Ecology, 88(4), 2007, pp.

918928
2007 by the Ecological Society of America
LIFE HISTORY AND VIABILITY OF A LONG-LIVED MARINE
INVERTEBRATE: THE OCTOCORAL PARAMURICEA CLAVATA
CRISTINA LINARES,
1,2,6
DANIEL F. DOAK,
3
RAFEL COMA,
1
DAVID DI

AZ,
4
AND MIKEL ZABALA
5
1
Centre dEstudis Avanc ats de Blanes, Acces Cala Sant Francesc 14, 17300 Blanes, Girona, Spain
2
Institut de Cie`ncies del Mar, Passeig Martim 37-49, 08003 Barcelona, Spain
3
Department of Ecology and Evolutionary Biology, University of California, Santa Cruz, California 95064 USA
4
Instituto Espanol de Oceanografa, C/ Moll de Ponent s/n, 07015 Palma de Mallorca, Spain
5
Departament dEcologia, Facultat de Biologia, Universitat de Barcelona, Avda Diagonal 645, 08028 Barcelona, Spain
Abstract. The red gorgonian Paramuricea clavata is a long-lived, slow-growing sessile
invertebrate of ecological and conservation importance in the northwestern Mediterranean
Sea. We develop a series of size-based matrix models for two Paramuricea clavata populations.
These models were used to estimate basic life history traits for this species and to evaluate the
viability of the red gorgonian populations we studied. As for many other slow-growing
species, sensitivity and elasticity analysis demonstrate that gorgonian population growth is far
more sensitive to changes in survival rates than to growth, shrinkage, or reproductive rates.
The slow growth and low mortality of red gorgonians results in low damping ratios, indicating
slow convergence to stable size structures (at least 50 years). The stable distributions predicted
by the model did not differ from the observed ones. However, our simulations point out the
fragility of this species, showing both populations in decline and high risk of extinction over
moderate time horizons. These declines appear to be related to a recent increase in
anthropogenic disturbances.
Relative to their life span, the values of recruitment elasticity for Paramuricea clavata are
lower than those reported for other marine organisms but are similar to those reported for
some long-lived plants. These values and the delayed age of sexual maturity, in combination
with the longevity of the species, show a clear fecundity/mortality trade-off. Full demographic
studies of sessile marine species are quite scarce but can provide insight into population
dynamics and life history patterns for these difcult and under-studied species. While our
work shows clear results for the red gorgonian, the variability in some of our estimates suggest
that future work should include data collection over longer temporal and spatial scales to
better understand the long-term effects of natural and anthropogenic disturbances on red
gorgonian populations.
Key words: elasticity; gorgonian; life history; longevity; matrix models; Mediterranean; Paramuricea
clavata; population viability analysis; recruitment; sensitivity; size-structured populations.
INTRODUCTION
While demographic models are widely used to assess
the population viability and life history patterns of many
terrestrial plant and animal species (Silvertown et al.
1996, Caswell 2001, Beissinger and McCullough 2002,
Doak et al. 2002), they have been used less commonly in
marine conservation than in other elds of population
ecology (Crowder et al. 1994, Thompson et al. 2000,
Holmes and York 2003, Gerber and Heppell 2004). In
particular, age- or stage-based matrix models have been
conducted on only a few groups of marine invertebrates,
including sea urchins (Pster and Bradbury 1996),
bivalves (Nakaoka 1993, 1997), bryozoans (Hughes
1990), hard corals (Hughes 1984, Done 1988, Babcock
1991, Hughes and Tanner 2000), and soft corals and
gorgonians (Gotelli 1991, Lasker 1991, McFadden
1991).
In this study, we develop a demographic matrix model
for populations of the long-lived, slow-growing Medi-
terranean red gorgonian coral, Paramuricea clavata
(henceforth, red gorgonian). In common with other
clonal plants and marine invertebrates (Hughes 1984,
Cook 1985, Hughes and Jackson 1985), gorgonians have
low recruitment rates, high longevity, and low mortality
rates (Lasker 1991, Yoshioka 1994, Coma et al. 1998,
2003, 2004). These traits have resulted in local endan-
germent of red gorgonians by SCUBA-based tourism in
marine reserves (Coma et al. 2004) and in regional
endangerment due to mass mortality events (Cerrano et
al. 2000, Perez et al. 2000, Linares et al. 2005). Red
gorgonians and other long-lived species serve as
ecosystem engineers (Jones et al. 1994) in Mediterra-
nean marine hard-bottom communities, with signicant
effects on the structure and biodiversity of their
communities.
Manuscript received 9 December 2005; revised 19 September
2006; accepted 29 September 2006. Corresponding Editor: S. R.
Thorrold.
6
E-mail: [email protected]
918
We have developed both deterministic and stochastic
size-based matrix models for two red gorgonian
populations in the northwestern Mediterranean Sea.
We use these models to address three goals. First, we
estimate basic life history traits for this species and use
the resulting models to evaluate the sensitivity of
population growth to different vital rates. Second, we
ask how population structure will develop through time
and analyze the viability of the red gorgonian popula-
tions we studied. Third, we compare the life history
patterns of red gorgonian to those of other sessile species
of invertebrates and of terrestrial plants to highlight the
signicance of their reproductive strategy and the
resemblance with species displaying similar life history
traits despite having very different environments and
evolutionary histories.
METHODS
The natural history of Paramuricea clavata
Paramuricea clavata (Anthozoa, Octocorallia) is one
of the key species of the highly diverse Mediterranean
coralligenous communities, playing an important role in
the maintenance of biomass and the structural com-
plexity in these habitats (True 1970, Ballesteros 2006).
Red gorgonians can be found in rocky substrates from
10 m to at least 110 m, in locations subjected to strong
currents (Weinberg 1991), and are widely distributed in
the western basin of the Mediterranean and in the
Adriatic Sea. In spite of the species broad geographic
range, red gorgonians have a strikingly patchy distribu-
tion (see Plate 1).
Red gorgonian colonies are discrete and generally
assume a two-dimensional arborescent form, reaching
heights of up to 1.5 m. At present, there is no method
available to directly estimate the age of living gorgonian
colonies. Past studies have estimated mean growth rates
of ;0.8 cm/yr in colony height (Coma et al. 2001; R.
Coma, unpublished data), suggesting ages of up to 50
100 yr for individual colonies. However, breakage of
branches on the main stalk is relatively common, so
colonies may be much older than they appear. Due to
variation in growth rates among individuals and the
possibility that individuals can undergo partial mor-
tality (loss of living tissue, in part due to branch
breakage), two colonies of the same size may be very
different in age. As for many other species with
indeterminate growth, the demographic characteristics
of gorgonian colonies (survival, growth, and fecundity)
appear to be inuenced more by size than by age.
Consequently, we classify colonies by size in all of our
models.
Gorgonians suffer high mortality rates as recruits and
juvenile colonies (Coma et al. 2001) and low natural
mortality as adults (Coma et al. 2004). While partial
mortality due to strong currents occurs at a low rate, this
damage can be dramatically increased by diving activity,
shing nets and lines, and mass mortality events
(Bavestrello et al. 1997, Coma et al. 2004, Linares et
al. 2005). Colonies can survive and potentially regrow
following the loss of living tissue, but previous data
indicate that regeneration depends on the extent of
injuries and is often fairly low (Bavestrello and Boero
1986, Linares et al. 2005).
The reproductive biology of this species has been well-
studied. The red gorgonian is dioecious, with a sex ratio
not signicantly different from 1:1 in the studied
PLATE 1. A red gorgonian (Paramuricea clavata) population in the Medes Islands (northwestern Mediterranean Sea). Photo
credit: Enric Ballesteros.
April 2007 919 DEMOGRAPHIC MODELS FOR AN OCTOCORAL
populations (Coma et al. 1995a; R. Coma, unpublished
data). Spawning is synchronous and occurs twice each
year in June around the new and full moons. Fertiliza-
tion is external and embryogenesis takes place on the
surface of the colony (Coma et al. 1995a). Red
gorgonian planulae have little potential for dispersal
and most settle near the parental colony (C. Linares,
unpublished data). While red gorgonians make signi-
cant investment in reproduction each year, recruitment
rates are typically low (Coma et al. 1995a, 2001).
Data
We analyzed the demographic parameters of red
gorgonian populations located at two marine protected
areas of the northwestern Mediterranean coast, Cap de
Creus and Medes Islands, separated by .25 km. Both
localities are typical of the habitats where well-devel-
oped red gorgonian populations are found along the
northwestern Mediterranean coast, and the studied
populations are located at the same depth range (15
25 m). Due to their beauty, red gorgonian populations
are among the most attractive areas for scuba divers,
making it difcult to study populations that are not
subjected to diving impacts. Although our study sites are
located within marine reserves, both are subject to
considerable levels of diving activity (Coma et al. 2004).
The demographic parameters used in this study are
based on the data from a previous study (Coma et al.
2003), with survivorship, growth, and recruitment of red
gorgonian colonies estimated from annual monitoring of
six 4 m long and 0.8 m wide permanent plots at Cap de
Creus (two plots from 20022004) and Medes Islands
(four plots from 20012004). All colonies present in a
plot were individually identied and mapped at the start
of the study and relocated each year to record their fates.
Each of the six plots contained between 80 and 200
colonies in each year of the study. The size of each
colony was measured as maximum height in centimeters,
and mortality was estimated both through the disap-
pearance of whole colonies and by the presence of
standing dead skeletons of colonies. Reproduction is
extremely difcult to quantify in the eld, so neither the
sex nor reproductive effort was measured in this study.
Previous work indicates that the size distributions and
hence the demography of male and female colonies are
similar (Coma et al. 2004). Newly settled primary polyps
are extremely small and thus are not readily observable
in the eld, with newly found colonies typically between
0.3 cm and 3 cm height. Given the growth rates of
newly settled individuals observed in the laboratory (C.
Linares, unpublished data), these newly found colonies
are likely to be in their second year of life.
Model formulation
We analyzed the demography of red gorgonians using
a set of seven age- and size-dened stage classes. The
rst class is age-based, representing the newborn
colonies (primary polyps) that we were unable to
directly observe in the eld. Class 2 colonies are at least
2 yr old and between 0.3 and 3 cm in height, and we
assumed that all newly observed colonies in the eld fall
into this class. The remaining colonies are all size-based:
class 3 colonies are between 3 and 10 cm in height and
are pre-reproductive, while classes 4 through 7 are 1120
cm, 2130 cm, 3140 cm, and .40 cm in height,
respectively, and have increasing reproductive output.
These classes are chosen to match those used during
previous work on this species (Coma et al. 1995b).
Further division of the larger colonies into narrower size
classes would be desirable, but the low numbers of
colonies at these sizes prevent their use.
Across each annual transition period, each gorgonian
colony could have one of ve fates: (1) survive and grow
to the next largest size class; (2) survive and shrink by
one size class; (3) survive and shrink by two size classes
(only applicable to classes 6 and 7); (4) survive but
remain in the same size class; or (5) die. Because these
overall fates are governed by combinations of survival,
growth, and shrinkage rates, we performed most of our
analysis on these underlying vital rates: s
i
, the probabil-
ity of a class i colony surviving; g
i
, the probability of a
class i colony growing (conditional on surviving); h
i
, the
probability of a class i colony shrinking by either one or
two size classes (conditional on surviving and not
growing); and h
2i
, the probability of a class i colony
shrinking by two size classes (conditional on surviving
and shrinking). The matrix model we used, dened in
terms of these vital rates, is shown in Table 1. These
rates were directly estimated from the data for each
annual transition using all individuals in either popula-
TABLE 1. Size class transition matrix wherein matrix elements represent four types of variables (f
i
, g
i
, h
i
, h
2i
, s
i
): f, fecundity; g,
growth; h, shrink into next smaller class; h
2,
shrink into class two intervals smaller; s, survival.
Size class
t 1
Size class at time t
1 2 3 4 5
1 0 0 0 f
4
f
5
2 s
1
3 g
1
s
2
3 (1 g
2
) s
3
3 (1 g
3
) 3 h
3
0 0
3 0 s
2
3 g
2
s
3
3 (1 g
3
) 3 (1 h
3
) s
4
3 (1 g
4
) 3 h
4
0
4 0 0 s
3
3 g
3
s
4
3 (1 g
4
) 3 (1 h
4
) s
5
3 (1 g
5
) 3 h
5
5 0 0 0 s
4
3 g
4
s
5
3 (1 g
5
) 3 (1 h
5
)
6 0 0 0 0 s
5
3 g
5
7 0 0 0 0 0
Note: Two red gorgonian populations were studied in the northwestern Mediterranean Sea.
CRISTINA LINARES ET AL. 920 Ecology, Vol. 88, No. 4
tion. For these estimates, we grouped data across all
plots within a population. Separate likelihood ratio chi-
square (G) tests for each of the ve site by annual
transition combinations (estimated with all zero fre-
quency cells dened as structural zeros in increasing
power) showed signicant plot effects on transition
probabilities only for the Medes population from 2002
2003 (P 0.012). We also used G tests to check the
assumption that demographic rates can be modeled as
rst-order Markov processes (Caswell 2001) and found
no signicant second-order effects for either population.
With the classes we used, the median and minimum
number of colonies starting each transition in each year
were 99 and 16, respectively, giving reasonable sample
sizes for vital rate estimation (the minimum number of
colonies were all for size class 2, newly found colonies).
As a check, we also used logistic regression to estimate
survival as a continuous function of size (log likelihood
2LL793.434, chi square 33.788, df 3, P ,0.001);
this resulted in survival estimates extremely similar to
those found with direct estimation. Since no data are
available on the demographic rates for yearling (prima-
ry) polyps due to the difculty of detecting them in the
eld, we made the optimistic assumption that their
survival is equal to that of size class 2 and that all
surviving class 1 polyps advance to class 2. Preliminary
analyses indicated that alterations of this assumption
had no important effects on our results.
To estimate reproduction, we used the relationship
between the colony size and the gonadal production
described in a previous study of this species (Coma et al.
1995b). Based on these data, we assume that the
reproductive term f
i
can be estimated as the product of
(1) the size-dependent gonadal production of oocytes
from female colonies (assuming a 1:1 sex ratio; Coma et
al. 1995a); (2) the fecundity of the gonads (fraction of
gonads converted to eggs (C. Linares, unpublished data);
and (3) the annual survival rate from eggs to primary
polyps that would be seen during the census just before
the reproductive pulse in June (C. Linares, unpublished
data). (Note that we do not include a correction of one-
half in this estimation of f to account for dioecy; while
only half of colonies are female, our model is for the
entire population of male and female colonies, such that
the halving of colony number to estimate the number of
female colonies is offset by the doubling of reproduction
to account for male offspring.) The resulting size-
dependent estimates of total production of class 1 polyp
were then used in each matrix estimated, but did not
vary with population or year.
From these vital rate estimates we constructed
transition matrices for each time interval for each
population (see Appendix A) and a mean matrix for
each population. We calculated several outputs from
these matrix models. First, we determined the long-term
deterministic growth rate, k
1
, predicted by each matrix.
Second, we calculated the damping ratio for each
matrix, q k
1
/jk
2
j (Caswell 2001), which provides a
measure of how fast a population will converge to the
stable stage distribution. Third, we quantied patterns
of importance to k
1
of changes in each vital rate for the
mean matrix for each population; given that there is no
single correct scaling for measures of relative importance
of different vital rates (Caswell 2001, Morris and Doak
2002), we present both sensitivity and elasticity values as
measures of importance. Finally, we asked how size
structure will develop following population establish-
ment, a critical question in interpreting observed
population structures. To explore the development of a
newly founded population, we used the mean matrix of
each population to simulate 100 years of population
growth, starting with 100 colonies of class 1 polyp. At
each time step the size distribution was recorded, and the
results were compared with the observed distribution of
each site. To compare the observed and predicted size
distributions, the class 1 (primary polyps) were not used,
since this class is not observable in the eld.
We also conducted a series of stochastic analyses.
First, we estimated the stochastic population growth
rate, k
S
for each population using equally likely random
draws of our annual matrices. These random-draw
simulations were also used to estimate the probabilities
of quasi-extinction for each population for up to 150
years, using an extinction threshold of 10% of the initial
population size and starting with a population size of
300 at stable stage distribution. Second, we used annual
estimates of each vital rate to calculate the temporal
means, variances, and covariances of vital rates and used
Tuljapurkars small noise approximation to calculate
stochastic sensitivity and elasticity values (Tuljapurkar
1990; programs from Morris and Doak 2002 and Doak
et al. 2005).
Finally, we employed the equations of Cochran and
Ellner (1992) to calculate several different life history
parameters for our mean matrices that are useful
descriptors of longevity. Using their formulas we estimat-
edthe survivorshipfunctionl
x
, whichgives the probability
that a newborn will survive to age x (Cochran and Ellner
[1992]: Eq. 2); the conditional total life span for newborn
colonies, whichis the meanage at deathfor acolony, given
that they survive long enoughtoenter size class i (Cochran
and Ellner [1992]: Eq. 6); and the mean age of residence in
each size class, which provides a method for relating the
TABLE 1. Extended.
Size class at time t
6 7
f
6
f
7
0 0
0 0
s
6
3 (1 g
6
) 3 h
6
3 h
26
0
s
6
3 (1 g
6
) 3 (1 h
2,6
) 3 h
6
s
7
3 h
7
3 h
2,7
s
6
3 (1 g
6
) 3 (1 h
6
) s
7
3 (1 h
2,7
) 3 h
7
s
6
3 g
6
s
7
3 (1 h
7
)
April 2007 921 DEMOGRAPHIC MODELS FOR AN OCTOCORAL
size and age of a given colony (Cochran and Ellner [1992]:
Eq. 29). To estimate these parameters we used the survival
matrixPfromthe meanmatrixfromeachsite andthe birth
matrix B, which contains the reproductive values in the
rst rows and for which all other entries are zero (see
Cochran and Ellner [1992] for details). These estimates
were performed with the program STAGECOACH
(Cochran and Ellner 1992).
We compared two life history patterns of red
gorgonians obtained from our analyses with those of
other sessile plants and animals. Following other studies
(Silvertown et al. 1993, Franco and Silvertown 1996) we
plotted expected life span (after the rst year) against the
summed elasticities of early life stage events (see Forbis
and Doak [2004] for methods). For this comparison we
gleaned data from the literature from demographic
studies of sessile marine invertebrates and algae,
terrestrial herbs, and terrestrial shrubs and trees (see
Appendix B for species and data). We also contrasted
the life span against the rst age at sexual maturity with
species for which we could glean the necessary data from
the literature (see Appendix C for species and data).
RESULTS
Life history and sensitivity analysis
Our model results indicate that red gorgonians are
long-lived. We found 6397% survival, with the lowest
values for class 2 colonies and survival rising to 83100%
in the largest size class. The populations showed a mean
survival rate of ;91% and a mean recruitment rate of
;3% (see Appendix D). In addition, growth rates were
slow, with the medians of the probability of growing
ranging from 0 to 41% across sizes and years. Just as
strikingly, the probability of shrinking was higher than
that of growing for many size classes in many years (see
Appendices A and D).
Red gorgonian individuals followed a characteristic
type III survivorship curve, with rapid declines in
survivorship during the rst years of life followed by
slow and constant declines of large and old individuals
and high longevity (Fig. 1a). Because survival of the
smaller size classes was low, mean total life span from
settlement was relatively short, but if colonies survived
to reach the largest size class, they lived .60 yr (Fig. 1b).
The high variance around each age-dependent life span
estimate indicates that the fates of individual colonies
are highly variable, with some colonies living consider-
ably longer than the average. The mean age of residence
in each class provided a direct way to examine the size
age relationship. The mean age at residence increased
progressively with size class, with a predicted value of
54.9 yr for colonies in size 7 (Fig. 1c). As for total life
span, the high variances in age at residence for larger
sizes mean that some colonies will be far older than the
mean and some far younger.
The sensitivities to mean vital rates estimated by
deterministic and stochastic models were quite similar
(Fig. 2a, b), indicating that the degree of variability in
FIG. 1. (a) Survivorship curve (log scale) for newly
established red gorgonian (Paramuricea clavata) colonies from
Eq. 2 of Cochran and Ellner (1992). (b) Total estimated
remaining life span, conditional on reaching a size class. (c) Age
at residence or the age of individuals expected to be found in a
size class for a population at stable stage distribution. The size
classes correspond to the seven age and size stage classes dened
in the model (see Methods: Model formulation for a thorough
description of each class). Error bars representing one standard
deviation, calculated using formulae of Cochran and Ellner
(1992), are shown above each estimate. Red gorgonian
populations were studied in the northwestern Mediterranean
Sea.
CRISTINA LINARES ET AL. 922 Ecology, Vol. 88, No. 4
demography observed over the study period was not
particularly important for red gorgonian population
dynamics. In addition, the sensitivities of k
S
to variances
and covariances of vital rates were small (results not
shown); consequently, we present only the deterministic
elasticity results. Elasticities showed a similar, but
stronger, pattern to that of the sensitivities, indicating
that gorgonian population growth is far more sensitive
to changes in survival rates than to growth, shrink, or
reproductive rates (Fig. 3). The survivorship of the
largest colonies (height . 40 cm) did not have the
highest elasticity values at either site, due to the relative
rarity of these largest individuals. For the Medes
Islands, the elasticity of survival rates for class 4 and 5
had the largest values, while in Cap de Creus, the highest
elasticities were for survival of classes 5 and 6. The
elasticity analyses also suggested that the reproductive
parameters (such as reproductive values and survival of
newborns and recruits) contributed little to the rate of
population increase over the examined period.
FIG. 2. Deterministic and stochastic sensitivities of red gorgonian population growth rate to mean vital rates for (a) Medes Island
and (b) Cap de Creus populations. Abbreviations are: f, fecundity; g, growth; h, shrink into next smaller class; h
2,
shrink into class two
intervals smaller; s, survival. The numbers following the vital rates abbreviations indicate the respective size class for each vital rate.
April 2007 923 DEMOGRAPHIC MODELS FOR AN OCTOCORAL
Population growth and stage structure
The annual multiplication rate associated with the
different site and time matrices varied considerably
(Table 2), but all matrices predicted declines in
gorgonian populations (k , 1). Notably, k
1
values for
the two mean matrices were almost identical to k
S
values
for the populations, again indicating the extremely small
effects of environmental stochasticity. k
1
values were
generally lower for the Medes Islands population than
for gorgonians at Cap de Creus. With k
S
values ,1,
long-term extinction is inevitable, and our simulations
suggested high risk of extinction over moderate time
horizons (Fig. 4). Quasi-extinction risks correspond to
the differences in k
S
, with median extinction times of 41
and 115 yr at Medes Islands and Cap de Creus,
respectively.
The damping ratios for all matrices were quite low,
ranging from 1.02 to 1.19, and indicating slow conver-
gence to a stable size structure (Table 2). The simulated
evolution of size distributions from recently established
populations showed that the proportion in each size
class slowly converges to a mean value (see Appendix E).
In both sites an approximately stable stage distribution
was obtained after 50 yr. The observed size distribution
was quite similar at Medes Islands and Cap de Creus,
showing a dominance of small reproductive colonies
FIG. 3. Elasticity values of deterministic population growth rate for vital rates. Abbreviations are: f, fecundity; g, growth; h,
shrink into next smaller class; h
2,
shrink into class two intervals smaller; s, survival. The subscripts of the vital rates indicate the
respective size class for each vital rate.
TABLE 2. Annual multiplication rates (k
1
, deterministic
lambda; k
S
, stochastic lambda) and damping ratios (q) of
transition matrices constructed from different sites and years.
Years k
1
k
S
q
Medes Islands
20012002 0.929 1.1961
20022003 0.988 1.1289
20032004 0.935 1.0290
Mean 0.937 0.936 1.1244
Cap de Creus
20022003 0.996 1.1409
20032004 0.933 1.0559
Mean 0.974 0.974 1.1496
Note: Mean k
1
values were estimated from the average
matrix of the available transitions for each site.
FIG. 4. Cumulative distribution function (CDF) for the
time to reach a quasi-extinction threshold of 10% of initial
population. For each population 50 000 simulations were run.
CRISTINA LINARES ET AL. 924 Ecology, Vol. 88, No. 4
ranging from 10 to 20 cm. The main difference between
the sites was the greater proportion of large colonies
(.20 cm) in Cap de Creus (see Appendix E). The
observed stage distributions were not signicantly
different from the predicted stable stage distributions
for either population (Medes Islands, Kolmogorov-
Smirnov [K-S] test, D 0.167, P . 0.1; Cap de Creus,
K-S test, D0.333, P . 0.10). Some differences can be
observed, such as a small proportion in the smallest
classes compared to that predicted for both populations
and larger proportion of colonies in class 5 in the
observed distribution than in the predicted one for the
Cap de Creus site. Nonetheless, these size distributions
indicate that the Medes Islands and the Cap de Creus
populations are likely to have existed at least 50 yr.
Comparative life history
Across the species surveyed, there was a general
decrease in reproductive elasticities with increasing life
span. Relative to its life span, the red gorgonian shows a
lower importance of reproductive vital rates than do
other marine invertebrates (Fig. 5a). However, it falls
within the expected range of long-lived trees and shrubs,
species that share its basic morphology, if differing in all
other aspects of their biology. In our second compar-
ison, of age at sexual maturity vs. life span (Fig. 5b), red
gorgonian again conforms to the relationship seen for
trees, but has a late age at sexual maturity compared
with the other marine invertebrate species.
DISCUSSION
While red gorgonians are important structural com-
ponents of Mediterranean hard-bottom communities,
we know little of the basic ecologic processes determin-
ing the resilience of these populations to natural and
anthropogenic disturbances. Therefore, the rst goal of
this study was to quantify the life history and population
viability of red gorgonians and to identify the life history
traits to which population growth is most sensitive.
Red gorgonian populations show a life history pattern
of high survival, low growth rates, and delayed onset of
reproduction. Our estimates of the conditional total life
span suggest that if colonies can attain the largest class
they may be 60100 yr old. Unlike most demographic
analyses of marine organisms (e.g., Hughes 1984, Gotelli
1991, Lasker 1991, McFadden 1991), we focused our
sensitivity and elasticity analyses on the vital rates that
correspond to demographic processes, not on the matrix
elements that are combinations of these rates. As noted
by other authors (Zudeima and Franco 2001, Franco
and Silvertown 2004), vital rate elasticities are more
informative than those for matrix elements because they
correspond better to fundamental demographic process-
es (fecundity, growth, shrinkage, and survival). Our
elasticity results suggest that population growth is
overwhelmingly dependent on survival rates, principally
of reproductive colonies (.10 cm). The differences
observed between sites, with survival of larger colonies
(.20 cm) in Cap de Creus showing higher elasticity
values than those for the Medes Islands, are due to the
different proportion of larger colonies (.20 cm) at the
two sites (49% and 42%, respectively) and the higher
survival rates of these larger colonies in Cap de Creus
(see Appendix D). Red gorgonians thus conform to the
generality that decreases in adult survival rate should
more negatively effect population growth for long-lived
species with low annual reproductive success than they
will for short-lived and highly fecund species (Heppel et
al. 2000).
In contrast to survival rates, population growth for
the red gorgonian is little affected by recruitment
variation, as has been found for other gorgonian species
(Gotelli 1991, Lasker 1991). Recruitment has been
considered widely as an important factor structuring
marine communities, with high fecundity of many
marine invertebrates combining with spatial and tem-
poral patchiness in larval supply to generate spectacular
recruitment pulses and crashes in local populations
(Caffey 1985, Caley et al. 1996). In contrast to short-
lived species, populations of some long-lived sessile
species have been shown to be buffered against such
uctuations in recruitment, even when normally low
yearly rates alternate with sporadic high peaks (Yoshio-
ka 1996, Connell et al. 1997, Hughes and Tanner 2000,
FIG. 5. (a) Comparison of elasticity values of recruitment or
seedling establishment and life span (see Appendix B). (b) Log
log comparisonof age at sexual maturity andlife spanof different
marine species, herbs, shrubs, and trees (see Appendix C).
April 2007 925 DEMOGRAPHIC MODELS FOR AN OCTOCORAL
Garrabou and Harmelin 2002). Our results suggest that
this pattern is comparable for red gorgonian. For such
taxa, recruitment variation has limited importance, at
least for local populations and relatively short time
scales. Instead, survival of large colonies is a key factor
for the population persistence, in part because they
determine how long a population can persist without
recruitment. Still, recruitment patterns may be more
important over longer time scales, governing the
establishment of new populations, in particular. While
the patchy distribution of red gorgonian populations
suggests that it may exist as a metapopulation, larger
temporal and spatial scale surveys of the species are
needed to investigate the role of recruitment in the
dynamics of this long-lived organism.
One of the most striking results from our study is the
similarity in life history traits between red gorgonians
and terrestrial plants. For instance, the values of
recruitment elasticity for Paramuricea clavata are lower
than those reported for other marine organisms but are
similar to those reported for long-lived plants (Silver-
town et al. 1993, Morris and Doak 1998, Forbis and
Doak 2004). Across species, fertility and recruitment
elasticities are strongly correlated with age at matura-
tion, mean fertility, generation time, and life expectancy.
As described for other organisms (Heppell et al. 2000,
Forbis and Doak 2004), populations with high mean
adult survival rates generally have low fertility elastic-
ities and higher adult survival elasticities. In particular,
our data are consistent with the relationship between
recruitment/seedling establishment elasticities and lon-
gevity seen for other species with similar life history
traits such as herbs, shrubs, and trees (Forbis and Doak
2004, Franco and Silvertown 2004) pointing to the
existence of a clear fecundity/mortality trade-off (Fig.
5a). This trade-off is also evident if we relate the life span
with the age at maturation and we compare red
gorgonian with other similar organisms such as plants
(Fig. 5b). This result, comparable to that reported for
other gorgonian species (Lasker 1990, Gotelli 1991,
Yoshioka 1998, Garrabou and Harmelin 2002), is in
concordance with the bet-hedging theory, which predicts
adults are selected for high survival when recruitment is
highly variable because a long reproductive life is needed
to counterbalance years of high juvenile mortality
(Stearns 1992).
We also used our demographic models to investigate
current population structures and likely future popula-
tion health. Our simulations show that populations need
;50 years to approach stable stage distributions and
that the observed population structures are quite similar
from predicted stable stage distributions at both sites.
The main difference is due to the very few colonies in the
rst observable class, possibly a consequence of recent
recruitment failures. Because anthropogenic changes in
mortality rates are likely to be relatively size-indepen-
dent, any changes in demographic rates due to these
impacts probably have had little effect on size distribu-
tions, also helping to explain the good match between
our observed and predicted population stage structures.
Given the observed vital rates, both populations
appear to be in decline, with both deterministic and
stochastic growth rates less than one. The explanation
for these declines is likely to be found in anthropogenic
impacts. As noted above, while both study areas are
located within marine protected areas, both are also
heavily visited by divers. Other authors have described
an increase in red gorgonian mortality rates in areas
with high numbers of divers, caused by mechanical
disturbances (Coma et al. 2004). The differences of the
population growth rates in the studied areas also suggest
the importance of the diving impact on the population
decline. While both areas have substantial diving
activity, the Medes Islands have higher visitation,
possibly explaining the higher mortality and lower
population growth rates at this study site. However,
we have to be cautious about this explanation due to its
correlative nature.
The fact that these populations were observed 30
years ago (M. Zabala, personal communication) with a
similar distribution as observed now indicates a high
persistence of these populations in a stable state. These
results contrast with the decline and the extremely high
extinction risk predicted by the model, but this apparent
divergence may be explained by recent increases in
mortality rates due to escalating anthropogenic distur-
bances. Better understanding of the general long-term
patterns on gorgonian populations will require the study
of other populations not affected by these kinds of
perturbations.
The major limitation in our results comes from the
relatively short time period of data collection. Temporal
variability had little effect on population growth rates or
sensitivity results over our study period. While recruit-
ment variation is unlikely to substantially inuence our
results, infrequent mass mortality events have been
observed in red gorgonian populations, and these pulses
of mortality could substantially alter the size composi-
tion and dynamics we predict. Nonetheless, the model
developed in this study allows us to better understand
the demography of this species and the likely importance
of different threats to population viability.
ACKNOWLEDGMENTS
Thanks to the people of the Doak lab for helpful discussions
and above all, for a pleasant working environment. We are
grateful to Joaquim Garrabou for the continuous encourage-
ment. Support for this work was provided by a research grant
(DURSI 2004 BE 00231) to C. Linares from the Generalitat de
Catalunya. This study was funded in part by the Departa-
ment de Medi Natural of the Generalitat de Catalunya, by a
research grant CTM2004-03120/MAR from the Ministerio de
Educacio n y Ciencia of Spain to R. Coma, and by NSF DEB-
0087078 and USDA 2002-00610 to D. F. Doak.
LITERATURE CITED
Babcock, R. C. 1991. Comparative demography of three species
of scleractinian corals using age and size-dependent classi-
cations. Ecological Monographs 61:225244.
CRISTINA LINARES ET AL. 926 Ecology, Vol. 88, No. 4
Ballesteros, E. 2006. Mediterranean coralligenous assemblages:
a synthesis of present knowledge. Oceanography and Marine
Biology 44:123195.
Bavestrello, G., and F. Boero. 1986. Necrosi e rigenerazione in
Eunicella cavolinii in Mar Ligure. Bollettino dei Musei e degli
Istituti Biologici dellUniversita di Genova 52:295300.
Bavestrello, G., C. Cerrano, D. Zanzi, and R. Cattaneo-Vietti.
1997. Damage by shing activities to the Gorgonian coral
Paramuricea clavata in the Ligurian Sea. Aquatic Conserva-
tion: Marine and Freshwater Ecosystems 7:253262.
Beissinger, S. R., and D. R. McCullough, editors. 2002.
Population viability analysis. University of Chicago Press,
Chicago, Illinois, USA.
Caffey, H. M. 1985. Spatial and temporal variation in
settlement and recruitment of intertidal barnacles. Ecological
Monographs 55:313332.
Caley, M. J., M. H. Carr, M. A. Hixon, T. P. Hughes, G. P.
Jones, and B. A. Menge. 1996. Recruitment and the local
dynamics of open marine populations. Annual Review of
Ecology and Systematics 27:477500.
Caswell, H. 2001. Matrix population models: construction,
analysis, and interpretation. Second edition. Sinauer, Sunder-
land, Massachusetts, USA.
Cerrano, C., et al. 2000. A catastrophic mass mortality episode
of gorgonians and other organisms in the Ligurian Sea (NW
Mediterranean), summer 1999. Ecology Letters 3:284293.
Cochran, M. E., and S. Ellner. 1992. Simple methods for
calculating age-based life history parameters for stage-
structured populations. Ecological Monographs 62:345364.
Coma, R., C. Linares, and E. Pola. 2003. Seguiment temporal
de la gorgo` nia Paramuricea clavata de les illes Medes Exercici
2003. Pages 6779 in M. Zabala, editor. Seguiment temporal
de la` rea marina protegida de les illes Medes. Informe anual
any 2003. Departament de Medi ambient, Generalitat de
Catalunya, Barcelona, Spain.
Coma, R., C. Linares, E. Pola, and M. Zabala. 2001. Seguiment
temporal de la gorgo` nia Paramuricea clavata de les illes
Medes Exercici 2001. Pages 5982 in M. Zabala, editor.
Seguiment temporal de la` rea marina protegida de les illes
Medes. Informe anual any 2001. Departament de Medi
ambient, Generalitat de Catalunya, Barcelona, Spain.
Coma, R., E. Pola, M. Ribes, and M. Zabala. 2004. Long-term
assessment of temperate octocoral mortality patterns, pro-
tected vs. unprotected areas. Ecological Applications 14:
14661478.
Coma, R., M. Ribes, M. Zabala, and J. M. Gili. 1995a.
Reproduction and cycle of gonadal development in the
Mediterranean gorgonian Paramuricea clavata. Marine
Ecology Progress Series 117:173183.
Coma, R., M. Ribes, M. Zabala, and J. M. Gili. 1998. Growth
in a modular colonial marine invertebrate. Estuarine, Coastal
and Shelf Science 47:459470.
Coma, R., M. Zabala, and J. M. Gili. 1995b. Sexual
reproductive effort in the Mediterranean gorgonian Para-
muricea clavata. Marine Ecology Progress Series 117:173
183.
Connell, J. H., T. P. Hughes, and C. C. Wallace. 1997. A 30-
year study of coral abundance, recruitment, and disturbance
at several scales in space and time. Ecological Monographs
67:461488.
Cook, R. E. 1985. Growth and development in clonal plant
populations. Pages 259296 in J. B. C. Jackson, L. W. Buss,
and R. E. Cook, editors. Population biology and evolution of
clonal organisms. Yale University Press, New Haven,
Connecticut, USA.
Crowder, L. B., D. T. Crouse, S. S. Heppell, and T. H. Martin.
1994. Predicting the impact of turtle excluder devices on
loggerhead sea turtle populations. Ecological Applications 4:
437445.
Doak, D. F., W. F. Morris, C. Pster, B. E. Kendall, and E. M.
Bruna. 2005. Correctly estimating how environmental
stochasticity inuences tness and population growth.
American Naturalist 166:E14E21.
Doak, D. F., D. M. Thomson, and E. S. Jules. 2002. PVA for
plants: understanding the demographic consequences of seed
banks for population health. Pages 312337 in S. R.
Beissinger and D. R. McCullough, editors. Population
viability analysis. University of Chicago Press, Chicago,
Illinois, USA.
Done, T. J. 1988. Simulation of recovery of pre-disturbance size
structure in populations of Porites spp. damaged by the
crown of thorns starsh Acanthaster planci. Marine Biology
100:5161.
Forbis, T. A., and D. F. Doak. 2004. Seedling establishment
and life history trade-offs in alpine plants. American Journal
of Botany 91:11471153.
Franco, M., and J. Silvertown. 1996. Life history variation in
plants: an exploration of the fastslow continuum hypothesis.
Philosophical Transactions of the Royal Society of London B
351:13411348.
Franco, M., and J. Silvertown. 2004. A comparative demogra-
phy of plants based upon elasticities of vital rates. Ecology
85:531538.
Garrabou, J., and J. G. Harmelin. 2002. A 20-year study on
life-history traits of a harvested long-lived temperate coral in
the NW Mediterranean: insights into conservation and
management needs. Journal of Animal Ecology 71:966978.
Gerber, L. R., and S. Heppell. 2004. The use of demographic
sensitivity analysis in marine species conservation planning.
Biological Conservation 120:121128.
Gotelli, N. J. 1991. Demographic models for Leptogorgia
virgulata, a shallow-water gorgonian. Ecology 72:457467.
Heppell, S. S., H. Caswell, and L. B. Crowder. 2000. Life
histories and elasticities patterns: perturbation analysis for
species with minimal demographic data. Ecology 81:654665.
Holmes, E. E., and A. E. York. 2003. Using age structure to
detect impacts on threatened populations: a case study with
steller sea lions. Conservation Biology 17:17941806.
Hughes, T. P. 1984. Population dynamics based on individual
size rather than age: a general model with a reef coral
example. American Naturalist 123:778795.
Hughes, T. P. 1990. Recruitment limitation, mortality and
population regulation in open systems: a case study. Ecology
71:1220.
Hughes, T. P., and J. B. C. Jackson. 1985. Population dynamics
and life histories of foliaceous corals. Ecological Mono-
graphs 55:141166.
Hughes, T. P., and J. E. Tanner. 2000. Recruitment failure, life
histories and long-term decline of Caribbean corals. Ecology
81:22502263.
Jones, C. J., J. H. Lawton, and M. Shachak. 1994. Organisms
as ecosystem engineers. Oikos 69:373386.
Lasker, H. R. 1990. Clonal propagation and population
dynamics of a gorgonian coral. Ecology 71:15781589.
Lasker, H. R. 1991. Population growth of a gorgonian coral:
equilibrium and non-equilibrium sensitivity to changes in life
history variables. Oecologia 86:503509.
Linares, C., R. Coma, D. D az, M. Zabala, B. Hereu, and L.
Dantart. 2005. Immediate and delayed effects of a mass
mortality event on gorgonian population dynamics and
benthic community structure in the NW Mediterranean.
Marine Ecology Progress Series 305:127137.
McFadden, C. S. 1991. A comparative demographic analysis of
clonal reproduction in a temperate soft coral. Ecology 72:
18491866.
Morris, W. F., and D. F. Doak. 1998. Life history of the long-
lived gynodioecious cushion plant, Silene acaulis (Caryo-
phyllaceae), inferred from size-based population projection
matrices. American Journal of Botany 85:784793.
Morris, W. F., and D. F. Doak. 2002. Quantitative conserva-
tion biology: theory and practice of population viability
analysis. Sinauer, Sunderland, Massachusetts, USA.
April 2007 927 DEMOGRAPHIC MODELS FOR AN OCTOCORAL
Nakaoka, M. 1993. Yearly variation in recruitment and its
effects on population-dynamics in Yoldia notabilis (Mollusca,
Bivalvia), analyzed using projection matrix model. Research-
es on Population Ecology 35:199213.
Nakaoka, M. 1997. Demography of the marine bivalve Yoldia
notabilis in uctuating environments: an analysis using a
stochastic matrix model. Oikos 79:5968.
Perez, T., J. Garrabou, S. Sartoretto, J. G. Harmelin, P.
Francour, and J. Vacelet. 2000. Mortalite massive dinver-
te bre s marins: un e ve nement sans pre ce dent en Me diterranee
nord-occidentale. Comptes Rendus de lAcade mie des
Sciences Serie III 323:853865.
Pster, C. A., and A. Bradbury. 1996. Harvesting red sea
urchins: recent effects and future predictions. Ecological
Applications 6:298310.
Silvertown, J., M. Franco, and E. Menges. 1996. Interpretation
of elasticity matrices as an aid to the management of plant
populations for conservation. Conservation Biology 10:591
597.
Silvertown, J., M. Franco, I. Pisanty, and A. Mendoza. 1993.
Comparative plant demography: relative importance of
lifecycle components to the nite rate of increase in woody
and herbaceous perennials. Journal of Ecology 81:465476.
Stearns, S. C. 1992. The evolution of life histories. Oxford
University Press, Oxford, UK.
Thompson, P. M., B. Wilson, K. Grellier, and P. S. Hammond.
2000. Combining power analysis and population viability
analysis to compare traditional and precautionary approach-
es to conservation of coastal cetaceans. Conservation Biology
14:12531263.
True, M. A. 1970. E

tude quantitative de quatre peuplements

sciaphiles sur substrat rocheux dans la re gion marseillaise.
Bulletin de LInstitute Oce anographique du Monaco 1410:1
48.
Tuljapurkar, S. 1990. Population dynamics in variable envi-
ronments. Lecture notes in Biomathematics 85. Springer,
New York, New York, USA.
Weinberg, S. 1991. Faut-il proteger les gorgones de Me diterra-
nee ? Pages 4752 in C. F. Boudouresque, M. Avonb, and V.
Gravez, editors. Les Espe` ces a` Prote ger en Me diterranee .
GIS Posidonie, Marseille, France.
Yoshioka, P. M. 1994. Size-specic life history pattern of
shallow-water gorgonian. Journal of Experimental Marine
Biology and Ecology 184:111122.
Yoshioka, P. M. 1996. Variable recruitment and its effect on the
population and community structure of shallow-water
gorgonians. Bulletin of Marine Science 59:433443.
Yoshioka, P. M. 1998. Are large colonies a key factor in the
dynamics of gorgonian populations? Revista de Biologia
Tropical 46:137143.
Zuidema, P. A., and M. Franco. 2001. Integrating vital rate
variability into perturbation analysis: an evaluation for
matrix population models of six plant species. Journal of
Ecology 89:9951005.
APPENDIX A
Size class transition matrices for Paramuricea clavata populations at Medes Islands and Cap de Creus (Ecological Archives E088-
057-A1).
APPENDIX B
A table showing the relation between life span and elasticity of recruitment (E
rec
) (Ecological Archives E088-057-A2).
APPENDIX C
A table showing the relation between life span and age at sexual maturity (Ecological Archives E088-057-A3).
APPENDIX D
A table showing vital rates of Paramuricea clavata populations at Medes Island and Cap de Creus over the study period
(Ecological Archives E088-057-A4).
APPENDIX E
Graphs of the observed size frequency distributions of colonies and predicted size distributions of two populations of
Paramuricea clavata, at Medes Islands and Cap de Creus (Ecological Archives E088-057-A5).
CRISTINA LINARES ET AL. 928 Ecology, Vol. 88, No. 4