Ecological Modelling 105 (1997) 65 82

A simple-biology, stage-structured population model of the

spring dynamics of Calanus chilensis at Mejillones del Sur Bay,
Chile
Victor H. Marn *
Depto. Ciencias Ecologicas, Facultad de Ciencias, Uni6ersidad de Chile, Casilla 653, Santiago, Chile
Accepted 29 July 1997

Abstract
The population dynamics of Calanus chilensis (Copepoda: Crustacea) was analyzed using a mixture of field data
and mathematical models. Field data were obtained in the Humboldt Current upwelling area, at Mejillones del Sur
Bay (northern Chilean coast) during the spring of 1990 and 1991. Those data were used to set the parameter values
and functions of a stage-structured population model (SSPM). The model was built and run with STELLA-II version
3.07, an interactive, iconographic modeling software. The results show that C. chilensis has a generation time of 20
days, and that its population dynamics in the spring is controlled by upwelling events, which affect both its growth
rate (food dependence) and its local population size (advection). 1997 Elsevier Science B.V.
Keywords: Calanus chilensis, upwelling, population model; Chilean coast; Copepods; STELLA II

1. Introduction

1.1. The problem

Population regulation in marine organisms is a
recurrent issue in biological oceanography,
fisheries ecology and conservation (Laevastu and
Larkins, 1981; Sinclair, 1988; Angel, 1994). Sinclair (1988) has proposed the member/vagrant
hypothesis, which emphasizes the effect of ocean
* E-mail: vmarin@antar.ciencias.uchile.cl.

circulation in the life cycle of marine animals. The

hypothesis proposes that the populations of species with complex life cycles are regulated, to a
large degree, by physical oceanographic processes.
The fact that holopelagic species are not ubiquitous has been used to suggest that their distribution patterns are actively maintained (Angel,
1994). The maintenance mechanism implies life
cycles finely tuned to water circulation patterns
(Verheye et al., 1991; Angel, 1994). Calanoid
copepods are good examples of animals with life
cycles tuned to the physical characteristics of

0304-3800/97/$17.00 1997 Elsevier Science B.V. All rights reserved.

PII S 0 3 0 4 - 3 8 0 0 ( 9 7 ) 0 0 1 3 9 - 7

66

V.H. Marn / Ecological Modelling 105 (1997) 6582

marine ecosystems (Marn, 1986, 1988; Conover,

1988). Different species, however, seem to have
evolved different successful solutions for this tuning process for a given ecosystem. For antarctic
calanoid copepods, Marn (1988) has shown that
three abundant species (Calanus propinquus,
Calanoides acutus and Rhincalanus gigas) have
different life cycles. Two of them (C. acutus and
R. gigas) show strong vertical seasonal (ontogenetic) migration, two (C. acutus and C, propinquus) have a 1-year life cycle, and the other has a
2-year cycle. All three species are endemic of the
Antarctic Ocean and do not occur north of the
Polar Front (Antarctic Convergence).
In some coastal areas, such as eastern boundary
currents, upwelling is the dominant advective factor that explains much of the spatial and temporal
changes in the ecosystem (Marn et al., 1993;
Strub et al., in press). The Humboldt Current,
which runs parallel to the coasts of Chile and
Peru, is one of the large eastern boundary current
systems (Shermann, 1995). Calanus chilensis is one
of the most abundant, endemic, calanoid copepods of the upwelling areas of this current and a
key herbivore (Gonzalez, 1993; Escribano and
Rodriguez, 1994; Marn et al., 1994). From the
standpoint of its general distribution, this species
can be described as neritic and epipelagic. That is,
it is abundant year round in the coastal zone
( B10 km offshore; Gonzalez, 1993; Escribano
and Rodriguez, 1994).
One of the mechanisms that explains the
maintenance of calanoids in upwelling areas is the
interaction between ontogenetic vertical migration
and the vertically separated onshore offshore water flows (Smith, 1984). In this case, adults are
carried offshore and sink to the onshore (deeper)
flow that then brings the eggs back to the upwelling area. Other authors (e.g. Verheye et al.,
1991) suggest a behavioral mechanism involving
diurnal vertical migration in relation to flow reversal.
Analysis of the distribution of C. chilensis in the
Chilean coast shows that this species does not
perform ontogenetic vertical migration (Gonzalez,
1993). However, the duration of the life cycle of
C. chilensis is uncertain and only few data are
available on its biology. In this situation, there are

two general approaches to propose for further

research: (1) keep collecting biological data on
topics related to the life cycle of the species,
together with current measurements; or (2) assemble the scanty information on both the biology
and physics of the system and construct a preliminary model that may, hopefully, shed some light
on what aspects of the research should be emphasized.

1.2. The approach

The philosophy put forward by Powell et al.
(1990) for the Global Ocean Ecosystems Dynamics Program (GLOBEC) has been followed: the
first step (in the GLOBEC Program) should be a
modeling effort to determine how well we are able
to put together our present knowledge of physical
oceanography with the known population biology
of marine organisms that have numerous, distinct,
planktonic life stages. However, the development,
calibration, and validation of a model against
existing data are complex activities that most of
the time separate the scientists into modelers and
empiricists (Rykiel, 1996). Thus, the successful
use of the GLOBEC approach requires a bridge
between these two groups of scientists. In my
opinion, easy-to-learn iconographic modeling
software such as STELLA II represents an important component of that bridge (Hannon and
Ruth, 1994). STELLA II substantially decreases
the amount of time that the researcher needs to
spend in computer programming, thus making
modeling available for a wider spectrum of people. Furthermore, even if a biologist is keen to use
the GLOBEC approach, the generation of threedimensional circulation models with incorporated
biology requires a powerful UNIX-based computer, with a minimum of 256 megabytes (MB) of
RAM (A. Sepulveda, EIMS Project, Dept. Cs.
Ecologicas, Universidad de Chile, personal communication). These classes of computers may not
be financially obtainable for many scientists working in developing nations. Simple software running on Windows-based personal computers is
then an alternative to start the modeling approach. This paper shows an example of how to
use STELLA II to build a simple-biology, popula-

V.H. Marn / Ecological Modelling 105 (1997) 6582

67

Fig. 1. Geographic location of oceanographic stations at Mejillones del Sur Bay (MB). Station B was located inside the Bay, station
P was located near the point where upwelling occurs.

tion dynamics model (Nival and Carlotti, 1993).

Simple biology means that development time is
considered constant, egg production is forced by
phytoplankton (chlorophyll-a) concentration
only, and mortality is a constant fraction of the
population. The specific example is a stage-structured population growth model (GLOBEC, 1993)
of C. chilensis. Using a mixture of field data
analysis and mathematical modeling, the interactions between coastal upwelling and the life cycle
of C. chilensis (Copepoda: Calanoida) are discussed.

2.1.2. En6ironmental data

Data on wind stress, temperature and phytoplankton (chlorophyll-a) were collected as part of
a research program on Mejillones Bay, Chile.
Most of that information, at least for the 1990
period, has been published elsewhere (Marn et
al., 1993). There a full account of the methods
used in measuring those variables is given. For the
sake of completeness, only a brief description of
them is given below.
Alongshore wind stress (ty) was calculated from
local wind records according to the formula:
ty = raCd(VyVy)

(1)

where: ra = air density (=1.22 kg m ), Cd =

empirical drag coefficient (= 0.0013) and Vy =
alongshore wind velocity. The value ty was used
to estimate the Ekman Transport (ME), which is a
parametric way to analyze the upwelling process,
according to the formula (Bowden, 1983):
3

2. Methods

2.1. Field data

2.1.1. Zooplankton samples
Vertically integrated (bottom, 40 m, to surface)
zooplankton samples were collected, with a 200
mm mesh open net, at two stations in the Mejillones del Sur Bay area (2305%S, 7020%W) during
SeptemberOctober 1990 and 1991 (Fig. 1). An
average of 10 samples was collected at each station each year. Individuals of C. chilensis were
counted and identified to adults, copepodite stage
and nauplius 6. Eggs or earlier naupliar stages
were not counted because the mesh size was too
coarse for their reliable quantification.

ME =

ty
f

(2)

where f=Coriolis parameter.

Sea surface temperature was measured at station B, in Mejillones Bay, by means of reversing
thermometers. Finally, water samples from at
least four levels within the euphotic zone, were
collected with Niskin bottles at Station B. Chlorophyll-a was determined from each sample using a
spectrophotometric technique (Strickland and
Parsons, 1972).

68

V.H. Marn / Ecological Modelling 105 (1997) 6582

Fig. 2. Diagram of the stage-structured population model (SSPM) built using STELLA II software. State variables appears as
squares, one for each developmental stage. Basic equations governing the model can be found in Appendix A.

2.2. A stage-structured population model (SSPM)

for C. chilensis

gathered in the field.

Time change (using Stella II notation), for each
variable, is calculated on a daily basis as:

The stage-structured population model (SSPM)

built for C. chilensis is shown in Fig. 2. The state
variables are the abundance of nauplius 6, copepodite stages (from I to V) and adult females.
These variables are functions of food-dependent
egg production, mortality, molting to the successive stage and wind-dominated advection. The
model has eight parameters and two forcing functions (Table 1). The choices of parameter values
are explained below. The model was designed to
run for 60 days, in order to compare the resulting
abundance of copepods with the information

Ci (t)= Ci (tDt)
+ (moltC(i 1) moltC(i + 1) mortCi
AdvecCi )Dt

(3)

where: Ci (t) is the abundance of the stage i at

time t, and Dt =1 day. The first two terms in the
right hand parenthesis represent molt from the
previous stage to the next, mortCi = mortality of
Ci and AdvecCi = advective losses. For the case of
nauplius 6, the first molting term is replaced by:
PopGrowth= EggProd/EggStDur

(4)

V.H. Marn / Ecological Modelling 105 (1997) 6582

69

Table 1
Parameters (p) and forcing functions (ff) for the stage-structured population model of Calanus chilensis (see text for details)
Parameter and forcing functions

Category

Explanation

Units

EggStDur
N6CIVStDur
CVStDur
mortRateN6
MortRateCICIV
MortRateCVFem
MIGc 1
MIGc 2
Food
Me

p
p
p
p
p
p
p
p
ff
ff

Pooled stage duration time from egg to N6

Stage duration time from nauplii 6 trough copepodite IV
Stage duration time for copepodite V
Mortality rate for nauplii 6
Mortality rate from copepodite I through IV
Mortality rate for copepodite V and females
Vertical migration parameter for copepodites I through III
Vertical migration parameter for copepodites IV, V and females
GRAPH(TIME)of chlorophyll-a concentration
GRAPH(TIME) of relative Ekman transport

Days
Days
Days
Days1
Days1
Days1
None

where:
EggProd= Females(0.08 +(Chla0.87))

(5)

and Chla= chlorophyll-a concentration (mg l 1).

2.2.1. Generation time and stages duration time

for C. chilensis
There are no data in the literature regarding the
generation time, Tgen, of C. chilensis. Several approximations were required to arrive at the value
used in the model. First, Tgen was estimated from
field-determined temporal changes of the stage
structure (see Section 3). Second, a literature
search was done to look for Tgen data for other
species of the genus Calanus, especially those
closely related to C. chilensis, such as C. australis
(Marn et al., 1994). Under laboratory conditions,
Peterson and Painting (1990) estimated Tgen for C.
australis to be 21 days at 15.5C, a temperature
close to those measured at Mejillones Bay (Fig.
3A). Bradford et al. (1988), also estimating Tgen
for this species, arrived at a similar number (23.8
days). Finally, the method discussed by Huntley
and Lopez (1992) was used. There they proposed
global equations relating temperature to a set of
population parameters for copepods. For Tgen
they suggest the following equation:
Tgen = 128.8 exp( 0.120T)

(6)

where T =temperature in C. If we evaluate Eq.

(6) for 15C, it gives a Tgen =21 days. This comes
close to the laboratory estimation of Peterson and

mg l1
None

Painting (1990) for C. australis and the lower Tgen

estimated for C. chilensis using field data. Thus,
for the SSPM model a generation time of 21 days
was used.
In order to build a stage-structured population
model, Tgen has to be divided into stage-by-stage
duration times, from eggs to adults. As explained
earlier, data were not available for egg abundance
nor for nauplii 15. Thus, to compare the model
with the field data, the stage duration times from
eggs to nauplius 6 were pooled in one parameter:
EggStDur (Table 1). Stage duration times from
nauplius 6 to copepodite V were estimated from
data on C. australis (Peterson and Painting, 1990).
SSPM stage duration times are shown in Table 4.
Those stage durations are in close agreement with
the temporal changes in stage structure observed
in our field data (see Section 3). The number of
individuals molted per unit of time were then
calculated as a fraction of the stage duration (see
molt terms in Appendix A).
Stage-specific mortality rates for Calanus copepods are not readily available in the literature
(Carlotti and Radach, 1996). Following Nival and
Carlotti (1993), mortality was considered as a
constant fraction of the population. However,
three different terms were used, based on information from Carlotti and Radach (1996): one applicable to nauplii 6, a second mortality term for
copepodite IIV and a third applicable to copepodite V and females. Normal-run, maximum and
minimum values for the stage-specific mortality
rates are given in Table 2.

V.H. Marn / Ecological Modelling 105 (1997) 6582

70

Fig. 3. Environmental variables measured at Mejillones del Sur Bay during September October 1990 and 1991. (A) Ekman
transport, (B) spectral analysis of the Ekman transport, (C) sea surface temperature, (D) integrated chlorophyll-a.

2.2.2. Population growth and chlorophyll-a

Growth of the C. chilensis population in the
SSPM model is accomplished by food-dependent
egg production (Eqs. (4) and (5)). The numerical
values and the form of Eq. (5) were obtained from
the work of Peterson and Bellantoni (1987) on C.
chilensis at Concepcion Bay (Southern Chilean
Coast). The maximum egg rate found by these
authors was of the order of 25 eggs female 1
day 1; this value is similar to that reported by
Table 2
Stage-specific mortality rates used in the SSPM (see text for
details)
Definition

Maximum
Normal run
Minimum

Stage
N6

CICIV

CVFem

0.10
0.08
0.05

0.10
0.06
0.02

0.05
0.04
0.02

Attwood and Peterson (1989) for C. australis in

the Benguela upwelling system. Thus, Eq. (5) was
used only to a food concentration up to 30 mg
Chla m 3; above this value, EggProd was assumed to be food independent and equal to a
maximum rate of 26 eggs female 1 day 1. The
model was also run with a constant maximum egg
production in order to analyze the idea of foodindependent population growth (Escribano and
Rodriguez, 1995).
Temporal changes in chlorophyll-a concentration (Chla in Eq. (5); Food in Appendix A), a
forcing function in the model, were incorporated
using the GRAPH(TIME) function of STELLA
II. The daily values of the function were those
measured in the field.

2.2.3. Ekman transport and the ad6ection terms

The advection terms (AdvecCi, Eq. (7)) were
considered as being negative all the time; these
terms remove organisms. This condition was

V.H. Marn / Ecological Modelling 105 (1997) 6582

Table 3
Ekman transport periods. Numbers in m3 s1 km1

3. Results

Period

3.1. Field data

I
II
III

Year
1990

1991

3839 382 (100%)

786 9495 (63%)
534 9311 (58%)

3509 304 (87%)

5869 223 (38%)
5009 533 (106%)

The ME time series was divided, according to the changes

observed in Fig. 3A, in: days 130 (period I); days 3140
(period II); day 41 61 (period II). Data correspond to
mean9 S.D. (coefficient of variation).

based on wind analysis of the Mejillones area

previously conducted by Marn et al. (1993).
There the authors show that favorable upwelling
conditions occur throughout the spring period
(SeptemberOctober). Thus, the model, as written, should not be used for other times of the year
when that assumption might not necessarily apply.
The Ekman transport, a forcing function in the
advection terms (see Eq. (7) below and Appendix
A) was built using the STELLA II GRAPH(TIME) function and field-determined ME data
(see Section 3; Table 3). ME series, for each year,
were divided by their maximum value to yield a
dynamic term with a range between 0 and 1. An
additional parameter in the advection term is the
migration component (MIGc 1 and MIG c2,
Table 1). All stages from copepodite I to females
were allowed to migrate vertically. In this simple
model, however, daily migration is set as a
parameter that interacts with the Ekman transport in the following way:
AdvecCi =Ci ME(1 MIG c i )

(7)

MIG c1 and MIGc 2 had values ranging from 0

(no migration or full wind effect) to 1 (wind effect
is absent). MIGc 1 applied to CI through CIII,
and MIG c2 to CIV through females. This separation of vertical migration capacities has been
documented for calanoid copepods in upwelling
areas (Verheye et al., 1991).

71

3.1.1. En6ironmental 6ariables

During the period studied, both in 1990 and
1991, ME was positive, indicating favorable upwelling conditions (Fig. 3A). The average transport was similar for both years (t-test= 1.0,
d.f.= 60, P\ 0.30). The fast Fourier transform
(FFT) of the 1990 and 1991 ME values, however,
revealed differences between years (Fig. 3B). The
spectral analysis for 1990 shows a dominant frequency centered on 0.25 day 1. On the other
hand, during 1991 the analysis showed a decreasing trend with frequency (P0.01), with peaks at
0.02, 0.08, 0.13, 0.27 and 0.43 day 1. For the
SSPM, the 1990 ME time series was used as the
wind forcing function in the normal run (see
Appendix A).
Sea surface temperature (SST) was very similar,
both in value (P =0.08, d.f. = 9) and temporal
dynamics, for 1990 and 1991 (Fig. 3C). During
the first 30 days, it remained relatively stable
(30-day average = 15.8C in 1990 and 16.2C in
1991), corresponding to the period of lower values
and higher variability in ME (Table 1). At the
beginning of October (day 32 onwards), ME was
more steady, presenting lower day-to-day variability (Table 3). Lower temperatures, corresponding
to the most intense upwelling event, characterized
that period in both years (Fig. 3C). After that
period, the Ekman transport decreased in intensity and increased in variability. Thus, SST
showed that a single main upwelling event (i.e.
decrease in temperature at day 40) could be identified against a constant background of conditions
favorable for upwelling (Fig. 3A).
Chlorophyll-a fluctuated during both the spring
periods of 1990 and 1991, following the dynamics
of the main upwelling event (Fig. 3D). During the
first 50 days in 1990 the average euphotic zone
integrated chlorophyll was 80 mg m 2; while in
1991 this value was 47 mg m 2. After the upwelling event, chlorophyll-a increased to 436 mg
m 2 in 1990, due to a red tide bloom (Marn et
al., 1993), and to 120 mg m 2 in 1991.

72

V.H. Marn / Ecological Modelling 105 (1997) 6582

Fig. 4. Temporal changes of stage specific abundance for Calanus chilensis at Mejillones del Sur Bay during 1990. Numbers in the
y-axis are in organisms per m3.
Fig. 5. Temporal changes of stage-specific abundance for Calanus chilensis at Mejillones del Sur Bay during 1991. Numbers in the
y-axis are in organisms per m3.

V.H. Marn / Ecological Modelling 105 (1997) 6582

3.1.2. Temporal fluctuations of abundance and

population stage structure of C. chilensis
Despite similarities between years, the spring
abundance of C. chilensis in 1990 (average= 748
organisms m 3, Fig. 4) was one order of magnitude higher than that in 1991 (average =81
copepod m 3, Fig. 5). Abundance at both stations (A and B) was not significantly different
(P \0.9). Thus, data from both stations were
pooled for further analysis. During 1990, nauplius 6 were most abundant during the first 30
days (average = 145 organisms m 3), then decreased to 10 organisms m 3, and finally increased to 50 organisms m 3 during the last
sampling date (Fig. 4A). The following year,
their abundance remained at around 1 organism
m 3, for the first 40 days, and increased to 108
organisms m 3 during the last 20 days (Fig.
5A). In both years, however, the abundance of
nauplii was lowest during the period when Ekman transport was at its highest.
Temporal changes in abundance of copepodites IIV were well correlated, both in 1990
and again 1991 (PB0.05; Figs. 4 and 5). The
oldest copepodite stage (CV) showed a somewhat different pattern. In 1990, their maximum
abundance was delayed 4 days from the other
stages, with the maximum for adults delayed by
2 days from that of CVs (Fig. 4C). During
1991, however, the abundance of copepodites
was very much depressed throughout the entire
period of study (Fig. 5). The average summed
abundance of stage I IV was close to 1.0 organisms m 3 during most of the period, with a
sharp increase toward the end. On the other
hand, CVs never reached 10 organisms m 3
even at their peak abundance.
Temporal changes in the abundance of females followed almost an inverse trend than
that of the nauplii (Fig. 4D and Fig. 5D). During 1990, adults remained below 50 organisms
m 3 for the first 30 days and increased to
nearly 150 organisms m 3 for the rest of the
period (Fig. 4D). On the contrary, during 1991
a single burst of adults was observed during the
12th day ( $200 organisms m 3); the rest of
the time their abundance remained below 10 organisms m 3 (Fig. 5D).

73

Differences between 1990 and 1991, regarding

C. chilensis, are even more striking if one looks
at the relative composition of all developmental
stages and their changes during the period studied (Fig. 6). At the beginning of spring in 1990
(Fig. 6A), C. chilensis population was mostly
copepodite I and II ($ 22 and 50%, respectively). As spring progressed, there was a continuous shift toward older developmental stages,
ending in October with a population composed
mostly of adults (between 70 and 80%). Conversely, during 1991 the population was quite
different, where it shifted from almost 100%
adults, during the first 30 days, to 80% young
developmental stages (nauplius 6$ 50%, copepodite I$ 14%, Copepodite II$ 16%), with
adults nearly absent (0.6%; Fig. 6B), for the remaining 30 days of the study period.
Another way of visualizing these differences is
by calculating the mean stage index, S
(Marn, 1987). This index was used to study the
mesoscale spatial population structure of
calanoid copepods. However, it can also be used
to study the temporal dynamics of a staged
structure population. The index is a simple
weighted average of the mean developmental
stage of the population. S was calculated
from the following equation:
S=

S6i = 0 Ni Si
S6i = 0 Ni

(8)

where Si = stage index, S0 (nauplius 6)=0.1, S1

(copepodite I)= 1, S6 (adults)=6, and Ni is
the corresponding abundance for each stage.
During 1990, S increased from 1.9 to 5.4
(S= 1.87+0.06 days; r 2 = 0.83), as shown in
Fig. 7. Two sharp declines of S during 1990,
which corresponds to recruitment of nauplii 6,
were also observed. If we consider the lapse between low values as an estimate of the population generation time, the data suggest a
generation time of either 20 or 36 days. The
data from 1991 showed the reversed trend: that
is, a population dominated by old stages at the
beginning of the period that switched to young
stages at the very end.

74

V.H. Marn / Ecological Modelling 105 (1997) 6582

Fig. 6. Temporal changes in relative abundance (%) of developmental stages of Calanus chilensis during 1990 (A) and 1991 (B). The
scale for the color pattern appears at the bottom of the figure.

3.2. Stage-structured population model (SSPM)

output
3.2.1. Normal run
Initial conditions and parameter values for the
normal SSPM run are shown in Table 4. The
initial conditions corresponded to the maximum
abundance observed during the first 10 days of
1990. The resulting time series for each developmental stage are shown in Fig. 8. Fluctuations of
abundance were qualitatively similar, and within
the same range, of those observed in 1990 (Fig. 4).
The model showed the secondary increase in nauplii 6 through copepodite III and also the lag in

peak abundance of copepodite IV and V. The

largest difference between the normal run of
SSPM and the 1990 data corresponded to adult
females. SSPM produced a peak abundance of
females within the first 10 days of the run (Fig. 8).
Such increases correspond to the recruitment of
CVs, a situation that was not observed in the
field. This lack of evidence for female recruitment
in the field data may be explained by either a
much larger mortality, or by a larger advection
effect. Those possibilities were studied through a
sensitivity analysis of the models parameters (see
below). The third alternative is the possibility of
field sampling errors.

V.H. Marn / Ecological Modelling 105 (1997) 6582

75

Fig. 7. Temporal changes in the mean stage index S (see text for details) for Calanus chilensis during 1990 and 1991.
Fig. 8. Outcome of the normal run of the SSPM. Numbers in the y-axis are in organisms per m3. See Figs. 4 and 5 for comparison
with real data from Mejillones Bay.

V.H. Marn / Ecological Modelling 105 (1997) 6582

76

A further visualization of the model behavior is

obtained by a plot of the mean stage index S
through time (Fig. 9). Again, the main difference
between the model and the 1990 data corresponded
to the effect of female recruitment at the beginning
of the run. The model, however, showed the sharp
decrease in S observed in 1990 after the upwelling event (see Figs. 7 and 9).

3.2.2. Sensiti6ity analysis

The effects of wind and vertical migration were
evaluated using the sensitivity analysis options of
STELLA II. Time changes of the total copepod
abundance were analyzed using: (1) normal values
(Table 4) and wind data from 1990 (st90) and 1991
(st91); (2) migration terms set to zero, with wind
data from 1990 (W90m0) and 1991 (W91m0); and
(3) migration terms set to 1 with wind data from
1990 (W90m1) and 1991 (W91m1). The results of
the analysis are shown in Fig. 10. These provide
evidence that there are no qualitative differences,
and that the curves are similar, although shifted in
Table 4
Initial conditions and parameter values for the normal run of
the SSPM
State variables (initial conditions)
Variable

Value

Units

N6
CI
CII
CIII
CIV
CV
FEM

70
350
700
1000
460
60
30

Organisms
Organisms
Organisms
Organisms
Organisms
Organisms
Organisms

Value

Units

m3
m3
m3
m3
m3
m3
m3

the Y-axis, when either winds from 1990 and 1991

are used (st90 and st91, Fig. 10). When vertical
migration is set to zero, i.e. full wind effect
(W90m0 and W91m0, Fig. 10), the population goes
to near zero in less than 20 days in 1990, and 40
days in 1991. On the other hand, when migration
is set to one, i.e. wind advection does not affect
organisms (W90m1 and W91m1, Fig. 10), the
population grows explosively, especially after the
main upwelling event. No combination of wind
effect and migration, reproduced the changes in
copepod abundance observed in 1991 (Fig. 5). The
sensitivity analysis of mortality rates showed no
differences among SSPM runs if one uses the range
of values from Table 2. Finally, in the case of a
population not regulated by food and with a
maximum and constant egg production of 26 eggs
per female, the results are different from all previous simulations and field observations (Fig. 11).

4. Discussion
The main objective of this work was to show an
example of how to use STELLA II to study the
population dynamics of a planktonic marine organism. The specific example chosen was the study
of the life cycle of a calanoid copepod, C. chilensis,
and its interaction with the coastal upwelling in the
northern Chilean coast. Given this objective, the
discussion will deal with two issues: (1) What new
information have we gained on the specific example chosen (i.e. upwelling dynamics and life cycles
of planktonic organisms)?; and (2) What is the
potential use of iconographic modeling software in
global programs that require a balanced mixture of
field data analysis and mathematical modeling?

Parameters
Parameter
EggStDur
N6CIVStDur
CVStDur
mortRateN6
mortRateCICIV
mortRateCVFem
MIGc1
MIGc2

10
2
3
0.08
0.06
0.04
0.05
0.60

Days
Days
Days
Days1
Days1
Days1
None
None

4.1. C. chilensis and the coastal upwelling in the

Chilean coast
The results of the SSPM model, together with
the analysis of field data, support the idea that C.
chilensis perform daily vertical migration. This
migration, together with a food-dependent growth
rate, explains the spring temporal variation observed at Mejillones del Sur Bay, Chile. These
results are similar to those obtained by Carlotti

V.H. Marn / Ecological Modelling 105 (1997) 6582

77

Fig. 9. Temporal changes in the mean stage index S of the normal run of the SSPM.

and Radach (1996) for their modeling of the

seasonal dynamics of C. finmarchicus.
Escribano and Rodriguez (1995) argued that
because C. chilensis seems to have a continuous (i.e.
year-round) production in the Humboldt upwelling
ecosystem, this implies that these animals do not
experience food-shortage periods (p. 380). On the
contrary, the SSPM output for C. chilensis (Figs.
4, 8 and 11) shows that food is a controlling factor
in the spring development of this species. A potential explanation for this disagreement is the difference in temporal scale between this investigation
and Escribano and Rodriguezs study; theirs was a
monthly scale, while this study was daily.
A generation time, Tgen, for C. chilensis of 21 days
at 15C was used in the SSPM. This Tgen produced
good agreement between field data and SSPM
results. Thus, given a Tgen of 21 days, it is rather
difficult to obtain population information with
monthly samples as used by Escribano and Rodriguez (1995). Here it is an example of the kind
of research guidance that field data analysis and
simple models can generate. In this specific case, it

is clear that a daily sampling program, or even

twice-a-day, of the C. chilensis population is necessary in order to determine Tgen more precisely.
The SSPM also shows that if the population
does not migrate vertically, the abundance falls to
near zero in about 10 days. Conversely, if the
population is allowed to migrate, the results compare favorably to those obtained in the field.
However, given the preliminary way in which
migration was set up in the model (see Section 2
and Appendix A), little can be said about the
specific migration pattern. Peterson et al. (1979),
in their study of the Oregon upwelling zone,
argued that different zooplankton species are
maintained within the upwelling zone by a specific
relationship between their distributions and the
circulation patterns. The results of this study suggest that, at least for C. chilensis, there may not
be such a clear relationship. Indeed, the SSPM
only required that the population may not be
affected by wind advection on a daily basis (Fig.
10). Thus, these model results suggest that although C. chilensis migrate vertically, changes in

78

V.H. Marn / Ecological Modelling 105 (1997) 6582

Fig. 10. Temporal changes of the total copepod abundance in the SSPM. The runs correspond to the following conditions: st90,
standard run; st91, standard run, using the wind conditions of 1991; W90m1, wind conditions for 1990 with migration terms set to
1; W90m0, wind conditions of 1990 with migration term set to 0; W91m1, wind conditions for 1991 with migration terms set to 1;
W91m0, wind conditions of 1991 with migration term set to 0.

abundance reflect the time evolution of the upwelling dynamics, both through food and advection effects. That is, vertical migration does not
compensate for the general drifting problem, but
it is more than likely an attenuation factor.

4.2. Iconographic modeling software

Through the development of a simple-biology,
stage-structured population model (SSPM), it has

been shown that iconographic modeling software,

such as STELLA II, can be used as a communication link between two scientific activities: field
data collection and mathematical modeling. The
model developed here (see Appendix A) can be
either improved, if one wants to follow a straight
modeling approach, or be used as a hypothesisgenerating mechanism that may highlight key unresolved problems, that then will require further
field sampling programs. There are many details

V.H. Marn / Ecological Modelling 105 (1997) 6582

79

Fig. 11. Outcome of the SSPM considering constant, maximum egg production. Numbers in the y-axis are in organisms per m3.

left aside by using this simple modeling approach,

such as three-dimensional oceanic circulation
equations, which requires more powerful modeling platforms than the Windows95-based, Pentium-PC used for this paper. However, by using
this approach, modelers may benefit from the
interaction with non-modelers and their insights
into the characteristics of the ecosystems that they
study. This may be especially true in fields such as
marine zooplankton and zooplankton circulation
interaction studies, where a large percentage of
the information that appears every year is still
descriptive in nature.

5. Conclusions
The simulation of changes in stage-specific abundance of copepods through a stage-structured population model, SSPM, shows that the maintenance
and growth of a local population of C. chilensis at
Mejillones del Sur Bay, Chile, can be explained as
a combination of advective forces and vertical
migration plus food-dependent egg production.
Easy-to-learn iconographic modeling software,
such as STELLA II, can be an effective tool in
closing the gap between empiricists and modelers whose interaction will be vital in the develop-

80

V.H. Marn / Ecological Modelling 105 (1997) 6582

ment of marine global programs. These tools are

also available to scientists working in third world
countries where budgets are unlikely to be available for large computers. However, the use of
these simple modeling tools should be considered
only in the preliminary phase of the studies, at
least for three-dimensional advective systems such
as marine ecosystems.

Acknowledgements
This work was partially funded by projects
FONDECYT 268/89 and 1049/92 from CONICYT-CHILE and also by project EIMS (Environmental Information and Modeling System)
awarded to Universidad de Chile by IBM Environmental Research Fund (IBM International
Foundation). The author is thankful to Donald
D. Adams (State University of New York, Plattsburgh) and to two anonymous reviewers for their
critical comments of the manuscript.

References
Angel, M.V., 1994. Spatial distribution of marine organisms.
In: Edwards, P.J., May, R.M., Webb, N.R. (Eds.), Largescale Ecology and Conservation Biology. Blackwell Science, New York, pp. 59110.
Attwood, C.G., Peterson, C.G., 1989. Reduction in fecundity
and lipids of the copepod Calanus australis (Brodskii) by
strong pulsed upwelling. J. Exp. Mar. Biol. Ecol. 129,
121 131.
Bowden, K.F., 1983. Physical Oceanography of Coastal Waters. Ellis Horwood Series in Marine Science. Wiley, New
York, 302 pp.
Bradford, J.M., Ohman, M.D., Jillett, J.B., 1988. Larval morphology and development of Neocalanus tonsus, Calanoides
macrocarinatus, and Calanus australis (Copepoda:
Calanoida) in the laboratory. N. Z. J. Mar. Freshwater
Res. 22, 301 320.
Carlotti, F., Radach, G., 1996. Seasonal dynamics of phytoplankton and Calanus finmarchicus in the North Sea as
revealed by a coupled one-dimensional model. Limnol.
Oceanogr. 41, 522 539.
Conover, R.J., 1988. Comparative life histories in the genera
Calanus and Neocalanus in high latitudes of the northern
hemisphere. Hydrobiologia 167/168, 127142.
Escribano, R., Rodriguez, L., 1994. Life cycle of Calanus
chilensis Brodsky in Bay of San Jorge, Antofagasta, Chile.
Hydrobiologia 292/293, 289294.

Escribano, R., Rodriguez, L., 1995. Seasonality of size and

growth of Calanus chilensis in northern Chile. Rev. Chil.
Hist. Nat. 68, 373 382.
GLOBEC, 1993. Population dynamics and physical variability. Report of the First Meeting of an International
GLOBEC Working Group, St. Johns College, Cambridge
University. GLOBEC Report No. 2. 104 pp.
Gonzalez, A., 1993. Distribucion espacial de dos copepodos
calanoideos Calanus chilensis (Brodsky) y Centropages
brachiatus (Dana) en el norte de Chile. Tesis de Grado
para optar al grado de Licenciado en Biologa Marina,
Universidad Austral de Chile, 69 pp.
Hannon, B., Ruth, M., 1994. Dynamic Modelling. SpringerVerlag, New York, 248 pp.
Huntley, M.E., Lopez, M.D.G., 1992. Temperature-dependent
production of marine copepods: a global synthesis. Am.
Nat. 140, 201 242.
Laevastu, T., Larkins, H.A., 1981. Marine Fisheries Ecosystem. Its Quantitative Evaluation and Management. Fishing
News Book, Farnham, 159 pp.
Marn, V., 1986. Distribution and life cycle of three antarctic
copepods (Calanoides acutus, Calanus propinquus and
Rhincalanus gigas). Ph.D. Dissertation, University of California, San Diego, CA, USA, 144 pp.
Marin, V., 1987. The oceanographic structure of the Eastern
Scotia Sea. IV. Distribution of copepod species in relation
to hydrography in 1981. Deep-Sea Res. 34, 105 121.
Marn, V., 1988. Independent life cycles: an alternative to the
asynchronism hypothesis for antarctic calanoid copepods.
Hydrobiologia 167/168, 161 168.
Marn, V., Rodriguez, L., Vallejo, L., Fuenteseca, J., Oyarce,
E., 1993. Efectos de la surgencia costera sobre la productividad primaria de Baha Mejillones del Sur (Antofagasta,
Chile). Rev. Chil. Hist. Nat. 66, 479 491.
Marn, V., Espinoza, S., Fleminger, A., 1994. Morphometric
study of Calanus chilensis males along the Chilean coast.
Hydrobiologia 292/293, 75 80.
Nival, P., Carlotti, F., 1993. Zooplankton models. In: Population Dynamics and Physical Variability. Report of the
First Meeting of an International GLOBEC Working
Group, St. Johns College, Cambridge University.
GLOBEC Report, No. 2, pp. 61 65.
Peterson, W.T., Bellantoni, D.C., 1987. Relationships between
water-column stratification, phytoplankton cell size and
copepod fecundity in Long Island Sound and off Central
Chile. S. African J. Mar. Sci. 5, 411 421.
Peterson, W.T., Painting, S.J., 1990. Developmental rates of
the copepod Calanus australis and Calanoides carinatus in
the laboratory, with discussion of methods used for calculation of development time. J. Plankton Res. 12, 283 293.
Peterson, W.T., Miller, C.B., Hutchinson, A., 1979. Zonation
and maintenance of copepod populations in the Oregon
upwelling zone. Deep-Sea Res. 26, 467 494.
Powell, T., Hofmann, E., Osborn, T., Price, J., Rotschild, B.,
Roughgarden, J., 1990. Theory and modeling in GLOBEC:
a first step. Report to the GLOBEC Steering Committee
from the Working Group on Theory and Modeling, 9 pp.

V.H. Marn / Ecological Modelling 105 (1997) 6582

Rykiel, E.J. Jr., 1996. Testing ecological models: the meaning
of validation. Ecol. Model. 90, 229244.
Shermann, K., 1995. Large marine ecosystems and fisheries. In:
Munasinghe, M., Shearer, W. (Eds.), Defining and Measuring Sustainability. The Biogeophysical Foundations. The
United Nations University and the World Bank, pp. 207
233.
Sinclair, M., 1988. Marine Populations. An Essay on Population
Regulation and Speciation. Books in Recruitment Fishery
Oceanography, Washington Sea Grant Program, Seattle,
WA, 252 pp.
Smith, S.L., 1984. Biological indications of active upwelling in
the northwestern Indian Ocean in 1964 and 1979, and a

81

comparison of zooplankton in the southeastern Bering Sea.

Deep-Sea Res. 31, 951 967.
Strickland, J., Parsons, T.R., 1972. A practical handbook of
seawater analysis. Bull. Fisheries Res. Board Can. 5, 1 128.
Strub, P., Mesas, J., Montecino, V., Rutllant, J. Coastal ocean
circulation off western South America. In: Robinson, A.,
Brink, K. (Eds.), The Sea. First Workshop on Coastal Ocean
Advanced Science and Technology Studies (COASTS). IOC,
Liege, Belgium, in press.
Verheye, H.M., Hutchings, L., Peterson, W.T., 1991. Life
history and population maintenance strategies of Calanoides
carinatus (Copepoda:Calanoida) in the Southern Benguela
ecosystem. S. African J. Mar. Sci. 11, 179 191.

Appendix A
Equations and terms of the stage-structured population model (SSPM) for C. chilensis. The model was
built and run using the modeling software STELLA II (version 3.0.7 for windows). Comments appear
in square brackets.
{Initialization}
[Values for the standard run are given in Table 4; GRAPH(TIME) data were taken from Fig. 3]
INIT N6=
INIT CI=
INIT CII=
INIT CIII=
INIT CIV=
INIT CV=
INIT Fem=
Food =GRAPH(TIME)
[(day, value)]
(0.00, 2.00), (3.16, 2.00), (6.32, 2.00), (9.47, 3.50), (12.6, 3.50), (15.8, 7.00), (18.9, 7.00), (22.1, 6.50),
(25.3, 6.50), (28.4, 9.00), (31.6, 6.00), (34.7, 7.00), (37.9, 3.00), (41.1, 1.50), (44.2, 2.00), (47.4, 2.00),
(50.5, 1.00), (53.7, 1.30), (56.8, 44.5), (60.0, 9.00)
Me = GRAPH(TIME)
[day, value]
(0.00, 0.222), (1.00, 0.203), (2.00, 0.291), (3.00, 0.138), (4.00, 0.11), (5.00, 0.894), (6.00, 0.715), (7.00,
0.014), (8.00, 0.003), (9.00, 0.005), (10.0, 0.32), (11.0, 0.377), (12.0, 0.006), (13.0, 0.384), (14.0,
0.002), (15.0, 0.07), (16.0, 0.465), (17.0, 0.39), (18.0, 0.121), (19.0, 0.32), (20.0, 0.193), (21.0, 0.171),
(22.0, 0.019), (23.0, 0.022), (24.0, 0.212), (25.0, 0.299), (26.0, 0.204), (27.0, 0.763), (28.0, 0.276), (29.0,
0.001), (30.0, 0.036), (31.0, 0.425), (32.0, 0.308), (33.0, 0.231), (34.0, 0.229), (35.0, 0.536), (36.0,
1.00), (37.0, 0.743), (38.0, 0.146), (39.0, 0.279), (40.0, 0.885), (41.0, 0.15), (42.0, 0.001), (43.0, 0.768),
(44.0, 0.401), (45.0, 0.273), (46.0, 0.276), (47.0, 0.349), (48.0, 0.322), (49.0, 0.229), (50.0, 0.241), (51.0,
0.603), (52.0, 0.479), (53.0, 0.107), (54.0, 0.078), (55.0, 0.443), (56.0, 0.425), (57.0, 0.212), (58.0,
0.201), (59.0, 0.388), (60.0, 0.549)

{Runtime equations}

82

V.H. Marn / Ecological Modelling 105 (1997) 6582

N6(t)= N6(tdt)+(PopGrowthmortN6moltN6AdvecN6)*dt
CI(t) =CI(tdt)+(moltN6mortCImoltCIAdvecCI)*dt
CII(t)= CII(tdt)+(moltCImortCIImoltCIIAdvecCII)*dt
CIII(t)=CIII(tdt)+(moltCIImortCIIImoltCIIIAdvecCIII)*dt
CIV(t)=CIV(tdt)+(moltCIIImortCIVmoltCIVAdvecCIV)*dt
CV(t)=CV(tdt)+(moltCIVmortCVmoltCI 5AdvecCV)*dt
Fem(t)= Fem(tdt)+(moltCI 5mortFemAdvecFem)*dt
Total= CI+CII+CIII+CIV+CV+Fem+N6
S= ((N6*0.1)+(CI*1.0)+(CII*2)+(CIII*3)+(CIV*4)+(CV*5)+(Fem*6))/Total
{PRODUCTION and MOLTING TERMS}
PopGrowth = (EggProd/EggStDur)
EggProd =IF (Food530) then (0.08+(Food*0.87))*Fem ELSE 26
moltN6= N6/N6CIVStDur
moltCI = CI/N6CIVStDur
moltCII =CII/N6CIVStDur
moltCIII = CIII/N6CIVStDur
moltCIV=CIV/N6CIVStDur
moltCV= CV/CVStDur
{MORTALITY TERMS}
mortN6=N6*mortRateN6
mortCI = CI*mortRateCICIV
mortCII =CII*mortRateCICIV
mortCIII =CIII*mortRateCICIV
mortCIV= CIV*mortRateCICIV
mortCV= CV*mortRateCVFem
mortFem= Fem*mortRateCVFem
{ADVECTION TERMS}
AdvecN6=N6*Me
AdvecCI= CI*Me*(1MIG c1)
AdvecCII= CII*Me*(1MIG c 1)
AdvecCIII= CIII*Me*(1MIG c1)
AdvecCIV= CIV*Me*(1MIG c2)
AdvecCV= CV*Me*(1MIG c 2)
AdvecFem=Fem*Me*(1MIG c2)
{PARAMETERS}
[See Table 4 for standard run parameter values and units]
EggStDur=
N6CIVStDur=
CVStDur =
mortRateN6=
mortRateCICIV=
mortRateCVFem=
MIGc 1=
MIG c2=