlement andrecruitment of the intertidal barnacle Semibalanus balanoides (L.) (Crustacea: Cirripedia) over aEuropean scale. J Exp Mar Biol Ecol 243:209–225

Kado R, Suzuki J, Suzuki Y, Nanba N, Ogawa H (2009) Reproductive cycle, larval distribution, settlementand initial growth of the barnacle Balanus rostratus in Mutsu bay, northern Japan. Nippon SuisanGakkaishi 75(3):432–442

Lagos N, Castilla J, Broitman B (2008) Spatial environment correlations of intertidal recruitment: a testusing barnacles in northern Chile. Ecol Monogr 78(2):245–261

Leslie H (2005) Positive intraspecific effects trump negative effects in high-density barnacle hummocks.Ecology 86(10):2716–2725

Lopez D (2008) Giant barnacle ‘‘picoroco’’ culture in Chile. In: Pham CK, Higgins RM, De Girolamo M,Isidro E (eds) Acta of the international workshop: developing a sustainable aquaculture industry in theAzores. Arquipelagos. Life and Mar Sci, pp 56–57, Supplement 7: xiii ?81 pp

Lopez D, Lopez B, Gonzalez M, Arriagada S (2005) Aquaculture diversification in Chile: potential culturegiant barnacles. Glob Aquac Advocate 8(2):73–74

Lopez D, Lopez B, Burgos I, Arriagada S, Gonzalez M (2007) Consequences of base modification inhummocks of barnacle Austromegabalanus psittacus. N Z J Mar Freshw Res 41(3):291–298

Lopez D, Lopez B, Pham C, Isidro C, De Girolamo M (2010) Barnacle culture: background, potential andchallenges. Aquac Res 41:e367–e375

Marin V (1997) A simple-biology, stage-structured population model of the spring dynamics of Calanuschilensis at Mejillones del Sur Bay, Chile. Ecol Model 105:65–82

Marin V, Grand V, Droven N (1998) Simulation of population dynamics of the parasite Haematoloechuscoloradensis in its three host species: effects of environmental temperature and precipitation. EcolModel 105:185–211

Marin V, Tironi A, Delgado L, Contreras M, Novoa F, Torres-Gomez M, Garreaud R, Vila I, Serey I (2009)On the sudden disappearance of Egeria densa from a Ramsar Wetland site of Southern Chile: aclimatic event trigger model. Ecol Model 220(15):1752–1763

McCulloch A, Shanks A (2003) Topographically generated fronts, very nearshore oceanography andthe distribution and settlement of mussel larvae and barnacle cyprids. J Plankton Res 25:1427–1439

Metaxas A, Saunders M (2009) Quantifying the ‘‘Bio-’’ components in biophysical models of larvaltransport in marine benthic invertebrates: advances and pitfalls. Biol Bull 216(3):257–272

Methratta F, Petraitis P (2008) Propagation of scale-dependent effects from recruits to adults in barnaclesand seaweeds. Ecology 89(11):3128–3137

Miron G, Bourget E, Archambault P (1996) Scale of observations and distribution of adult conspecific: theirinfluence in assessing and active settlement mechanism in the barnacle Balanus crenatus (Bruguiere).J Exp Mar Biol Ecol 201:137–158

Morecroft J (2007) Strategic modelling and business dynamics. A feedback system approach. Wiley,Chichester

Navarro J, Gutierrez A (1990) Cultivo de moluscos en Chile: sinopsis sobre el estado actual del cultivo demitılidos en Chile. In: Hernandez A (ed) Cultivo de moluscos en America Latina. Bogota, Colombia,pp 189–210

Pfeiffer-Herbert A, McManus M, Raimondi P, Chao Y, Chai I (2007) Dispersal of barnacle larvae along thecentral California coast: a modeling study. Limnol Oceanogr 52(4):1559–1569

Pham CK, Higgins RM, De Girolamo M, Isidro M (eds) (2008) Acta of the international workshop:developing a sustainable aquaculture industry in the Azores. Arquipelago–Life and Marine SciencesSupplement 7. Universidade dos Acores, Azores

Aquacult Int (2011) 19:1047–1060 1059

123

Pineda J (2000) Linking larval settlement to larval transport: assumption, potentials and pitfalls. Ocean EastPac 1:84–105

Pineda J (2007) Larval transport and dispersal in the coastal ocean and consequences for populationconnectivity. Oceanography 20(3):22–34

Pineda J, Reyns N, Starczack V (2009) Complexity and simplification in understanding recruitment inbenthic populations. Popul Ecol 51(1):17–32

Rittschof D, Branscomb E, Costlow J (1984) Settlement and behavior in relation to flow and surface inlarval barnacles, Balanus amphitrite Darwin. J Exp Mar Biol Ecol 82:131–146

Roughgarden J, Gaines S, Possinghan H (1988) Recruitment dynamics in complex life cycles. Science241:1460–1466

Satheesh S, Godwin S (2008) Seasonal variability in the recruitment of macrofouling community in Kun-dankulam waters east coast of India. Estuar Coastal Shelf S 79(3):518–524

Senge P (1990) The fifth discipline: the art and practice of the learning organization. Doubleday, New YorkSERNAPESCA (1998–2008). Anuario Estadıstico de Pesca. Servicio Nacional de Pesca. Subsecretaria de

Pesca, Ministerio de Economıa, Gobierno de Chile http://www.sernapesca.cl/index.php?option=com_remository&Itemid=54&func=select&id=2. Cited 20 Nov 2009

Sterman J (2000) Business dynamics: systems thinking and modeling for a complex world. Irwin McGraw-Hill, Boston

Toonen R, Pawlik J (1994) Foundations of gregariousness. Nature 370:511–512Tremblay R, Olivier F, Bourget E, Rittshof D (2007) Physiological condition of Balanus amphitrite cyprid

larvae determines habitat selection success. Mar Ecol Prog Ser 340:1–8Walker G (1995) Larval settlement: historical and future perspective. In: Schram FR, Høeg JT (eds) New

frontiers in Barnacle evolution. A.A. Balkema, Rotterdam, pp 69–85Winter J, Toro J, Navarro J, Valenzuela G, Chaparro O (1984) Recent developments status, and prospects of

molluscan aquaculture on the Pacific Coast of South America. Aquaculture 39(1–4):95–134

1060 Aquacult Int (2011) 19:1047–1060

123