Forgot your password?
Document Actions
macia
Journal of Experimental Marine Biology and Ecology
291 (2003) 29 – 56
www.elsevier.com/locate/jembe
˚
Thorn fish Terapon jarbua (Forskal) predation on
juvenile white shrimp Penaeus indicus H. Milne
Edwards and brown shrimp Metapenaeus monoceros
(Fabricius): the effect of turbidity, prey density,
substrate type and pneumatophore density
A. Macia a,*, K.G.S. Abrantes b, J. Paula b
a
´
Departamento de Ciencias Biologicas, Faculdade de Ciencias, Universidade Eduardo Mondlane,
ˆ ˆ
C.P. 257 Maputo, Mozambique
b
Departamento de Zoologia e Antropologia, Faculdade de Ciencias, Universidade de Lisboa,
ˆ
1749-016 Lisbon, Portugal
Received 27 February 2002; received in revised form 9 December 2002; accepted 7 February 2003
Abstract
A series of laboratory experiments was conducted at Inhaca Island Marine Biological Station,
Mozambique, in order to assess the separate effects of turbidity, prey density, substrate type,
pneumatophore density, and the combined effects of turbidity with the latter three, on rate of predation
˚
by the thorn fish Terapon jarbua (Forskal, 1775) on white shrimp Penaeus indicus and brown shrimp
Metapenaeus monoceros.
Significant interactions between turbidity and the other three factors on shrimp predation for
both prey species were detected. Regardless of prey density, increasing turbidity decreased
predation on P. indicus, but not on M. monoceros, for which increasing densities reduced the
protective effect of turbidity. Increasing prey density increased predation on P. indicus in clear
water, and increased predation on M. monoceros in low and high, but not in intermediate turbidity
or clear water. The presence of a substrate suitable for burying decreased predation on M.
monoceros in clear water, but not in the turbidity levels used. In clear water, solely sandy-shell
substrate afforded protection to P. indicus, while in turbid water, no substrate offered significant
protection and muddy substrate even increased prey vulnerability to fish probably as a result of
increased preys’ locomotor activity. Raising pneumatophores density seems to lower the protective
* Corresponding author.
E-mail address: Adriano@zebra.uem.mz (A. Macia).
0022-0981/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0022-0981(03)00097-2
30 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
value of turbidity for both species. In clear water, only low and high structure density provided a
deterrent effect on predation on P. indicus; in turbid water, intermediate and higher structure
density increased predation. Increasing structural complexity reduced predation on M. monoceros
linearly in clear water; but in low turbid water it increased. In high turbid waters, the increase was
only significant in intermediate pneumatophore density. High structural complexities impair the
pursuing capacity of fish and thus decreased predation rates. The results indicate that the effective
provision of shelter of different habitats depends not only on the various environmental
parameters analysed, but also on the way they interact and on the behaviour of prey and predator
as well.
D 2003 Elsevier Science B.V. All rights reserved.
Keywords: Teraponidae; Fish predation; Penaeid shrimp; Mangrove; Substratum; Turbidity; Shelter; Synergistic
effects
1. Introduction
Juvenile penaeid shrimps have been reported to have a marked preference for coastal
wetlands, such as mangroves, and adjacent intertidal habitats, such as nursery areas (Staples
et al., 1985; Robertson and Duke, 1987; De Freitas, 1986; Chong et al., 1990). Some of the
reasons why juveniles select these different nursery areas are believed to be related to food
availability (litter, detritus, primary productivity) and to high survival promoted by the
abundance of protective bottom substrates and structural complexity (substrate type,
turbidity, mangrove roots, submerged macrophytes, etc.) reducing or impeding their
predation (Williams, 1958; Macnae, 1974; Minello and Zimmerman, 1983; Minello et
al., 1989; Zimmerman and Zamora, 1984; Staples et al., 1985; De Freitas, 1986; Coles et
al., 1987; Dall et al., 1990; Laprise and Blaber, 1992; Robertson and Blaber, 1992; Vance et
al., 1996; Primavera, 1997; Ronnback et al., 1999).
¨ ¨
Predation is a key mechanism in structuring and maintaining diversity and stability in
aquatic communities as one of the effects on prey populations is intraspecific resource
partitioning (Stein, 1977). Several studies have suggested that fish predation on juvenile
penaeid shrimps may be one of the most important processes causing natural mortality and
consequent recruitment variability, both of which contribute to the annual fluctuations
observed in commercial catches (Minello and Zimmerman, 1983; Minello et al., 1989; Dall
et al., 1990). Many species of fish have been identified as shrimp predators (see Minello and
Zimmerman, 1983; Dall et al., 1990). Dall et al. (1990) identified at least 14 families of fish
that predate on juvenile penaeid shrimps from bottom fish commonly found in the same
areas as penaeids.
Penaeid shrimps respond to predation pressure (as a defence strategy) by means of a
series of specific behaviours that include reducing visibility through burying in the soft
substrate or using escape movements and vegetation structures to hide when attacked
(see Main, 1987; Dall et al., 1990). Accordingly, some penaeid species remain buried in
the substratum during the day to emerge at night (Minello et al., 1987; Dall et al., 1990;
Vance, 1992; Primavera and Lebata, 1995), while others occupy highly turbid waters
(Dall et al., 1990; Chong, 1995). Primavera and Lebata (1995) found that juvenile
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 31
Metapenaeus anchistus and Metapenaeus sp. burrow much more frequently than
Penaeus merguiensis, and Penaeus monodon during the day and night, respectively.
Metapenaeus monoceros also stays mostly buried during daylight (Joshi et al., 1979),
whereas Penaeus indicus have been reported to bury very seldom, if ever, but commonly
occur in turbid waters and close to mud banks (Hughes, 1966; De Freitas, 1986; Dall et
al., 1990).
Despite the lack of comparative studies of shrimp predation in mangrove and non-
mangrove areas, the importance of structural complexity in reducing predator efficiency is
well established (Vance et al., 1990; Ronnback et al., 1999), although this shelter function is
¨ ¨
species-specific, depending on prey behaviour and predator strategies and efficiency
(Minello and Zimmerman, 1983; Primavera and Lebata, 1995; Primavera, 1997). For
instance, Penaeus species seem to depend more on structural complexity for shelter when
compared to Metapenaeus species. The lack of confinement to vegetated habitats for M.
monoceros might thus probably be explained by behavioural differences between the genus
Penaeus and Metapenaeus. Accordingly, the protective value of a given habitat is
dependent upon the specific behaviour of the predator and prey species (e.g. Minello and
Zimmerman, 1983; Loneragan et al., 1998; Primavera, 1997).
Although visual defence against predators by penaeid shrimps is well documented,
only a few studies have attempted to assess substrate type and structural complexity
(resulting from prop roots, pneumatophores, seagrass, etc.) and turbidity (see Minello et
al., 1987; Laprise and Blaber, 1992; Chong, 1995; Primavera, 1997; Primavera and
Lebata, 1995) as refugia and how these factors interact with prey density in predator
avoidance.
This study was conducted in Mozambique, where the export of penaeid shrimps is a
significant source of foreign currency income contributing to ca. 21% of the current exports
(INE, 1999). The experiments focused on the two major commercial penaeid species, the
white shrimp P. indicus and the brown M. monoceros, which together comprise 85% of the
country’s shrimp harvest (Macia, 1990; Palha de Sousa, 1996), and their juvenile stages are
found predominantly associated with mangroves and estuarine areas as well as adjacent
intertidal flats and turbid waters (De Freitas, 1986; Hughes, 1966; Ronnback et al., 2002).
¨ ¨
The main objective was to quantify predation by thorn fish (Terapon jarbua—a voracious
and common shrimp predator) on juvenile stages of these two penaeid species under the
influence of different levels of turbidity, prey density, substrate type and pneumatophore
density. These data will allow for comparison of the shelter efficiency provided to shrimps
by the different levels of the environmental factors analysed, as well as the interactive effect
of turbidity.
The experiments were designed to test the following hypotheses: (i) that predation on
both prey species decreases with increasing turbidity and (ii) that it increases with
increasing prey density; (iii) that the presence of a substrate suitable for burrowing
decreases the vulnerability of burrowing shrimps to predation; (iv) that increases in the
fine fraction of sediment affects significantly the burrowing efficiency and consequently
predation rates on burrowing shrimps; (v) that increase in above ground structural
complexity results in a decrease in mortality rate for both prey species; and (vi) that there
are significant synergistic effects between turbidity and each of the other environmental
parameters considered on predation rates.
32 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
2. Materials and methods
This study was performed between October and December 2000 at the Marine Bio-
logical Station at Inhaca Island (lat. 26j00V long. 33j00V Maputo province, Southern
S; E),
Mozambique.
2.1. Experimental design
Shrimp mortality rates from thorn fish T. jarbua predation were tested at four turbidity
levels, three prey densities, three types of substrate and three pneumatophore densities, as
well as in every combination of the three lower levels of turbidity and each level of the
other factors, except the prey density factor, where 0.64 g lÀ 1 turbidity was used as well.
The factors were analysed for the respective responses of predation mechanisms studied
separately. In all experiments there were four replicates, which were run simultaneously,
and in each group of experiments the sequence of treatments applied was chosen
randomly. This procedure avoids potential biases caused by the temperature and photo-
period modifications over time. Order of addition of thorn fish and shrimps to exper-
imental tanks, as well as order of removal from tanks, was also changed in each trial to
avoid bias.
During the 24 h preceding the experiments, fish and shrimps were held in the tanks
where the experiments would be carried out, separated by a dividing net (1-mm mesh size).
Fish – shrimp exposure was initiated by removing the net and the animals were left to
interact for 12 h, after which the fish were removed with a small dip net and the surviving
shrimps counted in order to determine the number of shrimps predated. New animals (both
fish and shrimps) were used on each trial.
2.2. Experimental animals
The fish were obtained from the wild, captured by hand trawling (8-mm mesh size) in
Saco da Inhaca during the end of ebbing tides along the small draining channels and held in
a flow-through glass aquarium 110 Â 50 Â 50 cm set outdoors and were fed daily with live
juvenile shrimps.
Shrimps were captured in the mangrove creeks fringed by Rhizophora mucronata by
means of a small beach seine net (1-mm mesh size) during low tides, usually less than 24 h
before acclimation period, except in a few cases, where the shrimps were captured 2 days
before. Prior to the acclimation period, shrimps were held in a 60-l capacity circular flow-
through tank and fed with prawn pellets.
2.3. Experimental procedure
The experiments were performed in eight circular plastic tanks 30-l capacity (50 cm
in diameter of aperture; 0.14 m2 bottom surface; 16 cm deep). Each set of four tanks
was connected in a closed system of constant and permanent flow-through seawater
maintained by means of two identical water pumps (10 W). This allowed us to run two
levels of each factor simultaneously. From a mother-tank, the water was pumped into
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 33
each tank and a collector tube conducted it back, closing the circuit. The tanks were
covered with removable light brown nets (1-mm mesh aperture), which allowed the
penetration and use of natural light and photoperiod but prevented shrimps from
jumping out of the tanks when chased by fish. No attempt was made to regulate the
amount of light entering the troughs. All the experiments were undertaken during
daylight (started between 0500 and 0600 h), as most of the shrimp and visual fish
predator species have a diel activity pattern (e.g. Moller and Jones, 1975; Minello et
¨
al., 1987). At the beginning of the study, sunrise and sunset occurred at about 0520
and 1700 h, respectively, and at the end occurred at about 0400 and 1900 h,
respectively.
In each tank, a dividing net (1-mm mesh aperture) kept fish and shrimps apart during
acclimation so that the animals could exploit the new ‘‘environment’’ and get used to each
other’s presence. These nets divided the tanks in two halves, and were suspended by
horizontal sticks and anchored to the tanks’ bottoms by means of small leads, or glued with
silicone as in the case of the structure density study. Predator and prey were stocked
simultaneously at a ratio of 1 fish:6 shrimps, except for the study of prey density factor,
where 12 and 18 penaeids per tank were used as well. Preliminary trials revealed that the
maximum feeding capacity of fish was about seven P. indicus and about six M. monoceros
juveniles for 12 h (daylight) and, consequently, this latter density was set as the reference
density for all experiments.
The 24-h acclimation period enables both the fish and shrimps to recover from
handling stress and to explore the structures in the tank. Fish starvation during acclimation
and stocking of shrimps at higher densities (43 juveniles/m2) than those in the field (5 –
15/m2 among pneumatophores in Saco da Inhaca mangrove forest, according to Ronnback ¨ ¨
et al., 2002) ensured predation on shrimps during the experiments. At the end of the
experiments, the fish were removed by means of a small dip net, the tanks were drained
and the surviving shrimps counted. All shrimps missing from the tanks were considered
predated. Preliminary trials revealed that we could consistently recover all shrimps from
tanks that did not contain predators. Shrimps’ non-predation mortality was observed in
three clear water tanks. The dead shrimps were considered as not eaten, as the fish are
known to eat dead shrimps as well as live ones in clear water conditions (author’s
observation).
All fish and shrimps were measured to the nearest millimeter (total length) prior to
each experimental set. We attempted to use fish and shrimps as similar as possible in size
in each experiment and to keep the size range of fish and shrimps as narrow as possible
to avoid problems with size-selective predation. Hence, only fish within the 7.0 –7.9 cm
and shrimps within the 2.5– 3.5 cm (P. indicus) and 2.4– 3.6 cm (M. monoceros) size
range were used (total length). The sizes of fish and shrimps used were chosen such that
the total length of prey was between 30% and 50% of that of the predator, for it is
known that this is the preferred length of penaeid prey for fish (Minello et al., 1989; Dall
et al., 1990; Minello and Zimmerman, 1991). Salinity, temperature and turbidity were
monitored throughout the experiments in each tank, including the mother-tank. The
variation of turbidity was never higher than 0.03 g lÀ 1 throughout the experiments.
Table 1 summarizes the temperature and salinity means and ranges for each set of
experiments.
34 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Table 1
Summary data on mean values and ranges of temperature and salinity in each set of experiments
Experiment Temperature (jC) Salinity (x)
Min Max Mean Min Max Mean
Turbidity
P. indicus 19.2 24.3 22.4 32.2 35.5 33.4
M. monoceros 20.2 24.0 22.7 32.4 35.2 33.2
Prey density
P. indicus 19.2 27.1 21.8 32.2 34.5 33.5
M. monoceros 21.4 24.8 22.7 32.0 35.0 33.4
Pneumatophore density
P. indicus 21.6 26.2 24.1 30.7 34.4 32.8
M. monoceros 21.6 28.0 24.9 30.7 35.0 32.7
Substrate type
P. indicus 21.6 27.7 24.4 32.2 34.8 33.5
M. monoceros 21.8 27.9 24.4 35.4 28.3 32.3
Min—minimum; Max—maximum.
2.4. Turbidity
Shrimps’ predation rate in four different turbidity levels was investigated in order to
determine to what degree turbidity provides protection. The turbidity levels tested were
0.00 (control), 0.16, 0.32 and 0.64 g lÀ 1. The initial density of shrimps considered was 6
per tank. Turbidity was produced by means of fine sediment collected in the mangrove area
and sieved with a 250-Am mesh sieve in order to retain any particle or organism that could
interfere with the experiments. The sediment was carefully added to the running water until
the intended turbidity level was achieved. A water pump (28 W) placed inside the mother-
tank maintained the particle suspension, and associated turbidity level.
2.5. Prey density
In order to test the predation rate in different prey density conditions, experiments with
6, 12 and 18 shrimps per tank (43, 86 and 129 mÀ 2) were run for each species. Three levels
of turbidity were considered on the analysis of the synergistic effects of turbidity and prey
density: 0.00, 0.16 and 0.32 g lÀ 1.
2.6. Substrate type
This experiment attempted to compare the importance of three substrate types in
protecting the shrimps from the thorn fish. The manipulation yielded four very discrete
microhabitats: no substrate (control), mud, sand and a sandy-shell mixture. Each tank
contained a 5-cm layer of sediment (except the control tanks). Three levels of turbidity
were also tested in a two-way design with the substrate types: 0.00, 0.16 and 0.32 g lÀ 1.
The three substrates chosen are representative of bottoms that are widely but not
uniformly distributed over the island. The first substrate was taken from a sand bank, Banco
Xidjane, off the southwestern coast of the island, and consisted of shell debris and gravel to
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 35
coarse sand (shell-sand mixture 33.7% F size of >2000 Am, 35.7% F of 2000 – 1000 Am,
26.5% 1000– >500 Am, 3.3% 500 – >250 Am). It was obtained with a shovel from the
superficial layer of the bottom. The second was clean sand from the beach in front of the
Marine Biology Station, which was sieved and washed (sand and fine sand 9% F size of
500 –>250 Am, 90.1% F of 250 – 125 Am and 0.7% < 125 Am F). The third consisted of
mud (100% very fine sand, silt and clay fraction F < 125 Am) from the area of R.
mucronata in Saco da Inhaca mangrove. It was sieved, washed in salt water and left to rest
for some days, until it was sufficiently compact to be used on the experiments. New
substrates were used in each trial.
According to the Wentworth grade classification (Krumbein and Sloss, 1963), the
fraction passing the 0.125-mm sieve contains very fine sand, silt and clay. The combined
weight of this fraction, expressed as a percentage of the total weight, was used for
regression with shrimp mortality. In the text, this fraction is referred to as ‘‘fine fraction of
sediment’’.
2.7. Pneumatophore density
Relative predation rates by the thorn fish were studied in four pneumatophore density
treatments: 0 (control), 15, 30 and 45 per tank. Pneumatophores 16.1 F 0.2 cm long
collected from the fringing Avicennia marina mangrove were attached to the tanks’
bottoms with silicone, in an approximately homogenized distribution. The distribution,
density and size of pneumatophores of A. marina in the field are highly variable (80 – 180/
m2) and all densities used varied from lower, equal to or higher than those found naturally
in the Saco da Inhaca mangroves. In these experiments, the dividing nets were also
attached to the tanks’ bottoms by means of silicone, and after the experiment, the fish and
remaining shrimps were recovered by draining the tanks with a filtering net, as it was not
possible to use a dip net.
Three turbidity levels were also tested in a two-way design with the pneumatophore
density: 0.00, 0.16 and 0.32 g lÀ 1.
2.8. Statistical analysis
In the experiments concerning solely the turbidity, the data were analysed using the
mean numbers of shrimps eaten in a tank over the experimental period as the observation
in a one and two-way analysis of variance (ANOVA), where the main factors of turbidity
and prey species were considered. For each of the remaining factors analysed (prey
density, substrate type and pneumatophore density), the main effect of turbidity and the
respective environmental factor were considered in a two-way ANOVA on shrimps’
mortality rates. If a significant ( p V 0.05) F value resulted from the ANOVAs, an HSD
Tukey’s multiple comparison test was used to determine which means differed signifi-
cantly. Examination of residual plots and univariate tests (Cochran C, Hartley Fmax and
Bartlett chi-square) revealed whether the assumption of homogeneity of variance was
satisfied.
Differences in mortality rates were also tested in a multiple regression-based approach.
Accordingly, in order to assess the relative importance of each parameter as a factor which
36 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
might influence predation rates, both turbidity level and each of the other factors analysed
were included in a backward stepwise multiple regression of the form:
M ¼ a þ b1 T þ b2 X þ b3 T 2 þ b4 X 2 þ b5 TX
where M is mortality, T is turbidity (g lÀ 1) and X is the other physical factor, depending on
which experiment was being analysed. The terms T 2 and X 2 represent the quadratic effects
and TX the intersection effect. Those components of the model that were not significant
were eliminated from the equation one at a time, and were included only if significant at
the 5% level and if that accounted for more than 5% of the total variation. In order to
facilitate the perception of relationships, mortality response surfaces to turbidity and each
of the others parameters were constructed using the quadratic smoothing procedure, which
fits a second-order polynomial function to the data.
3. Results
Upon contact with the thorn fish, both shrimp species showed evasion by locomotion.
Each attack elicited an escape response that consisted in a rapid flexing of the abdomen
causing a jump through the water.
3.1. Turbidity
Shrimp mortality from thorn fish predation was affected by turbidity (Table 2) and
turbidity – prey species interaction (Table 3). In general, predation decreased with increas-
ing turbidity. However, that decrease was not similar for both species (Fig. 1). In 0.00, 0.16
and 0.64 g lÀ 1, predation rates were similar for both species, but in 0.32 g lÀ 1 conditions,
less P. indicus juveniles were eaten than M. monoceros. However, the Tuckey HSD test did
not detect a significant difference at the 5% level ( p = 0.0593), probably due to the fact that
M. monoceros’ mortality rate was 0.0% in one replica, and 66.7% in the remaining three
replicas. This difference between experiments was the largest of the series and was probably
responsible for the non-significance of the test.
3.2. P. indicus
A significant negative linear regression was detected between the mortality rate (number
of shrimps predated per tank) and turbidity (g lÀ 1) ( F(1,14) = 34.184; p = 0.0000), for which
Table 2
ANOVA tests of P. indicus and M. monoceros predation in different turbidity conditions (df = degree of freedom,
MS = mean squares, F =Fmax Hartley)
Species df effect MS effect F p level
P. indicus 3 23.75 33.53 < 0.0001
M. monoceros 3 17.56 14.29 0.0003
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 37
Table 3
Two-way ANOVA on differences in mean number of shrimps eaten in each set of experiments
Experiment Source df effect MS effect F p level
Turbidity Sp 1 2.53 2.61 0.1191
T 3 38.03 39.26 < 0.0001
Sp  T 3 3.28 3.39 0.0345
Penaeus indicus Prey density n T 3 87.69 161.9 < 0.0001
D 2 6.9 12.73 0.0001
TÂD 6 0.92 1.71 0.1481
% T 3 8536.25 107.28 < 0.0001
D 2 1143.1 14.37 < 0.0001
TÂD 6 728.58 9.16 < 0.0001
Substrate type T 2 33.77 54.64 < 0.0001
S 3 24.13 39.04 < 0.0001
TÂS 6 7.63 12.35 < 0.0001
Pneumatophore T 2 21.02 40.36 < 0.0001
density P 3 11.08 21.27 < 0.0001
TÂP 6 4.41 8.47 < 0.0001
Metapenaeus Prey density n T 3 21.52 26.49 < 0.0001
monoceros D 2 15.08 18.56 < 0.0001
TÂD 6 4.42 5.44 0.0004
% T 3 3141.28 70.98 < 0.0001
D 2 2182.78 49.32 < 0.0001
TÂD 6 1485.82 33.57 < 0.0001
Substrate type T 2 5.15 8.93 0.0007
S 3 23.24 40.32 < 0.0001
TÂS 6 6.53 11.34 < 0.0001
Pneumatophore T 2 7.77 7.77 0.0016
density S 3 22.25 22.25 < 0.0001
TÂS 6 6.6 6.6 0.0001
Sp corresponds to shrimp species, T is turbidity, D is the initial density of shrimps, n corresponds to the results of
the ANOVA on number of shrimps eaten and % corresponds to the ANOVA on percentage of shrimps eaten, S is
the substrate type (analysed as percentage of fine fraction) and P is the pneumatophore density (df = degrees of
freedom, MS = mean squares, F = Fmax Hartley test).
the regression coefficient (r) is 0.89 (slope = À 7.95, intercept = 4.35). Accordingly,
significantly more P. indicus juveniles were eaten in clear water (91.7%) than in 0.16
(37.5%), 0.32 (12.5%) and 0.64 g lÀ 1 (0%) turbidity conditions ( p = 0.0022, p = 0.0001 and
p = 0.0001, respectively). Predation rates were also higher in 0.16 than in 0.64 g lÀ 1
conditions, though the differences were not significant at the 5% level ( p = 0.0593). The
number of shrimps eaten in 0.32 g lÀ 1 turbidity was not statistically different from the one
detected for 0.16 and 0.64 g lÀ 1 conditions ( p = 0.4109 and p = 0.9556, respectively),
though the predation rate tended to decrease with increasing turbidity from 0.16 up to 0.64 g
lÀ 1 (see Fig. 1).
3.3. M. monoceros
Predation rates on M. monoceros juveniles also tended to decrease with increasing
turbidity. In fact, more shrimps were eaten in clear water (91.7%) than in 0.16 (25.0%),
38 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Fig. 1. Shrimp mortality (mean F S.E.) from thorn fish predation at different prey density (6, 12 and 18/tank) and
turbidity conditions. Above: number of shrimps eaten; below: percentage of shrimps eaten.
0.32 (50.0%) and 0.64 g lÀ 1 (12.5%) conditions ( p = 0.0003, p = 0.0270 and p = 0.0001,
respectively). The number of shrimps predated increased slightly from 1.50 F 0.29 to
3.00 F 1.00 (mean F S.E.) as turbidity increased from 0.16 to 0.32 g lÀ 1 (see Fig. 1), but
results of the Tuckey HSD test were not significant for this comparison ( p = 0.4109). As
the turbidity increased from 0.32 to 0.64 g lÀ 1, the number of shrimps eaten decreased
from 3.00 F 1.00 to 0.75 F 0.25 (mean F S.E.), but again, the differences were not
statistically significant at the 5% level ( p = 0.0593).
3.4. Prey density
3.4.1. P. indicus
Although according to the ANOVA and multiple regression on number of shrimps eaten
the predation rates were only affected by turbidity and prey density, the analysis of
proportional data also detected a significant turbidity – prey density interaction (Tables 3
and 4).
In the experiments with clear water, the number of shrimps eaten increased with
increasing prey density (Fig. 1). However, significant differences at the 5% level were
only found between the experiments with 6 and 12 shrimps per tank ( p = 0.0207), but not
between 12 and 18 shrimps per tank ( p = 0.0678).
In the three remaining levels of turbidity, the number of shrimps predated was not
significantly affected by increasing prey density. However, it was possible to observe that
in 0.32 and 0.64 g lÀ 1 conditions, it seemed to increase slightly with increasing prey
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 39
Table 4
2 2
Multiple backward correlation summary for the correlation model z = a + b1X1 + b2X2 + b3 X 1 + b 4X1X2 + b5 X 2 fit
to the data of mortality rates
Experiment Penaeus indicus Metapenaeus monoceros
z = 5.18 À 26.75t + 0.11d + 28.19t2 z = 5.08 À 10.12t + 0.45xtd
Density N
2 2
RA = 0.88, F(3,44) = 112.85, p < 0.0000 RA = 0.59, F(2,45) = 34.217, p < 0.0000
z = 103.33 À 345.49t À 3.40d + 265.72t2 + 7.30td z = 96.09 À 137.21t À 3.81d + 7.03td
P
2 2
RA = 0.86, F(4,43) = 71.382, p < 0.0000 RA = 0.58, F(3,43) = 22.644, p < 0.0000
z = 5.25 À 11.79t À 0.04s + 0.13ts z = 1.98 À 0.02t
Substrate type
2 2
RA = 0.56, F(3,44) = 20.691, p < 0.0000 RA = 0.19, F(1,46) = 12.243, p < 0.0011
z = 4.79 À 9.05t À 0.22tp À 0.001p2
z = 4.70 À 13.52t À 0.04p + 0.30tp
Pneumatophore
2 2
density RA = 0.46, F(3,44) = 14.595, p < 0.000 RA = 0.23, F(3,44) = 75.6670, p < 0.002
N corresponds to the results of correlation on number of shrimps eaten, P corresponds to the correlation on
percentage of shrimps eaten, t is the turbidity level, d is the density of prey, s is the fine fraction of the sediment
2
(as percent weight) and p is the pneumatophore density. The adjusted multiple correlation coefficient (RA) was
used as a measure of the explained variation.
density. The analysis of proportional data revealed that there was a decrease of percentage
of shrimps eaten in 0.16 g lÀ 1 conditions as prey density increased from 6 (37.5%) to 12
(14.6%) and to 18 (15.3%) per tank ( p = 0.0353 and p = 0.0462, respectively). In 0.32 and
0.64 g lÀ 1 conditions, however, no differences were detected between percentages of
shrimps eaten in different prey density conditions.
In the experiments with 12 shrimps per tank, predation rates decreased with increasing
turbidity from 0.00 (60.4%) to 0.16 g lÀ 1 (14.6%) ( p = 0.0001), but remained constant as
the turbidity increased further (10.4% shrimps eaten in 0.32 g lÀ 1 and 8.3% in 0.64 g lÀ 1
turbidity conditions). Similarly, in experiments with 18 shrimps per tank, the predation rates
were higher in clear water (41.7%) when compared to 0.16 (15.3%) ( p = 0.0084), 0.32
(11.1%) ( p = 0.0014) and 0.64 g lÀ 1 (8.3%) ( p = 0.0004) conditions, and between 0.16,
0.32 and 0.64 g lÀ 1 turbidity conditions, no significant differences were detected ( p>0.05).
The results of the multiple regression analyses are presented in Table 4, together with the
2
adjusted multiple regression coefficients (RA), F ratios and probability levels. The regression
models show the variation of prey mortality (in number and percentage of shrimps eaten) in
relation to turbidity, prey density and their interaction. To illustrate the relative effects of
these two variables on shrimp predation, response surface was obtained, representing a
smoothed image of the data following the quadratic smooth procedure (Fig. 2).
3.4.2. M. monoceros
The number and percentage of M. monoceros juveniles predated were affected by
turbidity, prey density and their interaction (Tables 3 and 4). In clear water experiments, the
mean number of juvenile shrimps eaten was maximum and corresponded to the maximum
feeding capacity of thorn fish. Consequently, it was similar in the different prey densities
analysed (5.50 F 0.29) (Fig. 1). In low (0.16 g lÀ 1) turbidity condition, the number of
shrimps eaten was significantly higher in tanks with 12 (5.25 F 0.48) and 18 (4.75 F 0.25)
than in tanks with 6 shrimps (1.50 F 0.29) ( p = 0.0002 and p = 0.0007, respectively), and
there were no significant differences between mortality rates in tanks with 12 and 18
shrimps. In intermediate (0.32 g lÀ 1) and high (0.64 g lÀ 1) turbidity conditions, the number
40 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Fig. 2. Contour plots of mortality response to turbidity and prey density, representing a smoothed image of the
number (above) and percentage (below) of the eaten shrimps data following the quadratic smoothing procedure.
of shrimps eaten increased proportionally with increasing prey density. In 0.32 g lÀ 1
conditions, the number of shrimps predated was significantly higher in tanks with 18
(4.00 F 0.41—22.2%) shrimps ( p1 = p2 = 0.0001) when compared to tanks with 12
(3.50 F 0.50—29.2%) and 6 (3.00 F 1.00—50.0%). Similarly, in 0.64 g lÀ 1 conditions,
the mean number of shrimps predated increased from 6 (0.75 F 0.25) to 12 (2.00 F 0.41)
and to 18 (4.00 F 0.41), but the only significant difference was detected between the 6 and
18 shrimps per tank experiments ( p = 0.0007).
In the analyses of proportional data, an outlier was removed from the original data
regarding 0.32 g lÀ 1 experiments with 6 shrimps per tank in order to satisfy the normality
and homoscedasticity assumptions. This lead to the detection of significant differences
between predation rates in 0.32 and 0.16 g lÀ 1 and also between 0.32 and 0.64 g lÀ 1 in
tanks with 6 shrimps by the HSD Tukey test, which were not evident in the analyses
regarding solely the turbidity. Accordingly, in experiments with 6 shrimps per tank,
mortality rates in 0.00 g lÀ 1 conditions (91.6%) were higher compared to 0.16 (25.0%)
( p = 0.0001), 0.32 (50.0%) ( p = 0.0012) and 0.64 g lÀ 1 (12.5%) ( p = 0.0001) conditions,
and were also higher in 0.32 than in 0.16 and 0.64 g lÀ 1 conditions ( p1 = p2 = 0.0001).
In intermediate prey density conditions (12 per tank), predation rate only decreased in
higher turbidity conditions, and even then the number of shrimps eaten was relatively high
(3.50 F 0.50 in 0.32 and 2.00 F 0.41 shrimps eaten in 0.64 g lÀ 1 conditions). Mortality
rate was higher in clear water (45.8%) and 0.16 g lÀ 1 (43.8%) than in 0.64 g lÀ 1 turbidity
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 41
(16.7%) ( p = 0.0003 and p = 0.0007, respectively). In higher prey density conditions (18
per tank), predation rates were not affected by turbidity. Table 4 shows the results of the
multiple regression analyses and Fig. 2 illustrates the response surfaces of mortality to
variations in turbidity and prey density.
3.5. Substrate type
3.5.1. P. indicus
According to the ANOVA results, there was a significant effect of turbidity, substrate
type and their interaction on predation rates (Table 3). In muddy substrate conditions, the
water turbidity had no effect on predation rates as these were statistically similar on the
various turbidities analysed (Fig. 3). The sandy sediment, however, had no effect on
predation protection for P. indicus as it did not increase nor decrease the predation relatively
to control conditions in the entire analysed turbidity spectrum. In this manner, and as
happened in control tanks, in sandy substrate conditions, more shrimps were eaten in 0.00
(87.5%) when compared to 0.16 (20.8%) and 0.32 g lÀ 1 turbidity conditions (16.7%)
( p1 = p2 = 0.0001)—the predation curves in the experiments without sediment and with
sandy sediment are identical. In sandy-shell substrate predation rates were higher in 0.0 and
0.32 g lÀ 1 (91.7% in both cases) conditions than in 0.16 g lÀ 1 (54.2%) ( p1 = p2 = 0.0120).
In clear water conditions, only sandy-shell substrate provided a significant protection to
P. indicus juveniles, as significantly fewer shrimps were eaten in this substrate (29.2%)
when compared to control (87.5%), muddy (91.7%) and sandy substrate (91.7%)
( p = 0.0001 for all comparisons). Predation rates in the three latter conditions (control,
muddy and sandy substrates) were statistically similar. In 0.16 g lÀ 1 turbidity conditions,
greater predation rates were observed in the presence of muddy sediment (91.7%) when
compared to control (20.8%) ( p = 0.0001), sandy substrate conditions (37.5%) ( p = 0.0002)
and sandy-shell (4.2%) ( p = 0.0001), and in control when compared to sandy substrate
conditions ( p = 0.0385). In 0.32 g lÀ 1 conditions, predation was higher in tanks with
muddy sediment (54.2%) than in tanks with no sediment (21.5%) ( p = 0.0036) and with
Fig. 3. Shrimp mortality (mean F S.E.) from thorn fish predation at different turbidity and substrate type
conditions.
42 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
sandy-shell sediment (16.7%) ( p = 0.0121), but not than in sandy substrate conditions
(33.3%) ( p = 0.5281).
Table 4 shows the results of a multiple regression analysis, according to which turbidity
and fine fraction of sediment have a significant effect on P. indicus mortality from thorn
fish predation. Note that in this analysis, the data of predation rates in control tanks were
not considered. Fig. 4 illustrates the response surfaces of mortality rate as a function of
these variables (for easier identification on this relationship, percentage of fine fraction is
arcsine-transformed).
3.5.2. M. monoceros
There was a significant effect of turbidity, substrate type and their interactions on M.
monoceros juveniles’ predation rates (Table 3). In clear water conditions, more shrimps
were eaten in tanks with no sediment (91.6%) than in tanks with any type of sediment
(0.0% in mud, 20.8% in sand and 4.2% in sandy-shell conditions) ( p = 0.0001 for each
comparison) (Fig. 3). In 0.16 g lÀ 1 turbidity conditions, predation rate was lower in
muddy substrate (0.0%) when compared to sandy (50.0%), sandy-shell (37.5%) and
control experiments (41.7%) ( p = 0.0002, p = 0.0082 and p = 0.0023, respectively), but in
0.32 g lÀ 1 conditions, the presence or type of substrate had no significant effect on
predation rates.
In muddy and sandy sediment conditions, predation rates were not affected by turbidity.
In sandy-shell sediment conditions, however, these were higher in 0.16 g lÀ 1 (37.5%)
compared to 0.00 in 0.32 conditions (0.0% in both cases) ( p1 = p2 = 0.0082).
Fig. 4. Predation on shrimps (%) as a function of turbidity and substrate type (expressed as arcsine-transformed
percentage of fine fraction—ffs) based upon the quadratic smoothing procedure fitted to the mortality data.
According to the multiple backward regression, the following equations are the quadratic polynomials that best
describe the response surfaces:
pffiffiffiffiffi
– P. indicus: M ¼ 54:53 À 108:5lt þ 10:20ðsinð ffsÞÞÀ1 (RA = 0.59, p(2,33) = 25.76, p < 0.0000);
2
Fffiffiffiffiffi
pffiffiffiffiffi À1 2
– M. monoceros: M ¼ 24:96 À 2:57ððsinð ffsÞÞ Þ þ 0:12tðsinð ffsÞÞÀ1 (RA = 0.66, F(2,33) = 30.405,
2
p < 0.0000).
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 43
Table 4 shows the results of a multiple regression analysis, according to which there is a
significant effect of substrate type and turbidity –substrate type interaction on shrimp
mortality. Fig. 4 illustrates these results.
3.6. Pneumatophore density
In all experiments, fish were able to swim easily between the pneumatophores but the
pursuing activity in higher pneumatophore density conditions was more difficult.
3.6.1. P. indicus
Shrimp mortality from thorn fish predation was affected by turbidity, pneumatophore
density and their interaction (Tables 3 and 4). More shrimps were captured in the absence of
pneumatophores and in the presence of 30 pneumatophores (91.7% for both cases) than in
tanks with 15 (45.8%) ( p1 = p2 = 0.0003) or 45 pneumatophores (41.7%) ( p1 = p2 = 0.0001)
(Fig. 5).
In low turbidity conditions (0.16 g lÀ 1), fish captured P. indicus juveniles more
effectively in tanks with 30 (58.3%) than in tanks with 15 pneumatophores (20.8%)
( p = 0.0045), and in all the remaining comparisons no differences were detected. In 0.32 g
lÀ 1 conditions, greater predation was observed in intermediate and high pneumatophore
density conditions (50.0% and 45.8% mortality, respectively), compared to control (12.5%)
( p = 0.0045 and p = 0.0170, respectively). More shrimps were also eaten in conditions of 30
than in conditions of 15 pneumatophores per tank (16.7%) ( p = 0.0170).
In the experiments with 15 pneumatophores per tank, the mortality rate decreased with
increasing turbidity, though no differences were detected at the 5% level ( p = 0.0580 for the
comparison between predation in 0.32 and 0.00 g lÀ 1). In 30 pneumatophores per tank
conditions, predation rates were higher in clear water (91.7%) compared to turbid waters
(58.3% in 0.16 g lÀ 1 and 50.0% in 0.32 g lÀ 1) ( p = 0.0170 and p = 0.0012, respectively).
Turbidity had no detectable effect on predation rates in 45 pneumatophores per tank
conditions. Fig. 6 illustrates the mortality response surfaces to turbidity and pneumatophore
density.
Fig. 5. Shrimp mortality (mean F S.E.) from thorn fish predation in different turbidity and pneumatophore density
conditions.
44 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Fig. 6. Predation rates as a function of turbidity and pneumatophore density based upon the quadratic smoothing
procedure fitted to the mortality data.
3.6.2. M. monoceros
Shrimp mortality from fish predation varied according to turbidity, pneumatophore
density and their interaction (Tables 3 and 4). Fig. 6 illustrates the response surfaces of
mortality to variations on turbidity and pneumatophore density.
In clear water conditions, the vulnerability of M. monoceros shrimps to predation
decreased with increasing habitat complexity (Fig. 5). However, the only significant
difference found by the ANOVA was between predation rates in tanks with 45 and tanks
without pneumatophores ( p = 0.0010). A significant negative relation of regression between
predation rates and pneumatophore density was detected ( F(1,14) = 12.06; p = 0.0037), for
which the regression coefficient (r) is 0.68 (slope = À 0.06, intercept = 5.18).
In 0.16 g lÀ 1 conditions, the predation rates increased with increasing pneumatophore
density from 0 to 15 and to 30 per tank (25.0 – 58.3– 87.5%), though the only significant
difference found was between tanks without and tanks with 30 pneumatophores
( p = 0.0004). However, as the density of pneumatophores increased from 30 to 45 per
tank, the predation induced mortality decreased significantly (87.5 –8.3%; p = 0.0001).
Predation rates in 45 pneumatophores per tank experiments were also lower compared to
those in tanks with 15 pneumatophores ( p = 0.0071), and were not statistically different
from those in experiments with no pneumatophores ( p = 0.9530). In 0.32 g lÀ 1 conditions,
the predation rates decreased as structure density increased up to 15 per tank (50 –16.7%),
though the difference was not significant at the 5% level ( p = 0.2179). As pneumatophore
density increased from 15 to 30 per tank, predation rates also increased (16.7 –70.8%)
( p = 0.0027), and then decreased again as pneumatophore density increased up to 45 per
tank (70.8 –25.0%) ( p = 0.0183).
In the presence of 15 pneumatophores per tank, the predation rates remained constant as
the turbidity increased from 0.00 to 0.16 g lÀ 1 (54.2 – 58.3%) and decreased as it increased
from 0.16 to 0.32 g lÀ 1 (16.7%) ( p = 0.0449). In 30 and 45 pneumatophores per tank
conditions, the turbidity had no effect on predation rates.
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 45
4. Discussion
4.1. Turbidity
Turbidity provides predatory protection to P. indicus and M. monoceros juveniles as it
decreases predation rates by thorn fish, but its effect on predator– prey relationships
depends upon the behaviour of the prey. In fact, the visual reactive distance of the predator
is reduced in turbid water (as reported by Moore and Moore, 1976) and thus prey detection
capacity decreases confirming the strict reliance on visual capacities of this predator in prey
detection.
The reduced predator efficiency on capturing P. indicus and M. monoceros in turbid
waters agrees with Minello et al.’s (1987) findings for the pinfish predation on Penaeus
aztecus. However, the prey response to predation by thorn fish showed different pattern
between the two prey species. P. indicus exhibit an approximately linear negative response
to predation as far as turbidity increases while predation rate on M. monoceros decreases
with increasing turbidity up to 0.16 g lÀ 1. An increase in predation rate as water turbidity
reached 0.32 g lÀ 1 was observed, which can be related to a specific effect of turbidity on M.
monoceros juveniles behaviour. In fact, it seems to be a result of a significant increase in
locomotor activity of shrimps due to the decrease in light intensity. A similar predation
response was found by Minello et al. (1987), where high turbidity increased predation rates
of the southern flounder Paralichthys lethostigma, an ambush predator, on brown shrimp P.
aztecus. As happens with these species, many benthic invertebrates are negatively photo-
tactic and their locomotor activity decreases with increasing turbidity (e.g. Wickham and
Minckler, 1975; Minello et al., 1987; Brewer et al., 1989). Hence, at 0.32 g lÀ 1 turbidity,
the fish were not capable to see the prey perfectly, but because shrimps were more active
than in lower turbidity conditions, the encounter rates and consequently the predation rates
increased. The synchronization of the circadian activity rhythms of many shrimp species is
actually known to be controlled mainly by light intensity (e.g. Hughes, 1968; Moller and
¨
Jones, 1975), but other environmental parameters can also influence it (e.g. Fuss and Ogren,
1966; Wickham and Minckler, 1958; Wickham, 1967; Williams and Naylor, 1967). In this
manner, the adaptiveness of the visual sense of fish as a mechanism of prey detection is
apparently lowered. However, some species of fish that use visual cues are able to detect,
pursuit and capture prey at very low light intensity, and according to Minello and
Zimmerman (1983), reduced light intensity does not always restrict predation by visual
feeders during the night. On the other hand, shrimps do not avoid solely visual predators.
It also seems that there is a turbidity threshold above which the thorn fish actually
changes its foraging tactics due to a decline on its visual acuity becoming in very turbid
water conditions an ambush type predator. This type of behavioural shift by the predator is
relatively common (see Stoner, 1979, 1982; Anderson, 1984) and often lead to a real
threshold response of predation rate to turbidity (Nelson and Bonsdorff, 1990). Accord-
ingly, as the turbidity level increased from 0.32 to 0.64 g lÀ 1, the predation rates on M.
monoceros juveniles are lowered as a consequence of a decrease of the fish’s reactive
distance. By affecting prey and predator interactions, turbidity shows to plays an important
role on regulating distribution and mortality of P. indicus and M. monoceros at Saco da
Inhaca.
46 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
4.2. Prey density
As with most fish (see Minello et al., 1989; Bailey and Houde, 1989), predation rate of
thorn fish on shrimps increases with increasing prey density. This functional response is
known to be one of the two basic components of predation (Holling, 1959a,b), and in the
present study, it seems to vary according to turbidity level for both prey species, indicating
that a clear turbidity– prey density interaction on predation rate is present. These changes in
functional response probably lead to important modifications in the shape of the mortality
curve of prey in natural conditions. Tidal variations in the mangroves and adjacent intertidal
areas may affect water turbidity and cyclically modifies shrimps’ density.
Although it was not possible to clearly detect the exact shape of the prey density –
predation rate function, in clear water conditions, predation seems to increase with
increasing prey density until a plateau corresponding to fish’ maximum feeding capacity,
or satiation, is attained (linear or type I response) (Holling, 1959a,b). To this value
corresponds the number of shrimp juveniles needed to feed one thorn fish for 12 h.
Accordingly, around seven P. indicus or five/six M. monoceros juveniles are needed to feed
one of these fish for that period of time. The higher mortality rate of P. indicus in higher
density conditions can be explained by the higher probability that a fish might encounter the
prey. Predation on M. monoceros juveniles in low and high turbidity conditions also
increased probably as a result of higher encounter rates between prey and predator. In
contrast, in high turbidity conditions, the increase is much smoother due to the more gradual
increase of encounter rates between predator and prey (as the visual capacity of predator is
no longer good as previously referred) and a behavioural shift has already taken place in
which the predator changed from active chasing to ambush.
A minimum shrimp density is required before T. jarbua starts feeding on juvenile shrimps
in higher turbidity conditions. For P. indicus juveniles prey that minimum is probably
between 6 and 12 shrimps per tank (43 and 86/m2), as no shrimp was eaten in 6 shrimps per
tank conditions, and for M. monoceros juveniles that minimum is below 6 shrimps per tank.
In intermediate turbidity conditions, predation on M. monoceros juveniles was not
affected by prey density because the shrimps’ locomotor activity was higher than in lower
turbidity conditions and, hence, encounter rates between prey and predator increased.
However, this effect was reduced as prey density increased, because in higher prey density
conditions, changes in prey activity should only have negligible effect on encounter rates
with predators. Consequently, predation rates were constant along the analysed turbidity
spectrum.
M. monoceros juveniles are probably more visible to fish than P. indicus as they are dark
pigmented, and P. indicus, in addition to being a very transparent species, is a faster
swimmer (author’s observation). Thus, in higher turbidity conditions, predation rates on P.
indicus juveniles were often lower than those of M. monoceros. Previous studies with other
species have actually shown that the increased visibility associated with pigmentation can
be more important than either size or density in prey detection and selection by fish (e.g.
Zaret and Kerfoot, 1975; Kislalioglu and Gibson, 1976; Kneib, 1987). Results also seem to
indicate that the increase in locomotor activity of M. monoceros juveniles with increasing
turbidity (decreasing light intensity) is much steeper than that of P. indicus, as even in
higher turbidity conditions the number of M. monoceros juveniles eaten was elevated.
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 47
In turbid waters, the number of P. indicus eaten is not affected by increasing prey
density, at least from 6 to 18 per tank (43 – 128/m2). As only four levels of turbidity and
three levels of prey density were analysed, it was not possible to assert which functional
response corresponds exactly to each turbidity condition analysed. Hence, additional
experiments with higher levels of replication and the analysis of more levels of prey
density are needed. These findings could also indicate that predation in turbid areas
(mangrove areas) only acts as a fine-tuner of abundance when shrimp densities exceed a
given threshold (as suggested by Strong, 1984), from which it does serve a useful purpose
in holding prey population sizes low, but quickly loses its impact on the prey population
when these grow rapidly.
Effects of crowding or gregariousness of these species in nature are not well known,
though individuals are found in close proximity to each other when swimming freely,
resting on the surface, or when burrowed in the ground (personal observations). In aquaria,
the most prominent anti-predator behaviour observed in P. indicus juveniles was the
formation of both aggregations and schools. Formation of aggregation was also observed
for M. monoceros juveniles, but not schools. School and aggregation formation are
actually known to be very important anti-predator processes (e.g. Cushing and Harden
Jones, 1968; Pitcher, 1973; Taylor, 1976; Sullivan and Atchinson, 1978; Turchin and
Kareiva, 1989), as by grouping together, prey can reduce the risk of being eaten. In clear
water conditions, the fish were able to pursue and capture shrimps very intensively and,
hence, the number of shrimps eaten increased with increasing prey density, though the
school effect was still present. In low turbidity conditions, on the other hand, the fish
detection capacity was lower than in clear water and, consequently, the protection effect of
schools was increased. In this manner, the number of shrimps eaten was almost constant
along the prey density spectrum analysed. These results seem to indicate that the
functional response is dependent on prey density and that it therefore contributes to
population regulation (see Oaten and Murdoch, 1975). In intermediate and high turbidity
conditions, predation rates varied proportionally to prey density, suggesting that the
schooling behaviour was not present. The same result was found for M. monoceros
juveniles in low and high turbidity, confirming that this species does not school.
Demographic factors are thus important variables determining predator– prey interactions.
The constancy of P. indicus mortality rate with increasing turbidity from 0.16 to 0.64 g
lÀ 1 in experiments with intermediate and higher prey density suggests that as the turbidity
increases there is a compensation of the decrease in predator’s visual reactive distance with
the increase in shrimps’ locomotor activity. Consequently, the probability that a fish might
encounter the shrimps remains constant, and turbidity has no apparent effect on predation.
With 6 shrimps per tank, for instance, the mortality decreased with increasing turbidity
along the entire turbidity spectrum analysed because prey density was low and even
though the activity level of shrimps increased with increasing turbidity, encounter rates
could not increase significantly in order to maintain the plateau level.
4.3. Substrate type
Substrate type influences predation on shrimp juveniles, and hence these should prefer
substrates that offer maximum protection. Actually, several authors have identified
48 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
correlations between sediment characteristics and shrimp abundance for several species
(e.g. Williams, 1958; Moller and Jones, 1975; Branford, 1981a,b).
¨
Because P. indicus juveniles do not bury in the laboratory (Hughes, 1966; personal
observations), only sandy-shell substrate provided a deterrent effect on predation in clear
water conditions. This was probably a result of the cryptic coloration, which combined
with immobilization permitted blending in of the shrimps with their background, avoiding
detection by visual predators (see Clements and Livingston, 1984; Russo, 1987). In fact,
during preliminary trials, the animals were observed to remain immobile during the
majority of the time, except when the predator approximated. Several studies have shown
that movement of prey appears to elicit predator strikes and to evoke feeding behaviour in
a variety of visual predators (e.g. Zaret and Kerfoot, 1975; Stein and Magnuson, 1976;
Main, 1987). These results are not in accordance with P. indicus juvenile’s distribution in
nature, as according to Hughes (1966), Branford (1981a) and De Freitas (1986), they seem
to prefer very muddy areas within mangrove swamps, both in the primary channels or the
smaller creeks of the upper reaches. According to Ronnback et al. (2002) (author’s
¨ ¨
unpublished data) at Inhaca Island, P. indicus in the mangrove forest does not show any
preference to muddy substrate.
However, it is important to note that this species buries in natural conditions, though
not completely (Hughes, 1966; author’s observation). The presence of a light yellow sandy
substrate had no effect on P. indicus vulnerability to predation, probably because this
species does not bury and the substrate colour did not allow shrimps to blend with the
background. The presence of sandy-shell substrate, however, decreased shrimps’ vulner-
ability to thorn fish predation, probably as a result of a significant camouflage provided to
the semi-transparent body, which made prey detection by fish more difficult. In this
manner, the presence of this type of substrate may only affect predation by modifying the
shrimps’ activity levels. However, higher turbidity conditions already provided predation
protection. In Maputo Bay, P. indicus is mostly captured in areas of turbid waters
(commercial fisheries fleet) close to Maputo River estuary.
Whether shrimps bury, rest, perch or swim was found previously to depend on species,
size, presence or absence of vegetation or other structures and time of the day, among other
parameters (e.g. Moller and Jones, 1975; Minello et al., 1987). M. monoceros juveniles
¨
avoid detection by visual fish predators by burrowing during daylight (personal observa-
tions) and thus decreasing their apparent availability to visual feeding predators (see Minello
and Zimmerman, 1984). This anti-predation behaviour has previously been observed in
several penaeid species (see Ruello, 1973; Moller and Jones, 1975; Minello and Zimmer-
¨
man, 1983; Minello et al., 1987; Dall et al., 1990). Moller and Jones (1975), working with
¨
Penaeus semisulcatus, suggested that burrowing is triggered by increasing light intensity,
while emergence from the substratum at dusk is mainly governed by endogenous control.
The thorn fish were not able to detect or remove buried shrimps from the substratum
and only attacked shrimps lying on the bottom or swimming in the water column,
confirming again the Whitfield and Blaber’s (1978) assessment that this species is a
strictly visual feeder. However, some fish species that also rely on chemosensory abilities
can locate and capture buried shrimps by moving the sand with their snout and pelvic fins.
M. monoceros juveniles can burrow completely both in muddy and sandy substrate, but
in sandy-shell substrate, the burying is not complete as part of the dorsal area, eyes and
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 49
rostrum remain visible above the surface. The ability of brown shrimp to bury is thus
affected by the substrate characteristics just as happens with several other burrowing
penaeids (e.g. Penaeus setiferus, P. aztecus and Penaeus dourarum: Williams, 1958;
Penaeus japonicus: Egusa and Yamamoto, 1961). In clear water, the presence of any
substrate type provided deterrent effect on predation. Accordingly, the selection by
juvenile M. monoceros of a substratum suitable for burrowing reduces significantly the
predation success of fish that rely especially in visual cues for prey detection, and a soft
substratum especially protects this species in clear water as predation rates in all
substratum types were significantly different from those found in control experiments.
These findings agree with the distribution of juveniles and sub-adults in Maputo Bay and
Saco da Inhaca, where M. monoceros is more widespread than other species and can be
found in a diverse number of habitats, from areas with submerged macrophytes to the
deeper reaches of the mangrove swamps (Hughes, 1966; De Freitas, 1986; author’s
unpublished data).
In low turbidity conditions, the presence of substratum did not especially protect M.
monoceros juveniles, confirming that in the absence of sediment the shrimps are not very
active due to the high light intensity and that the fish’ visual detection capacity is lowered
due to the turbidity effect. As turbidity increased further, the M. monoceros’ vulnerability
to predation was minimum in the presence of muddy sediment, probably as a result of the
shrimps’ cryptic coloration that permits blending in with the dark background and avoid
detection. Hence, in muddy substrate conditions, predation rates were not affected by
increasing turbidity. In high turbidity conditions, the locomotor activity of shrimps was
already high due to the low light intensity and, consequently, predation rates were higher
in tanks with no sediment and with sandy-shell sediment when compared to low turbidity
conditions. Actually, Macnae and Kalk (1962) and Joshi et al. (1979) have demonstrated
that this species favours a muddy substratum, though that result was not confirmed by De
Freitas (1986) work, in which it is shown that this species can be found in a wide variety of
substrates.
In the presence of muddy and sandy sediment, turbidity had no effect on thorn fish’s
predation success, but in sandy-shell sediment, where the M. monoceros’ burrowing is
not complete, the effect of turbid water on predation rate was apparently increased by a
reduction in shrimps’ burrowing. In fact, burrowing by M. monoceros juveniles was
reduced in turbid water, as happens with P. aztecus (Minello et al., 1987) and other
penaeids. In this way, substrate type, as light intensity, also regulates the locomotor
activity rhythms of shrimps. Actually, several studies have shown that the activity
rhythms of many penaeid species are modified in the absence of substrate (e.g.
Metapenaeus bennettae, Metapenaeus macleayi and Penaeus plebejus: Racek, 1959;
P. dourarum: Fuss and Ogren, 1966; Crangon crangon: Hagerman, 1970; P. semi-
sulcatus and P. monodon: Moller and Jones, 1975). These activity rhythms that depend
¨
upon the presence and type of substrate and upon the negative phototaxy are probably
an efficient protection against visual predation, and the interaction between turbidity and
substrate type on predation rates can thus influence distribution and abundance patterns
of prey.
So, fish predation on the burying shrimps can be affected by turbidity, substrate type, as
well as by prey and predator species and their interactions.
50 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
4.4. Pneumatophore density
The use of mangrove structures (pneumatophores) by juvenile penaeid shrimps as refuge
from predation was documented for the first time by Primavera (1997). Results of the
present study indicate that, as assessed in her work, the presence of pneumatophores does
provide shrimp a significant amount of protection from fish predators. However, the quality
of refuge provided by pneumatophores in a certain microhabitat seems to depend upon its
complexity, the prey species and its density.
In clear water conditions, increased structural complexity reduced the consumption of
M. monoceros juveniles linearly, a result that is well documented by other experimental
predator –prey studies (e.g. Crowder and Cooper, 1982; Coull and Wells, 1983; Anderson,
1984; Nelson and Bonsdorff, 1990). However, this result still lacks support from the field
studies since at Saco da Inhaca M. monoceros appears in higher densities at the intertidal
flats than within mangrove areas (Ronnback et al., 2002). A thorn fish can see much
¨ ¨
further in low structure conditions, and in high structural complexity, it probably goes
through lengthy periods of search during which no shrimp prey is visible. Hence, the
structures of the environment have an important role in reducing visual contact with prey
and prey vulnerability is lower in these conditions simply because random visual
encounters between predator and prey are reduced. On the other hand, as pneumatophore
density increases, predator activity declines due to a decrease in behaviour associated with
visual contact with prey. In fact, at low structural complexities, the thorn fish were
observed to be active searchers whereas at high densities they became ambush predators,
because in these conditions the pursuit of prey by the predator seems to be inhibited. Thus,
the thorn fish also modifies its foraging tactics with changes in structural complexity,
besides turbidity. These behavioural shifts were observed both in clear and turbid water
and are probably a result of differential energy costs of the different foraging tactics in
differentially structured habitats.
According to Anderson (1984), largemouth bass Micropterus salmoides in an environ-
ment with moderate density of vegetation had higher prey encounter rates than largemouth
bass in a highly structured environment, and that the optimal behaviour in high structure
was more complex than the best foraging mode in low structure. Some authors have also
shown that the diet breadths of fish predators increase with increasing structural complex-
ity (Vince et al., 1976; Anderson, 1984). In nature, structure should then mitigate the
effect of predation on fish’s preferred prey species, possibly resulting in changes in the
overall composition of the prey community (e.g. Crowder and Cooper, 1982; Anderson,
1984).
In low turbidity conditions, increasing pneumatophore density up to 30 per tank seems
to provide predators with increasing levels of cover, enabling them to catch more M.
monoceros juveniles, as the locomotor activity of shrimps is already high due to the
turbidity conditions and so they probably ‘‘feel safe’’ to search for food and/or a suitable
substrate to burrow. In higher turbidity conditions, predation rates initially decreased with
increasing pneumatophore density up to 15 per tank because in these conditions the prey
detection and pursuing activity was more difficult. As the structural complexity increased
up to 30 pneumatophores per tank, the predation rates increased significantly, suggesting
that in those conditions the shrimps’ locomotor activity increased significantly and,
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 51
consequently, predation rates also increased. Besides, the foraging behaviour of fish has
also changed to ambush, allowing them to capture more prey. In turbid waters, as
happened in clear water, there was a break in capture rates at higher pneumatophore
density as a result of the lowered reactive distance of thorn fish and the consequent
behavioural shift.
In low structural complexity conditions, the predation rates were not affected by
increasing turbidity up to 0.16 g lÀ 1, suggesting that the shrimps’ locomotor activity
increases with increasing turbidity, as it is known that the reactive distance of fish
decreased. In intermediate and higher structural complexity the turbidity level had no
significant effect on predation rates, suggesting that the locomotor activity of shrimps is
highly increased with increasing structural complexity, and consequently as it increases,
the locomotor activity increase caused by the increasing turbidity becomes negligible.
Unlike M. monoceros, P. indicus vulnerability to predation in clear water conditions
was not a linear function of increasing habitat complexity. Actually, the results indicate
that small juvenile white shrimps would suffer higher rates of predation in conditions of
null and 30 pneumatophores density than in conditions of low and high densities, which is
apparently related to changes in prey and predator behaviour. Low structural complexity
provides shrimps with cover, allowing the predation rates to decrease, but intermediate
structure density probably provides predators with cover enabling them to catch more prey.
According to Anderson (1984), fish learn as juveniles to forage on a certain group of prey
species, or to apply different strategies in different structural complexity conditions, and
thus the observed predation response to habitat complexity can be explained by
modifications of thorn fish predator behaviour. As the survival in intermediate density
was not significantly different from that on control tanks, shrimps selecting areas with
intermediate pneumatophore density may not survive at greater rates than those found in
areas without any type of structures. Thus, it may be expected that predators would be
attracted to those intermediate structured areas, which provide them with cover, as long as
the structure density is insufficient to reduce foraging efficiency to less than it would be in
other habitats.
In high turbidity water (0.32 g lÀ 1), the cover provided to fish in intermediate and higher
levels of structural complexity seems to be highly significant. Thus, because P. indicus
juveniles are not able to detect the predator that furthermore is immobile due to the
behavioural shift, their locomotor activity increases, resulting in an increase in predation
rates. However, it is not known if P. indicus schooling behaviour changes with structural
complexity variation.
Since P. indicus juveniles do not burrow completely, in natural conditions the need for
adequate cover is great, especially since many mangrove as well as seagrass predators
consume primarily epifaunal species (Nelson, 1979; Stoner, 1979; Coen et al., 1981). In
fact, according to Ronnback et al. (2002), this species exhibits a stronger preference for
¨ ¨
mangrove forest than M. monoceros. The pneumatophores provide three dimensions
within which P. indicus juveniles may hide and space themselves. These shrimps were
observed to attach their bodies parallel to the pneumatophores when they perceived the
stalking behaviour of thorn fish and to remain immobile, avoiding detection. This strategy
would protect P. indicus juveniles from many ‘‘active chase’’ predators that require a
visual cue for prey detection. However, prey species that are protected from one search and
52 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
attack strategy (e.g. active chase or pursuit) may be extremely vulnerable to other mode of
prey capture (e.g. stalking or ambush) (e.g. Neill and Cullen, 1974; Primavera, 1997) and,
consequently, that anti-predator behaviour is not successful against all fish predation
strategies.
In low and intermediate structural complexities, the P. indicus mortality rates’ decrease
with increasing turbidity is much smoother than in no pneumatophores conditions, as a
result of an increase in shrimps’ locomotor activity due to increasing turbidity combined
with the cover provided to predator. These two factors seem to have an additive effect in
increasing prey activity and in decreasing predators’ prey visual detection and capture
capacity, so that in higher structure density the turbidity effect was not detected. Thus, the
effective provision of shelter of different habitats is highly variable and depends not only
upon structure, density, but also on the behaviour of predator and prey as well (e.g. Minello
and Zimmerman, 1983; Main, 1987; Primavera, 1997) and the way these factors interact.
5. Conclusions
Turbidity provides protection from thorn fish T. jarbua predation for both species of
shrimps studied.
The borrowing behaviour of shrimps influences the predation rate of Thorn fish.
Accordingly, the presence of a suitable substrate for burying decreases vulnerability to
predation of M. monoceros, a borrowing shrimp, but not of the non-burying P. indicus.
The presence of pneumatophores does provide shrimp a significant amount of
protection from fish predation. Thus the density and distribution of juvenile shrimps in
the field can be partly explained on basis of predation pressure (Crowder and Cooper,
1982).
Our results confirms previous findings that comparing areas with regard to their
protective capacity for juvenile shrimp are complicated as stated by Minello et al. (1987),
due to the highly significant interactions among habitats structures, predators and prey
species (different levels of each factor alter others). This study provides some information
for the assessment of the ecological value of mangroves and adjacent intertidal substrate as
well as estuaries as suitable shelter for juvenile penaeid shrimps (Minello et al., 1987;
Primavera, 1997).
Validation of these conclusions is the subject of conjectures since the laboratory con-
ditions are artificial and therefore making difficult extrapolations to natural environment.
Acknowledgements
´
We thank the Head of Marine Biological Station at Inhaca, Dr. Tomas Muacanhia, for
logistic support. In particular, Prof. Nils Kautsky, Dr. Steve Telford and two anonymous
reviewers are acknowledged for their helpful comments and English corrections on the
early version of this work. This work is funded by SIDA/SAREC and is part of the long-
term research programme ‘‘The development of biological research capacity at the
Department of Biological Sciences-Eduardo Mondlane University.’’ [AU]
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 53
References
Anderson, O., 1984. Optimal foraging by largemouth bass in structured environments. Ecology 65, 851 – 861.
Bailey, K.M., Houde, E.D., 1989. Predation on eggs and larvae of marine fishes and the recruitment problem.
Adv. Mar. Biol. 25, 1 – 83.
Branford, J.R., 1981a. Sediment preferences and morphometric equations for Penaeus monodon and Penaeus
indicus from creeks of the Red Sea. Estuar. Coast. Shelf Sci. 13, 473 – 476.
Branford, J.R., 1981b. Sediment and the distribution of penaeid shrimp in the sudanese Red Sea. Estuar. Coast.
Shelf Sci. 13, 349 – 354.
Brewer, D.T., Blaber, S.J.M., Salini, J.P., 1989. Feeding biology of Caranx bucculentus Alleyne and Macleay
(Teleostei: Carangidae) in Albatross Bay, Gulf of Carpentaria, with special reference to predation on penaeid
prawns. Aust. J. Mar. Freshw. Res. 40, 657 – 668.
Chong, V.C., 1995. The prawn – mangrove connection—fact or fallacy? Seminar on Sustainable Utilization of
Coastal Ecosystems for Agriculture, Forestry and Fisheries in Developing Regions, held in Kuala Lumpur,
Malaysia. Japan International Research Center for Agricultural Science (JIRCAS).
Chong, V.C., Sasekumar, A., Leh, M.U.C., D’Cruz, R., 1990. The fish and prawn communities of a Malaysian
coastal mangrove system, with comparisons to adjacent mud flats and inshore waters. Estuar. Coast. Shelf Sci.
31, 703 – 722.
Clements, W.H., Livingston, R.J., 1984. Prey selectivity of fringed filefish Monacanthus ciliatus (Piscis: Mon-
acantidae): role of prey accessibility. Mar. Ecol. Prog. Ser. 16, 291 – 295.
Coen, L.D., Heck Jr., K.L., Abele, L.G., 1981. Experiments on competition and predation among shrimps of
seagrass meadows. Ecology 62 (6), 1484 – 1493.
Coles, R.G., Lee Long, W.J., Squire, B.A., Squire, L.C., Bibby, J.M., 1987. Distribution of seagrasses and
Associated Juvenile Commercial Penaeid Shrimps in North-eastern Quensland Waters. Aust. J. Mar. Freshw.
Res. 38, 103 – 119.
Coull, B.C., Wells, J.B.J., 1983. Refuges from predation: experiments with phytal meiofauna from the New
Zealand rocky intertidal. Ecology 64, 1599 – 1609.
Crowder, L.B., Cooper, W.E., 1982. Habitat structural complexity and the interaction between bluegills and their
prey. Ecology 63, 1802 – 1813.
Cushing, D.H., Harden Jones, F.R., 1968. Why do fish school? Nature 218, 918 – 920.
Dall, W., Hill, D.J., Rothlisberg, P.C., Staples, D.J., 1990. The biology of the penaeidae. Adv. Mar. Biol. 27, 1 – 489
(Academic Press, London).
De Freitas, A.J., 1986. Selection of nursery areas by six southeast African Peaeidae. Estuar. Coast. Shelf Sci. 23,
901 – 908.
Egusa, S., Yamamoto, T., 1961. Studies on the respiration of the ‘kuruma’ prawn, Penaeus japonicus Bate.
Burrowing behaviour with special reference to its relation to environmental oxygen concentration. Bull. Jpn.
Soc. Sci. Fish. 27, 22 – 26.
Fuss, C.M., Ogren, L.H., 1966. Factors affecting activity and burrowing habits of the pink shrimp, Penaeus
dourarum. Biol. Bull. Mar. Biol. Lab., Woods Hole 130, 170 – 191.
Hagerman, L., 1970. Locomotory activity patterns of Crangon vulgaris (Crustacea, Natantia). Ophelia 8,
255 – 266.
Holling, C.S., 1959a. The components of predation as revealed by a study of small mammal predation of the
European pine sawfly. Can. Entomol. 91, 293 – 320.
Holling, C.S., 1959b. Some characteristics of simple types of predation and parasitism. Can. Entomol. 91,
385 – 398.
Hughes, D.A., 1966. Investigations of the nursery areas and habitat preferences of juvenile shrimps in Mozam-
bique. J. Appl. Ecol. 3, 349 – 354.
Hughes, D.A., 1968. Factors controlling emergence of pink shrimp, Penaeus dourarum, from the substrate. Biol.
Bull. Mar. Biol. Lab., Woods Hole 134, 48 – 59.
´ ´ ´
INE, 1999. Anuario Estatıstico. Instituto Nacional de Estatıstica, Mocambique. 134 pp.
¸
Joshi, P.K., Kulkarni, G.K., Nagabhushanam, R., 1979. Studies on the behaviour and substratum prefer-
ence in juveniles of the marine prawn, Metapenaeus monoceros (Fabricius). Hydrobiologia 65,
195 – 198.
54 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Kislalioglu, M., Gibson, R.N., 1976. Some factors governing prey selection by the 15-spined stickleback,
Spinachia spinachia (L.). J. Exp. Mar. Biol. Ecol. 25, 159 – 169.
Kneib, R.T., 1987. Predation risk and use of intertidal habitats by young fishes and shrimp. Ecology 68, 379 – 386.
Krumbein, W.C., Sloss, L.L., 1963. Stratigraphy and Sedimentation. Freeman, San Francisco, CA. 497 pp.
Laprise, R., Blaber, S.J.M., 1992. Predation by Moses perch, Lutjanus russelli, and blue-spotted trevally, Caranx
bucculentus, on juenile brown tiger prawn, Penaeus esculentus: effects of habitat structure and time of day.
J. Fish Biol. 40, 627 – 635.
Loneragan, N.R., Kenyon, R.A., Staples, D.J., Poiner, I.R., Conacher, C.A., 1998. The influence of seagrass type
on the distribution and abundance of postlarval and juvenile tiger prawns (Penaeus esculentus and P. semi-
sulcatus) in the western Gulf of Carpentaria, Australia. J. Exp. Mar. Biol. Ecol. 228 (2), 175 – 195.
´ ˜
Macia Jr., A.A., 1990. Estudo de duas especies de camarao (Penaeus indicus e Metapenaeus monoceros)
´
capturadas nas pescarias artesanal e semi-industrial da Baıa de Maputo. Tese de Licenciatura. Universidade
Eduardo Mondlane, Maputo. 53 pp.
Macnae, W., 1974. Mangrove Forests and Fisheries. FAO, Rome. IOFC/DEV/74/34, 35 pp.
Macnae, W., Kalk, M., 1962. The ecology of the mangrove swamps at Inhaca Island, Mocambique. J. Ecol. 50,
¸
19 – 34.
Main, K.L., 1987. Predator avoidance in seagrass meadows: prey behaviour, microhabitat selection and cryptic
coloration. Ecology 68, 170 – 180.
Minello, T.J., Zimmerman, R.J., 1983. Fish predation on juvenile brown shrimp, Penaeus aztecus Ives: the effect
of simulated Spartina structure on predation rates. J. Exp. Mar. Biol. Ecol. 72, 211 – 231.
Minello, T.J., Zimmerman, R.J., 1984. Selection for brown shrimp, Penaeus aztecus, as prey by the spotted
seatrout, Cynoscion nebulosus. Contrib. Mar. Sci. 27, 159 – 167.
Minello, T.J., Zimmerman, R.J., 1991. The role of estuarine habitats in regulating growth and survival of juvenile
penaeid shrimps. In: De Loach, P.F., Dougherty, W.J., Davidson, M.A. (Eds.), Frontiers of Shrimp Research.
Elsevier, Amsterdam, pp. 1 – 16.
Minello, T.J., Zimmerman, R.J., Martinez, E.X., 1987. Fish predation on juvenile brown shrimp Penaeus
aztecus Ives: effects of turbidity and substratum on predation rates. US Natl. Mar. Fish Serv. Fish.
Bull. 85, 59 – 70.
Minello, T.J., Zimmerman, R.J., Martinez, E.X., 1989. Mortality of young brown shrimp Penaeus aztecus in
estuarine nurseries. Trans. Am. Fish. Soc. 118, 693 – 708.
Moore, J.W., Moore, I.A., 1976. The basis of food selection in flounders, Platichthys flesus (L.), in the Seven
Estuary. J. Fish Biol. 9, 139 – 156.
Moller, T.H., Jones, D.A., 1975. Locomotory rhythms and burrowing habits of Penaeus semisulcatus (de Hann)
¨
and P. monodon (Fabricius). J. Exp. Mar. Biol. Ecol. 18, 61 – 77.
Neill, S.R., Cullen, J.M., 1974. Experiments on whether schooling by their prey affects the hunting behaviour of
cephalopods and fish predators. J. Zool., Lond. 172, 549 – 569.
Nelson, W.G., 1979. Experimental studies of selective predation on amphipods: consequences for amphipod
distribution and abundance. J. Exp. Mar. Biol. Ecol. 3, 225 – 245.
Nelson, W.G., Bonsdorff, E., 1990. Fish predation and habitat complexity: are complexity thresholds real? J. Exp.
Mar. Biol. Ecol. 141, 183 – 194.
Oaten, A., Murdoch, W.W., 1975. Functional response and stability in predator – prey systems. Am. Nat. 109,
19 – 30.
Palha de Sousa, L., 1996. Estado actual de conhecimento dos recursos pesqueiros de Mocambique, prespectivas
¸
para o futuro. In: Dias, D., Scarlet, P., Hatton, J., Macia, A. (Eds.), Proceedings do Workshop sobre: O papel
ˆ
¸˜ ˜
da Investigacao na gestao da zona costeira, Maputo, 24 e 25 de Abril 1996. Departamento de Ciencias
´
Biologicas, Universidade Eduardo Mondlane, Mozambique, pp. 50 – 58. Funded by: IDRC/NORAD.
Pitcher, T.J., 1973. The three dimensional structure of schools in the minnow, Phoxinus phoxinus. Anim. Behav.
21, 673 – 686.
Primavera, J.H., 1997. Fish predation on mangrove-associated penaeids—the role of structures and substrate.
J. Exp. Mar. Biol. Ecol. 215 (2), 205 – 216.
Primavera, J.H., Lebata, J., 1995. Diel activity patterns in Metapenaeus and Penaeus juveniles. Hydrobiologia
295, 295 – 302.
Racek, A.A., 1959. Prawn investigations in Eastern Australia. Res. Bull. St. Fish. N. S. W., No. 6, 57 pp.
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 55
Robertson, A.I., Blaber, S.J.M., 1992. Plankton, epibenthos and fish communities. In: Robertson, A.I., Alongi,
D.M. (Eds.), Tropical Mangrove Ecosystems. Coastal and Estuarine Studies, vol. 41. American Geophysical
Union, Washington, DC, pp. 63 – 100.
Robertson, A.I., Duke, N.C., 1987. Mangroves as nursery sites: comparison of abundance and species compo-
sition of fish and crustaceans in mangroves and other near shore habitats in tropical Australia. Mar. Biol. 96,
193 – 205.
Ronnback, P., Troell, M., Kautsky, N., Primavera, J.H., 1999. Distribution pattern of shrimps and fish among
¨ ¨
Avicennia and Rizophora microhabitats in the Pagbilao mangroves, Philippines. Estuar. Coast. Shelf Sci. 38,
223 – 234.
Ronnback, P., Macia, A., Amqvist, G., Schultz, L., Troell, M., 2002. Do penaeid shrimps have a preference for
¨ ¨
mangrove habitats? Distribution pattern analysis on Inhaca Island, Southern Mozambique. Estuar. Coast.
Shelf Sci. 55, 427 – 436.
Ruello, N.V., 1973. Burrowing, feeding and spatial distribution of the school prawn, Metapenaeus maclearyi
(Haswell) in the Hunger River region, Australia. J. Exp. Mar. Biol. Ecol. 13, 189 – 206.
Russo, A.R., 1987. Role of habitat complexity in mediating predation by the gray damselfish Abudefduf sordidus
on epiphytal amphipods. Mar. Ecol. Prog. Ser. 36, 101 – 105.
Staples, D.J., Vance, D.J., Heales, D.S., 1985. Habitat requirements of juvenile penaeid shrimps and their
relationship to offshore fisheries. In: Rothlisberg, P.C., Hill, B.J., Staples, D.J. (Eds.), Second Australian
National Shrimp Seminar National Prawn Seminar. CSIRO, Cleveland, pp. 47 – 54.
Stein, R.A., 1977. Selective predation, optimal foraging, and the predator – prey interaction between fish and
crayfish. Ecology 58, 1237 – 1253.
Stein, R.A., Magnuson, J.J., 1976. Behavioral response of crayfish to a fish predator. Ecology 57,
751 – 761.
Stoner, A.W., 1979. Species specific predation on amphipod crustacea by the pinfish Lagodon rhomboides:
mediation by macrophyte standing crop. Mar. Biol. 55, 201 – 207.
Stoner, A.W., 1982. The influence of benthic macrophytes on the foraging behaviour on pinfish, Lagodon
rhomboides (Linnaeus). J. Exp. Mar. Biol. Ecol. 58, 271 – 284.
Strong, D.R., 1984. Density-vague ecology and liberal population regulation in insects. In: Price, P.W., Slobod-
chikoff Pand, C.N., Gaud, W.S. (Eds.), New Ecology. Wiley, New York, pp. 313 – 327.
Sullivan, J.F., Atchinson, G.J., 1978. Predator – prey behaviour of fathead minnows, Pimphales promelas and
largemouth bass, Micropterus salmoides in a model ecosystem. J. Fish Biol. 13, 249 – 253.
Taylor, R.J., 1976. Value of clumping to prey and the evolutionary response to ambush predators. Am. Nat. 110,
13 – 29.
Turchin, P., Kareiva, P., 1989. Aggregation in Aphis varians: an effective strategy for reducing predation risk.
Ecology 70, 1008 – 1016.
Vance, D.J., 1992. Activity patterns of juvenile penaeid prawns in response to artificial tidal and day – night
cycles: a comparison of three species. Mar. Ecol. Prog. Ser. 87, 215 – 226.
Vance, D.J., Haywood, M.D.E., Staples, D.J., 1990. Use of mangrove estuary as a nursery area by postlarval and
juvenile banana prawns, Penaeus merguiensis de Man, in northern Australia. Estuar. Coast. Shelf Sci. 31,
689 – 701.
Vance, D.J., Haywood, M.D.E., Staples, D.J., 1996. How far do prawns and fish move in mangrove? Distribution
of banana prawn P. merguiensis and fish in tropical mangrove forest in northern Australia. Mar. Ecol. Prog.
Ser. 131, 115 – 124.
Vince, S., Valiela, I., Backus, M., Teal, J.M., 1976. Predation by the salt marsh killing fish Fundulus heteroclitys
(L.) in relation to prey size and habitat structure: consequences for prey distribution and abundance. J. Exp.
Mar. Biol. Ecol. 23, 255 – 266.
˚
Whitfield, A.K., Blaber, J.M., 1978. Scale-eating habits of the marine teleost Terapon jarbua (Forskal). J. Fish
Biol. 12, 61 – 70.
Wickham, D.A., 1967. Observations on the activity patterns in juveniles of the pink shrimp Penaeus dourarum.
Bull. Mar. Sci. 17, 769 – 786.
Wickham, D.A., Minckler III, F.C., 1958. Substrates as a factor in shrimp distribution. Limnol. Oceanogr. 3,
283 – 290.
Wickham, D.A., Minckler III, F.C., 1975. Laboratory observations on daily patterns of burrowing and locomotor
56 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
activity of pink shrimp, Penaeus douradum, brown shrimp Penaeus aztecus, and white shrimp, Penaeus
setiferus. Contrib. Mar. Sci. 19, 21 – 35.
Williams, A.B., 1958. Substrates as factors in shrimp distribution. Limnol. Oceanogr. vol. 3, pp. 283 – 290.
Williams, A.B., Naylor, E., 1967. Spontaneously induced rhythm of tidal periodicity in the laboratory reared
Carcinus. J. Exp. Biol. 47, 229 – 234.
Zaret, T.M., Kerfoot, W.C., 1975. Fish predation on Bosmina longirostris: body-size selection versus visibility
selection. Ecology 56, 232 – 237.
Zimmerman, R.J., Zamora, G., 1984. Selection of vegetated habitat by brown shrimp, Penaeus aztecus, in a
Galveston Bay salt marsh. Fish. Bull., U.S. 82, 325 – 336.
291 (2003) 29 – 56
www.elsevier.com/locate/jembe
˚
Thorn fish Terapon jarbua (Forskal) predation on
juvenile white shrimp Penaeus indicus H. Milne
Edwards and brown shrimp Metapenaeus monoceros
(Fabricius): the effect of turbidity, prey density,
substrate type and pneumatophore density
A. Macia a,*, K.G.S. Abrantes b, J. Paula b
a
´
Departamento de Ciencias Biologicas, Faculdade de Ciencias, Universidade Eduardo Mondlane,
ˆ ˆ
C.P. 257 Maputo, Mozambique
b
Departamento de Zoologia e Antropologia, Faculdade de Ciencias, Universidade de Lisboa,
ˆ
1749-016 Lisbon, Portugal
Received 27 February 2002; received in revised form 9 December 2002; accepted 7 February 2003
Abstract
A series of laboratory experiments was conducted at Inhaca Island Marine Biological Station,
Mozambique, in order to assess the separate effects of turbidity, prey density, substrate type,
pneumatophore density, and the combined effects of turbidity with the latter three, on rate of predation
˚
by the thorn fish Terapon jarbua (Forskal, 1775) on white shrimp Penaeus indicus and brown shrimp
Metapenaeus monoceros.
Significant interactions between turbidity and the other three factors on shrimp predation for
both prey species were detected. Regardless of prey density, increasing turbidity decreased
predation on P. indicus, but not on M. monoceros, for which increasing densities reduced the
protective effect of turbidity. Increasing prey density increased predation on P. indicus in clear
water, and increased predation on M. monoceros in low and high, but not in intermediate turbidity
or clear water. The presence of a substrate suitable for burying decreased predation on M.
monoceros in clear water, but not in the turbidity levels used. In clear water, solely sandy-shell
substrate afforded protection to P. indicus, while in turbid water, no substrate offered significant
protection and muddy substrate even increased prey vulnerability to fish probably as a result of
increased preys’ locomotor activity. Raising pneumatophores density seems to lower the protective
* Corresponding author.
E-mail address: Adriano@zebra.uem.mz (A. Macia).
0022-0981/03/$ - see front matter D 2003 Elsevier Science B.V. All rights reserved.
doi:10.1016/S0022-0981(03)00097-2
30 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
value of turbidity for both species. In clear water, only low and high structure density provided a
deterrent effect on predation on P. indicus; in turbid water, intermediate and higher structure
density increased predation. Increasing structural complexity reduced predation on M. monoceros
linearly in clear water; but in low turbid water it increased. In high turbid waters, the increase was
only significant in intermediate pneumatophore density. High structural complexities impair the
pursuing capacity of fish and thus decreased predation rates. The results indicate that the effective
provision of shelter of different habitats depends not only on the various environmental
parameters analysed, but also on the way they interact and on the behaviour of prey and predator
as well.
D 2003 Elsevier Science B.V. All rights reserved.
Keywords: Teraponidae; Fish predation; Penaeid shrimp; Mangrove; Substratum; Turbidity; Shelter; Synergistic
effects
1. Introduction
Juvenile penaeid shrimps have been reported to have a marked preference for coastal
wetlands, such as mangroves, and adjacent intertidal habitats, such as nursery areas (Staples
et al., 1985; Robertson and Duke, 1987; De Freitas, 1986; Chong et al., 1990). Some of the
reasons why juveniles select these different nursery areas are believed to be related to food
availability (litter, detritus, primary productivity) and to high survival promoted by the
abundance of protective bottom substrates and structural complexity (substrate type,
turbidity, mangrove roots, submerged macrophytes, etc.) reducing or impeding their
predation (Williams, 1958; Macnae, 1974; Minello and Zimmerman, 1983; Minello et
al., 1989; Zimmerman and Zamora, 1984; Staples et al., 1985; De Freitas, 1986; Coles et
al., 1987; Dall et al., 1990; Laprise and Blaber, 1992; Robertson and Blaber, 1992; Vance et
al., 1996; Primavera, 1997; Ronnback et al., 1999).
¨ ¨
Predation is a key mechanism in structuring and maintaining diversity and stability in
aquatic communities as one of the effects on prey populations is intraspecific resource
partitioning (Stein, 1977). Several studies have suggested that fish predation on juvenile
penaeid shrimps may be one of the most important processes causing natural mortality and
consequent recruitment variability, both of which contribute to the annual fluctuations
observed in commercial catches (Minello and Zimmerman, 1983; Minello et al., 1989; Dall
et al., 1990). Many species of fish have been identified as shrimp predators (see Minello and
Zimmerman, 1983; Dall et al., 1990). Dall et al. (1990) identified at least 14 families of fish
that predate on juvenile penaeid shrimps from bottom fish commonly found in the same
areas as penaeids.
Penaeid shrimps respond to predation pressure (as a defence strategy) by means of a
series of specific behaviours that include reducing visibility through burying in the soft
substrate or using escape movements and vegetation structures to hide when attacked
(see Main, 1987; Dall et al., 1990). Accordingly, some penaeid species remain buried in
the substratum during the day to emerge at night (Minello et al., 1987; Dall et al., 1990;
Vance, 1992; Primavera and Lebata, 1995), while others occupy highly turbid waters
(Dall et al., 1990; Chong, 1995). Primavera and Lebata (1995) found that juvenile
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 31
Metapenaeus anchistus and Metapenaeus sp. burrow much more frequently than
Penaeus merguiensis, and Penaeus monodon during the day and night, respectively.
Metapenaeus monoceros also stays mostly buried during daylight (Joshi et al., 1979),
whereas Penaeus indicus have been reported to bury very seldom, if ever, but commonly
occur in turbid waters and close to mud banks (Hughes, 1966; De Freitas, 1986; Dall et
al., 1990).
Despite the lack of comparative studies of shrimp predation in mangrove and non-
mangrove areas, the importance of structural complexity in reducing predator efficiency is
well established (Vance et al., 1990; Ronnback et al., 1999), although this shelter function is
¨ ¨
species-specific, depending on prey behaviour and predator strategies and efficiency
(Minello and Zimmerman, 1983; Primavera and Lebata, 1995; Primavera, 1997). For
instance, Penaeus species seem to depend more on structural complexity for shelter when
compared to Metapenaeus species. The lack of confinement to vegetated habitats for M.
monoceros might thus probably be explained by behavioural differences between the genus
Penaeus and Metapenaeus. Accordingly, the protective value of a given habitat is
dependent upon the specific behaviour of the predator and prey species (e.g. Minello and
Zimmerman, 1983; Loneragan et al., 1998; Primavera, 1997).
Although visual defence against predators by penaeid shrimps is well documented,
only a few studies have attempted to assess substrate type and structural complexity
(resulting from prop roots, pneumatophores, seagrass, etc.) and turbidity (see Minello et
al., 1987; Laprise and Blaber, 1992; Chong, 1995; Primavera, 1997; Primavera and
Lebata, 1995) as refugia and how these factors interact with prey density in predator
avoidance.
This study was conducted in Mozambique, where the export of penaeid shrimps is a
significant source of foreign currency income contributing to ca. 21% of the current exports
(INE, 1999). The experiments focused on the two major commercial penaeid species, the
white shrimp P. indicus and the brown M. monoceros, which together comprise 85% of the
country’s shrimp harvest (Macia, 1990; Palha de Sousa, 1996), and their juvenile stages are
found predominantly associated with mangroves and estuarine areas as well as adjacent
intertidal flats and turbid waters (De Freitas, 1986; Hughes, 1966; Ronnback et al., 2002).
¨ ¨
The main objective was to quantify predation by thorn fish (Terapon jarbua—a voracious
and common shrimp predator) on juvenile stages of these two penaeid species under the
influence of different levels of turbidity, prey density, substrate type and pneumatophore
density. These data will allow for comparison of the shelter efficiency provided to shrimps
by the different levels of the environmental factors analysed, as well as the interactive effect
of turbidity.
The experiments were designed to test the following hypotheses: (i) that predation on
both prey species decreases with increasing turbidity and (ii) that it increases with
increasing prey density; (iii) that the presence of a substrate suitable for burrowing
decreases the vulnerability of burrowing shrimps to predation; (iv) that increases in the
fine fraction of sediment affects significantly the burrowing efficiency and consequently
predation rates on burrowing shrimps; (v) that increase in above ground structural
complexity results in a decrease in mortality rate for both prey species; and (vi) that there
are significant synergistic effects between turbidity and each of the other environmental
parameters considered on predation rates.
32 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
2. Materials and methods
This study was performed between October and December 2000 at the Marine Bio-
logical Station at Inhaca Island (lat. 26j00V long. 33j00V Maputo province, Southern
S; E),
Mozambique.
2.1. Experimental design
Shrimp mortality rates from thorn fish T. jarbua predation were tested at four turbidity
levels, three prey densities, three types of substrate and three pneumatophore densities, as
well as in every combination of the three lower levels of turbidity and each level of the
other factors, except the prey density factor, where 0.64 g lÀ 1 turbidity was used as well.
The factors were analysed for the respective responses of predation mechanisms studied
separately. In all experiments there were four replicates, which were run simultaneously,
and in each group of experiments the sequence of treatments applied was chosen
randomly. This procedure avoids potential biases caused by the temperature and photo-
period modifications over time. Order of addition of thorn fish and shrimps to exper-
imental tanks, as well as order of removal from tanks, was also changed in each trial to
avoid bias.
During the 24 h preceding the experiments, fish and shrimps were held in the tanks
where the experiments would be carried out, separated by a dividing net (1-mm mesh size).
Fish – shrimp exposure was initiated by removing the net and the animals were left to
interact for 12 h, after which the fish were removed with a small dip net and the surviving
shrimps counted in order to determine the number of shrimps predated. New animals (both
fish and shrimps) were used on each trial.
2.2. Experimental animals
The fish were obtained from the wild, captured by hand trawling (8-mm mesh size) in
Saco da Inhaca during the end of ebbing tides along the small draining channels and held in
a flow-through glass aquarium 110 Â 50 Â 50 cm set outdoors and were fed daily with live
juvenile shrimps.
Shrimps were captured in the mangrove creeks fringed by Rhizophora mucronata by
means of a small beach seine net (1-mm mesh size) during low tides, usually less than 24 h
before acclimation period, except in a few cases, where the shrimps were captured 2 days
before. Prior to the acclimation period, shrimps were held in a 60-l capacity circular flow-
through tank and fed with prawn pellets.
2.3. Experimental procedure
The experiments were performed in eight circular plastic tanks 30-l capacity (50 cm
in diameter of aperture; 0.14 m2 bottom surface; 16 cm deep). Each set of four tanks
was connected in a closed system of constant and permanent flow-through seawater
maintained by means of two identical water pumps (10 W). This allowed us to run two
levels of each factor simultaneously. From a mother-tank, the water was pumped into
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 33
each tank and a collector tube conducted it back, closing the circuit. The tanks were
covered with removable light brown nets (1-mm mesh aperture), which allowed the
penetration and use of natural light and photoperiod but prevented shrimps from
jumping out of the tanks when chased by fish. No attempt was made to regulate the
amount of light entering the troughs. All the experiments were undertaken during
daylight (started between 0500 and 0600 h), as most of the shrimp and visual fish
predator species have a diel activity pattern (e.g. Moller and Jones, 1975; Minello et
¨
al., 1987). At the beginning of the study, sunrise and sunset occurred at about 0520
and 1700 h, respectively, and at the end occurred at about 0400 and 1900 h,
respectively.
In each tank, a dividing net (1-mm mesh aperture) kept fish and shrimps apart during
acclimation so that the animals could exploit the new ‘‘environment’’ and get used to each
other’s presence. These nets divided the tanks in two halves, and were suspended by
horizontal sticks and anchored to the tanks’ bottoms by means of small leads, or glued with
silicone as in the case of the structure density study. Predator and prey were stocked
simultaneously at a ratio of 1 fish:6 shrimps, except for the study of prey density factor,
where 12 and 18 penaeids per tank were used as well. Preliminary trials revealed that the
maximum feeding capacity of fish was about seven P. indicus and about six M. monoceros
juveniles for 12 h (daylight) and, consequently, this latter density was set as the reference
density for all experiments.
The 24-h acclimation period enables both the fish and shrimps to recover from
handling stress and to explore the structures in the tank. Fish starvation during acclimation
and stocking of shrimps at higher densities (43 juveniles/m2) than those in the field (5 –
15/m2 among pneumatophores in Saco da Inhaca mangrove forest, according to Ronnback ¨ ¨
et al., 2002) ensured predation on shrimps during the experiments. At the end of the
experiments, the fish were removed by means of a small dip net, the tanks were drained
and the surviving shrimps counted. All shrimps missing from the tanks were considered
predated. Preliminary trials revealed that we could consistently recover all shrimps from
tanks that did not contain predators. Shrimps’ non-predation mortality was observed in
three clear water tanks. The dead shrimps were considered as not eaten, as the fish are
known to eat dead shrimps as well as live ones in clear water conditions (author’s
observation).
All fish and shrimps were measured to the nearest millimeter (total length) prior to
each experimental set. We attempted to use fish and shrimps as similar as possible in size
in each experiment and to keep the size range of fish and shrimps as narrow as possible
to avoid problems with size-selective predation. Hence, only fish within the 7.0 –7.9 cm
and shrimps within the 2.5– 3.5 cm (P. indicus) and 2.4– 3.6 cm (M. monoceros) size
range were used (total length). The sizes of fish and shrimps used were chosen such that
the total length of prey was between 30% and 50% of that of the predator, for it is
known that this is the preferred length of penaeid prey for fish (Minello et al., 1989; Dall
et al., 1990; Minello and Zimmerman, 1991). Salinity, temperature and turbidity were
monitored throughout the experiments in each tank, including the mother-tank. The
variation of turbidity was never higher than 0.03 g lÀ 1 throughout the experiments.
Table 1 summarizes the temperature and salinity means and ranges for each set of
experiments.
34 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Table 1
Summary data on mean values and ranges of temperature and salinity in each set of experiments
Experiment Temperature (jC) Salinity (x)
Min Max Mean Min Max Mean
Turbidity
P. indicus 19.2 24.3 22.4 32.2 35.5 33.4
M. monoceros 20.2 24.0 22.7 32.4 35.2 33.2
Prey density
P. indicus 19.2 27.1 21.8 32.2 34.5 33.5
M. monoceros 21.4 24.8 22.7 32.0 35.0 33.4
Pneumatophore density
P. indicus 21.6 26.2 24.1 30.7 34.4 32.8
M. monoceros 21.6 28.0 24.9 30.7 35.0 32.7
Substrate type
P. indicus 21.6 27.7 24.4 32.2 34.8 33.5
M. monoceros 21.8 27.9 24.4 35.4 28.3 32.3
Min—minimum; Max—maximum.
2.4. Turbidity
Shrimps’ predation rate in four different turbidity levels was investigated in order to
determine to what degree turbidity provides protection. The turbidity levels tested were
0.00 (control), 0.16, 0.32 and 0.64 g lÀ 1. The initial density of shrimps considered was 6
per tank. Turbidity was produced by means of fine sediment collected in the mangrove area
and sieved with a 250-Am mesh sieve in order to retain any particle or organism that could
interfere with the experiments. The sediment was carefully added to the running water until
the intended turbidity level was achieved. A water pump (28 W) placed inside the mother-
tank maintained the particle suspension, and associated turbidity level.
2.5. Prey density
In order to test the predation rate in different prey density conditions, experiments with
6, 12 and 18 shrimps per tank (43, 86 and 129 mÀ 2) were run for each species. Three levels
of turbidity were considered on the analysis of the synergistic effects of turbidity and prey
density: 0.00, 0.16 and 0.32 g lÀ 1.
2.6. Substrate type
This experiment attempted to compare the importance of three substrate types in
protecting the shrimps from the thorn fish. The manipulation yielded four very discrete
microhabitats: no substrate (control), mud, sand and a sandy-shell mixture. Each tank
contained a 5-cm layer of sediment (except the control tanks). Three levels of turbidity
were also tested in a two-way design with the substrate types: 0.00, 0.16 and 0.32 g lÀ 1.
The three substrates chosen are representative of bottoms that are widely but not
uniformly distributed over the island. The first substrate was taken from a sand bank, Banco
Xidjane, off the southwestern coast of the island, and consisted of shell debris and gravel to
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 35
coarse sand (shell-sand mixture 33.7% F size of >2000 Am, 35.7% F of 2000 – 1000 Am,
26.5% 1000– >500 Am, 3.3% 500 – >250 Am). It was obtained with a shovel from the
superficial layer of the bottom. The second was clean sand from the beach in front of the
Marine Biology Station, which was sieved and washed (sand and fine sand 9% F size of
500 –>250 Am, 90.1% F of 250 – 125 Am and 0.7% < 125 Am F). The third consisted of
mud (100% very fine sand, silt and clay fraction F < 125 Am) from the area of R.
mucronata in Saco da Inhaca mangrove. It was sieved, washed in salt water and left to rest
for some days, until it was sufficiently compact to be used on the experiments. New
substrates were used in each trial.
According to the Wentworth grade classification (Krumbein and Sloss, 1963), the
fraction passing the 0.125-mm sieve contains very fine sand, silt and clay. The combined
weight of this fraction, expressed as a percentage of the total weight, was used for
regression with shrimp mortality. In the text, this fraction is referred to as ‘‘fine fraction of
sediment’’.
2.7. Pneumatophore density
Relative predation rates by the thorn fish were studied in four pneumatophore density
treatments: 0 (control), 15, 30 and 45 per tank. Pneumatophores 16.1 F 0.2 cm long
collected from the fringing Avicennia marina mangrove were attached to the tanks’
bottoms with silicone, in an approximately homogenized distribution. The distribution,
density and size of pneumatophores of A. marina in the field are highly variable (80 – 180/
m2) and all densities used varied from lower, equal to or higher than those found naturally
in the Saco da Inhaca mangroves. In these experiments, the dividing nets were also
attached to the tanks’ bottoms by means of silicone, and after the experiment, the fish and
remaining shrimps were recovered by draining the tanks with a filtering net, as it was not
possible to use a dip net.
Three turbidity levels were also tested in a two-way design with the pneumatophore
density: 0.00, 0.16 and 0.32 g lÀ 1.
2.8. Statistical analysis
In the experiments concerning solely the turbidity, the data were analysed using the
mean numbers of shrimps eaten in a tank over the experimental period as the observation
in a one and two-way analysis of variance (ANOVA), where the main factors of turbidity
and prey species were considered. For each of the remaining factors analysed (prey
density, substrate type and pneumatophore density), the main effect of turbidity and the
respective environmental factor were considered in a two-way ANOVA on shrimps’
mortality rates. If a significant ( p V 0.05) F value resulted from the ANOVAs, an HSD
Tukey’s multiple comparison test was used to determine which means differed signifi-
cantly. Examination of residual plots and univariate tests (Cochran C, Hartley Fmax and
Bartlett chi-square) revealed whether the assumption of homogeneity of variance was
satisfied.
Differences in mortality rates were also tested in a multiple regression-based approach.
Accordingly, in order to assess the relative importance of each parameter as a factor which
36 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
might influence predation rates, both turbidity level and each of the other factors analysed
were included in a backward stepwise multiple regression of the form:
M ¼ a þ b1 T þ b2 X þ b3 T 2 þ b4 X 2 þ b5 TX
where M is mortality, T is turbidity (g lÀ 1) and X is the other physical factor, depending on
which experiment was being analysed. The terms T 2 and X 2 represent the quadratic effects
and TX the intersection effect. Those components of the model that were not significant
were eliminated from the equation one at a time, and were included only if significant at
the 5% level and if that accounted for more than 5% of the total variation. In order to
facilitate the perception of relationships, mortality response surfaces to turbidity and each
of the others parameters were constructed using the quadratic smoothing procedure, which
fits a second-order polynomial function to the data.
3. Results
Upon contact with the thorn fish, both shrimp species showed evasion by locomotion.
Each attack elicited an escape response that consisted in a rapid flexing of the abdomen
causing a jump through the water.
3.1. Turbidity
Shrimp mortality from thorn fish predation was affected by turbidity (Table 2) and
turbidity – prey species interaction (Table 3). In general, predation decreased with increas-
ing turbidity. However, that decrease was not similar for both species (Fig. 1). In 0.00, 0.16
and 0.64 g lÀ 1, predation rates were similar for both species, but in 0.32 g lÀ 1 conditions,
less P. indicus juveniles were eaten than M. monoceros. However, the Tuckey HSD test did
not detect a significant difference at the 5% level ( p = 0.0593), probably due to the fact that
M. monoceros’ mortality rate was 0.0% in one replica, and 66.7% in the remaining three
replicas. This difference between experiments was the largest of the series and was probably
responsible for the non-significance of the test.
3.2. P. indicus
A significant negative linear regression was detected between the mortality rate (number
of shrimps predated per tank) and turbidity (g lÀ 1) ( F(1,14) = 34.184; p = 0.0000), for which
Table 2
ANOVA tests of P. indicus and M. monoceros predation in different turbidity conditions (df = degree of freedom,
MS = mean squares, F =Fmax Hartley)
Species df effect MS effect F p level
P. indicus 3 23.75 33.53 < 0.0001
M. monoceros 3 17.56 14.29 0.0003
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 37
Table 3
Two-way ANOVA on differences in mean number of shrimps eaten in each set of experiments
Experiment Source df effect MS effect F p level
Turbidity Sp 1 2.53 2.61 0.1191
T 3 38.03 39.26 < 0.0001
Sp  T 3 3.28 3.39 0.0345
Penaeus indicus Prey density n T 3 87.69 161.9 < 0.0001
D 2 6.9 12.73 0.0001
TÂD 6 0.92 1.71 0.1481
% T 3 8536.25 107.28 < 0.0001
D 2 1143.1 14.37 < 0.0001
TÂD 6 728.58 9.16 < 0.0001
Substrate type T 2 33.77 54.64 < 0.0001
S 3 24.13 39.04 < 0.0001
TÂS 6 7.63 12.35 < 0.0001
Pneumatophore T 2 21.02 40.36 < 0.0001
density P 3 11.08 21.27 < 0.0001
TÂP 6 4.41 8.47 < 0.0001
Metapenaeus Prey density n T 3 21.52 26.49 < 0.0001
monoceros D 2 15.08 18.56 < 0.0001
TÂD 6 4.42 5.44 0.0004
% T 3 3141.28 70.98 < 0.0001
D 2 2182.78 49.32 < 0.0001
TÂD 6 1485.82 33.57 < 0.0001
Substrate type T 2 5.15 8.93 0.0007
S 3 23.24 40.32 < 0.0001
TÂS 6 6.53 11.34 < 0.0001
Pneumatophore T 2 7.77 7.77 0.0016
density S 3 22.25 22.25 < 0.0001
TÂS 6 6.6 6.6 0.0001
Sp corresponds to shrimp species, T is turbidity, D is the initial density of shrimps, n corresponds to the results of
the ANOVA on number of shrimps eaten and % corresponds to the ANOVA on percentage of shrimps eaten, S is
the substrate type (analysed as percentage of fine fraction) and P is the pneumatophore density (df = degrees of
freedom, MS = mean squares, F = Fmax Hartley test).
the regression coefficient (r) is 0.89 (slope = À 7.95, intercept = 4.35). Accordingly,
significantly more P. indicus juveniles were eaten in clear water (91.7%) than in 0.16
(37.5%), 0.32 (12.5%) and 0.64 g lÀ 1 (0%) turbidity conditions ( p = 0.0022, p = 0.0001 and
p = 0.0001, respectively). Predation rates were also higher in 0.16 than in 0.64 g lÀ 1
conditions, though the differences were not significant at the 5% level ( p = 0.0593). The
number of shrimps eaten in 0.32 g lÀ 1 turbidity was not statistically different from the one
detected for 0.16 and 0.64 g lÀ 1 conditions ( p = 0.4109 and p = 0.9556, respectively),
though the predation rate tended to decrease with increasing turbidity from 0.16 up to 0.64 g
lÀ 1 (see Fig. 1).
3.3. M. monoceros
Predation rates on M. monoceros juveniles also tended to decrease with increasing
turbidity. In fact, more shrimps were eaten in clear water (91.7%) than in 0.16 (25.0%),
38 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Fig. 1. Shrimp mortality (mean F S.E.) from thorn fish predation at different prey density (6, 12 and 18/tank) and
turbidity conditions. Above: number of shrimps eaten; below: percentage of shrimps eaten.
0.32 (50.0%) and 0.64 g lÀ 1 (12.5%) conditions ( p = 0.0003, p = 0.0270 and p = 0.0001,
respectively). The number of shrimps predated increased slightly from 1.50 F 0.29 to
3.00 F 1.00 (mean F S.E.) as turbidity increased from 0.16 to 0.32 g lÀ 1 (see Fig. 1), but
results of the Tuckey HSD test were not significant for this comparison ( p = 0.4109). As
the turbidity increased from 0.32 to 0.64 g lÀ 1, the number of shrimps eaten decreased
from 3.00 F 1.00 to 0.75 F 0.25 (mean F S.E.), but again, the differences were not
statistically significant at the 5% level ( p = 0.0593).
3.4. Prey density
3.4.1. P. indicus
Although according to the ANOVA and multiple regression on number of shrimps eaten
the predation rates were only affected by turbidity and prey density, the analysis of
proportional data also detected a significant turbidity – prey density interaction (Tables 3
and 4).
In the experiments with clear water, the number of shrimps eaten increased with
increasing prey density (Fig. 1). However, significant differences at the 5% level were
only found between the experiments with 6 and 12 shrimps per tank ( p = 0.0207), but not
between 12 and 18 shrimps per tank ( p = 0.0678).
In the three remaining levels of turbidity, the number of shrimps predated was not
significantly affected by increasing prey density. However, it was possible to observe that
in 0.32 and 0.64 g lÀ 1 conditions, it seemed to increase slightly with increasing prey
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 39
Table 4
2 2
Multiple backward correlation summary for the correlation model z = a + b1X1 + b2X2 + b3 X 1 + b 4X1X2 + b5 X 2 fit
to the data of mortality rates
Experiment Penaeus indicus Metapenaeus monoceros
z = 5.18 À 26.75t + 0.11d + 28.19t2 z = 5.08 À 10.12t + 0.45xtd
Density N
2 2
RA = 0.88, F(3,44) = 112.85, p < 0.0000 RA = 0.59, F(2,45) = 34.217, p < 0.0000
z = 103.33 À 345.49t À 3.40d + 265.72t2 + 7.30td z = 96.09 À 137.21t À 3.81d + 7.03td
P
2 2
RA = 0.86, F(4,43) = 71.382, p < 0.0000 RA = 0.58, F(3,43) = 22.644, p < 0.0000
z = 5.25 À 11.79t À 0.04s + 0.13ts z = 1.98 À 0.02t
Substrate type
2 2
RA = 0.56, F(3,44) = 20.691, p < 0.0000 RA = 0.19, F(1,46) = 12.243, p < 0.0011
z = 4.79 À 9.05t À 0.22tp À 0.001p2
z = 4.70 À 13.52t À 0.04p + 0.30tp
Pneumatophore
2 2
density RA = 0.46, F(3,44) = 14.595, p < 0.000 RA = 0.23, F(3,44) = 75.6670, p < 0.002
N corresponds to the results of correlation on number of shrimps eaten, P corresponds to the correlation on
percentage of shrimps eaten, t is the turbidity level, d is the density of prey, s is the fine fraction of the sediment
2
(as percent weight) and p is the pneumatophore density. The adjusted multiple correlation coefficient (RA) was
used as a measure of the explained variation.
density. The analysis of proportional data revealed that there was a decrease of percentage
of shrimps eaten in 0.16 g lÀ 1 conditions as prey density increased from 6 (37.5%) to 12
(14.6%) and to 18 (15.3%) per tank ( p = 0.0353 and p = 0.0462, respectively). In 0.32 and
0.64 g lÀ 1 conditions, however, no differences were detected between percentages of
shrimps eaten in different prey density conditions.
In the experiments with 12 shrimps per tank, predation rates decreased with increasing
turbidity from 0.00 (60.4%) to 0.16 g lÀ 1 (14.6%) ( p = 0.0001), but remained constant as
the turbidity increased further (10.4% shrimps eaten in 0.32 g lÀ 1 and 8.3% in 0.64 g lÀ 1
turbidity conditions). Similarly, in experiments with 18 shrimps per tank, the predation rates
were higher in clear water (41.7%) when compared to 0.16 (15.3%) ( p = 0.0084), 0.32
(11.1%) ( p = 0.0014) and 0.64 g lÀ 1 (8.3%) ( p = 0.0004) conditions, and between 0.16,
0.32 and 0.64 g lÀ 1 turbidity conditions, no significant differences were detected ( p>0.05).
The results of the multiple regression analyses are presented in Table 4, together with the
2
adjusted multiple regression coefficients (RA), F ratios and probability levels. The regression
models show the variation of prey mortality (in number and percentage of shrimps eaten) in
relation to turbidity, prey density and their interaction. To illustrate the relative effects of
these two variables on shrimp predation, response surface was obtained, representing a
smoothed image of the data following the quadratic smooth procedure (Fig. 2).
3.4.2. M. monoceros
The number and percentage of M. monoceros juveniles predated were affected by
turbidity, prey density and their interaction (Tables 3 and 4). In clear water experiments, the
mean number of juvenile shrimps eaten was maximum and corresponded to the maximum
feeding capacity of thorn fish. Consequently, it was similar in the different prey densities
analysed (5.50 F 0.29) (Fig. 1). In low (0.16 g lÀ 1) turbidity condition, the number of
shrimps eaten was significantly higher in tanks with 12 (5.25 F 0.48) and 18 (4.75 F 0.25)
than in tanks with 6 shrimps (1.50 F 0.29) ( p = 0.0002 and p = 0.0007, respectively), and
there were no significant differences between mortality rates in tanks with 12 and 18
shrimps. In intermediate (0.32 g lÀ 1) and high (0.64 g lÀ 1) turbidity conditions, the number
40 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Fig. 2. Contour plots of mortality response to turbidity and prey density, representing a smoothed image of the
number (above) and percentage (below) of the eaten shrimps data following the quadratic smoothing procedure.
of shrimps eaten increased proportionally with increasing prey density. In 0.32 g lÀ 1
conditions, the number of shrimps predated was significantly higher in tanks with 18
(4.00 F 0.41—22.2%) shrimps ( p1 = p2 = 0.0001) when compared to tanks with 12
(3.50 F 0.50—29.2%) and 6 (3.00 F 1.00—50.0%). Similarly, in 0.64 g lÀ 1 conditions,
the mean number of shrimps predated increased from 6 (0.75 F 0.25) to 12 (2.00 F 0.41)
and to 18 (4.00 F 0.41), but the only significant difference was detected between the 6 and
18 shrimps per tank experiments ( p = 0.0007).
In the analyses of proportional data, an outlier was removed from the original data
regarding 0.32 g lÀ 1 experiments with 6 shrimps per tank in order to satisfy the normality
and homoscedasticity assumptions. This lead to the detection of significant differences
between predation rates in 0.32 and 0.16 g lÀ 1 and also between 0.32 and 0.64 g lÀ 1 in
tanks with 6 shrimps by the HSD Tukey test, which were not evident in the analyses
regarding solely the turbidity. Accordingly, in experiments with 6 shrimps per tank,
mortality rates in 0.00 g lÀ 1 conditions (91.6%) were higher compared to 0.16 (25.0%)
( p = 0.0001), 0.32 (50.0%) ( p = 0.0012) and 0.64 g lÀ 1 (12.5%) ( p = 0.0001) conditions,
and were also higher in 0.32 than in 0.16 and 0.64 g lÀ 1 conditions ( p1 = p2 = 0.0001).
In intermediate prey density conditions (12 per tank), predation rate only decreased in
higher turbidity conditions, and even then the number of shrimps eaten was relatively high
(3.50 F 0.50 in 0.32 and 2.00 F 0.41 shrimps eaten in 0.64 g lÀ 1 conditions). Mortality
rate was higher in clear water (45.8%) and 0.16 g lÀ 1 (43.8%) than in 0.64 g lÀ 1 turbidity
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 41
(16.7%) ( p = 0.0003 and p = 0.0007, respectively). In higher prey density conditions (18
per tank), predation rates were not affected by turbidity. Table 4 shows the results of the
multiple regression analyses and Fig. 2 illustrates the response surfaces of mortality to
variations in turbidity and prey density.
3.5. Substrate type
3.5.1. P. indicus
According to the ANOVA results, there was a significant effect of turbidity, substrate
type and their interaction on predation rates (Table 3). In muddy substrate conditions, the
water turbidity had no effect on predation rates as these were statistically similar on the
various turbidities analysed (Fig. 3). The sandy sediment, however, had no effect on
predation protection for P. indicus as it did not increase nor decrease the predation relatively
to control conditions in the entire analysed turbidity spectrum. In this manner, and as
happened in control tanks, in sandy substrate conditions, more shrimps were eaten in 0.00
(87.5%) when compared to 0.16 (20.8%) and 0.32 g lÀ 1 turbidity conditions (16.7%)
( p1 = p2 = 0.0001)—the predation curves in the experiments without sediment and with
sandy sediment are identical. In sandy-shell substrate predation rates were higher in 0.0 and
0.32 g lÀ 1 (91.7% in both cases) conditions than in 0.16 g lÀ 1 (54.2%) ( p1 = p2 = 0.0120).
In clear water conditions, only sandy-shell substrate provided a significant protection to
P. indicus juveniles, as significantly fewer shrimps were eaten in this substrate (29.2%)
when compared to control (87.5%), muddy (91.7%) and sandy substrate (91.7%)
( p = 0.0001 for all comparisons). Predation rates in the three latter conditions (control,
muddy and sandy substrates) were statistically similar. In 0.16 g lÀ 1 turbidity conditions,
greater predation rates were observed in the presence of muddy sediment (91.7%) when
compared to control (20.8%) ( p = 0.0001), sandy substrate conditions (37.5%) ( p = 0.0002)
and sandy-shell (4.2%) ( p = 0.0001), and in control when compared to sandy substrate
conditions ( p = 0.0385). In 0.32 g lÀ 1 conditions, predation was higher in tanks with
muddy sediment (54.2%) than in tanks with no sediment (21.5%) ( p = 0.0036) and with
Fig. 3. Shrimp mortality (mean F S.E.) from thorn fish predation at different turbidity and substrate type
conditions.
42 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
sandy-shell sediment (16.7%) ( p = 0.0121), but not than in sandy substrate conditions
(33.3%) ( p = 0.5281).
Table 4 shows the results of a multiple regression analysis, according to which turbidity
and fine fraction of sediment have a significant effect on P. indicus mortality from thorn
fish predation. Note that in this analysis, the data of predation rates in control tanks were
not considered. Fig. 4 illustrates the response surfaces of mortality rate as a function of
these variables (for easier identification on this relationship, percentage of fine fraction is
arcsine-transformed).
3.5.2. M. monoceros
There was a significant effect of turbidity, substrate type and their interactions on M.
monoceros juveniles’ predation rates (Table 3). In clear water conditions, more shrimps
were eaten in tanks with no sediment (91.6%) than in tanks with any type of sediment
(0.0% in mud, 20.8% in sand and 4.2% in sandy-shell conditions) ( p = 0.0001 for each
comparison) (Fig. 3). In 0.16 g lÀ 1 turbidity conditions, predation rate was lower in
muddy substrate (0.0%) when compared to sandy (50.0%), sandy-shell (37.5%) and
control experiments (41.7%) ( p = 0.0002, p = 0.0082 and p = 0.0023, respectively), but in
0.32 g lÀ 1 conditions, the presence or type of substrate had no significant effect on
predation rates.
In muddy and sandy sediment conditions, predation rates were not affected by turbidity.
In sandy-shell sediment conditions, however, these were higher in 0.16 g lÀ 1 (37.5%)
compared to 0.00 in 0.32 conditions (0.0% in both cases) ( p1 = p2 = 0.0082).
Fig. 4. Predation on shrimps (%) as a function of turbidity and substrate type (expressed as arcsine-transformed
percentage of fine fraction—ffs) based upon the quadratic smoothing procedure fitted to the mortality data.
According to the multiple backward regression, the following equations are the quadratic polynomials that best
describe the response surfaces:
pffiffiffiffiffi
– P. indicus: M ¼ 54:53 À 108:5lt þ 10:20ðsinð ffsÞÞÀ1 (RA = 0.59, p(2,33) = 25.76, p < 0.0000);
2
Fffiffiffiffiffi
pffiffiffiffiffi À1 2
– M. monoceros: M ¼ 24:96 À 2:57ððsinð ffsÞÞ Þ þ 0:12tðsinð ffsÞÞÀ1 (RA = 0.66, F(2,33) = 30.405,
2
p < 0.0000).
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 43
Table 4 shows the results of a multiple regression analysis, according to which there is a
significant effect of substrate type and turbidity –substrate type interaction on shrimp
mortality. Fig. 4 illustrates these results.
3.6. Pneumatophore density
In all experiments, fish were able to swim easily between the pneumatophores but the
pursuing activity in higher pneumatophore density conditions was more difficult.
3.6.1. P. indicus
Shrimp mortality from thorn fish predation was affected by turbidity, pneumatophore
density and their interaction (Tables 3 and 4). More shrimps were captured in the absence of
pneumatophores and in the presence of 30 pneumatophores (91.7% for both cases) than in
tanks with 15 (45.8%) ( p1 = p2 = 0.0003) or 45 pneumatophores (41.7%) ( p1 = p2 = 0.0001)
(Fig. 5).
In low turbidity conditions (0.16 g lÀ 1), fish captured P. indicus juveniles more
effectively in tanks with 30 (58.3%) than in tanks with 15 pneumatophores (20.8%)
( p = 0.0045), and in all the remaining comparisons no differences were detected. In 0.32 g
lÀ 1 conditions, greater predation was observed in intermediate and high pneumatophore
density conditions (50.0% and 45.8% mortality, respectively), compared to control (12.5%)
( p = 0.0045 and p = 0.0170, respectively). More shrimps were also eaten in conditions of 30
than in conditions of 15 pneumatophores per tank (16.7%) ( p = 0.0170).
In the experiments with 15 pneumatophores per tank, the mortality rate decreased with
increasing turbidity, though no differences were detected at the 5% level ( p = 0.0580 for the
comparison between predation in 0.32 and 0.00 g lÀ 1). In 30 pneumatophores per tank
conditions, predation rates were higher in clear water (91.7%) compared to turbid waters
(58.3% in 0.16 g lÀ 1 and 50.0% in 0.32 g lÀ 1) ( p = 0.0170 and p = 0.0012, respectively).
Turbidity had no detectable effect on predation rates in 45 pneumatophores per tank
conditions. Fig. 6 illustrates the mortality response surfaces to turbidity and pneumatophore
density.
Fig. 5. Shrimp mortality (mean F S.E.) from thorn fish predation in different turbidity and pneumatophore density
conditions.
44 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Fig. 6. Predation rates as a function of turbidity and pneumatophore density based upon the quadratic smoothing
procedure fitted to the mortality data.
3.6.2. M. monoceros
Shrimp mortality from fish predation varied according to turbidity, pneumatophore
density and their interaction (Tables 3 and 4). Fig. 6 illustrates the response surfaces of
mortality to variations on turbidity and pneumatophore density.
In clear water conditions, the vulnerability of M. monoceros shrimps to predation
decreased with increasing habitat complexity (Fig. 5). However, the only significant
difference found by the ANOVA was between predation rates in tanks with 45 and tanks
without pneumatophores ( p = 0.0010). A significant negative relation of regression between
predation rates and pneumatophore density was detected ( F(1,14) = 12.06; p = 0.0037), for
which the regression coefficient (r) is 0.68 (slope = À 0.06, intercept = 5.18).
In 0.16 g lÀ 1 conditions, the predation rates increased with increasing pneumatophore
density from 0 to 15 and to 30 per tank (25.0 – 58.3– 87.5%), though the only significant
difference found was between tanks without and tanks with 30 pneumatophores
( p = 0.0004). However, as the density of pneumatophores increased from 30 to 45 per
tank, the predation induced mortality decreased significantly (87.5 –8.3%; p = 0.0001).
Predation rates in 45 pneumatophores per tank experiments were also lower compared to
those in tanks with 15 pneumatophores ( p = 0.0071), and were not statistically different
from those in experiments with no pneumatophores ( p = 0.9530). In 0.32 g lÀ 1 conditions,
the predation rates decreased as structure density increased up to 15 per tank (50 –16.7%),
though the difference was not significant at the 5% level ( p = 0.2179). As pneumatophore
density increased from 15 to 30 per tank, predation rates also increased (16.7 –70.8%)
( p = 0.0027), and then decreased again as pneumatophore density increased up to 45 per
tank (70.8 –25.0%) ( p = 0.0183).
In the presence of 15 pneumatophores per tank, the predation rates remained constant as
the turbidity increased from 0.00 to 0.16 g lÀ 1 (54.2 – 58.3%) and decreased as it increased
from 0.16 to 0.32 g lÀ 1 (16.7%) ( p = 0.0449). In 30 and 45 pneumatophores per tank
conditions, the turbidity had no effect on predation rates.
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 45
4. Discussion
4.1. Turbidity
Turbidity provides predatory protection to P. indicus and M. monoceros juveniles as it
decreases predation rates by thorn fish, but its effect on predator– prey relationships
depends upon the behaviour of the prey. In fact, the visual reactive distance of the predator
is reduced in turbid water (as reported by Moore and Moore, 1976) and thus prey detection
capacity decreases confirming the strict reliance on visual capacities of this predator in prey
detection.
The reduced predator efficiency on capturing P. indicus and M. monoceros in turbid
waters agrees with Minello et al.’s (1987) findings for the pinfish predation on Penaeus
aztecus. However, the prey response to predation by thorn fish showed different pattern
between the two prey species. P. indicus exhibit an approximately linear negative response
to predation as far as turbidity increases while predation rate on M. monoceros decreases
with increasing turbidity up to 0.16 g lÀ 1. An increase in predation rate as water turbidity
reached 0.32 g lÀ 1 was observed, which can be related to a specific effect of turbidity on M.
monoceros juveniles behaviour. In fact, it seems to be a result of a significant increase in
locomotor activity of shrimps due to the decrease in light intensity. A similar predation
response was found by Minello et al. (1987), where high turbidity increased predation rates
of the southern flounder Paralichthys lethostigma, an ambush predator, on brown shrimp P.
aztecus. As happens with these species, many benthic invertebrates are negatively photo-
tactic and their locomotor activity decreases with increasing turbidity (e.g. Wickham and
Minckler, 1975; Minello et al., 1987; Brewer et al., 1989). Hence, at 0.32 g lÀ 1 turbidity,
the fish were not capable to see the prey perfectly, but because shrimps were more active
than in lower turbidity conditions, the encounter rates and consequently the predation rates
increased. The synchronization of the circadian activity rhythms of many shrimp species is
actually known to be controlled mainly by light intensity (e.g. Hughes, 1968; Moller and
¨
Jones, 1975), but other environmental parameters can also influence it (e.g. Fuss and Ogren,
1966; Wickham and Minckler, 1958; Wickham, 1967; Williams and Naylor, 1967). In this
manner, the adaptiveness of the visual sense of fish as a mechanism of prey detection is
apparently lowered. However, some species of fish that use visual cues are able to detect,
pursuit and capture prey at very low light intensity, and according to Minello and
Zimmerman (1983), reduced light intensity does not always restrict predation by visual
feeders during the night. On the other hand, shrimps do not avoid solely visual predators.
It also seems that there is a turbidity threshold above which the thorn fish actually
changes its foraging tactics due to a decline on its visual acuity becoming in very turbid
water conditions an ambush type predator. This type of behavioural shift by the predator is
relatively common (see Stoner, 1979, 1982; Anderson, 1984) and often lead to a real
threshold response of predation rate to turbidity (Nelson and Bonsdorff, 1990). Accord-
ingly, as the turbidity level increased from 0.32 to 0.64 g lÀ 1, the predation rates on M.
monoceros juveniles are lowered as a consequence of a decrease of the fish’s reactive
distance. By affecting prey and predator interactions, turbidity shows to plays an important
role on regulating distribution and mortality of P. indicus and M. monoceros at Saco da
Inhaca.
46 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
4.2. Prey density
As with most fish (see Minello et al., 1989; Bailey and Houde, 1989), predation rate of
thorn fish on shrimps increases with increasing prey density. This functional response is
known to be one of the two basic components of predation (Holling, 1959a,b), and in the
present study, it seems to vary according to turbidity level for both prey species, indicating
that a clear turbidity– prey density interaction on predation rate is present. These changes in
functional response probably lead to important modifications in the shape of the mortality
curve of prey in natural conditions. Tidal variations in the mangroves and adjacent intertidal
areas may affect water turbidity and cyclically modifies shrimps’ density.
Although it was not possible to clearly detect the exact shape of the prey density –
predation rate function, in clear water conditions, predation seems to increase with
increasing prey density until a plateau corresponding to fish’ maximum feeding capacity,
or satiation, is attained (linear or type I response) (Holling, 1959a,b). To this value
corresponds the number of shrimp juveniles needed to feed one thorn fish for 12 h.
Accordingly, around seven P. indicus or five/six M. monoceros juveniles are needed to feed
one of these fish for that period of time. The higher mortality rate of P. indicus in higher
density conditions can be explained by the higher probability that a fish might encounter the
prey. Predation on M. monoceros juveniles in low and high turbidity conditions also
increased probably as a result of higher encounter rates between prey and predator. In
contrast, in high turbidity conditions, the increase is much smoother due to the more gradual
increase of encounter rates between predator and prey (as the visual capacity of predator is
no longer good as previously referred) and a behavioural shift has already taken place in
which the predator changed from active chasing to ambush.
A minimum shrimp density is required before T. jarbua starts feeding on juvenile shrimps
in higher turbidity conditions. For P. indicus juveniles prey that minimum is probably
between 6 and 12 shrimps per tank (43 and 86/m2), as no shrimp was eaten in 6 shrimps per
tank conditions, and for M. monoceros juveniles that minimum is below 6 shrimps per tank.
In intermediate turbidity conditions, predation on M. monoceros juveniles was not
affected by prey density because the shrimps’ locomotor activity was higher than in lower
turbidity conditions and, hence, encounter rates between prey and predator increased.
However, this effect was reduced as prey density increased, because in higher prey density
conditions, changes in prey activity should only have negligible effect on encounter rates
with predators. Consequently, predation rates were constant along the analysed turbidity
spectrum.
M. monoceros juveniles are probably more visible to fish than P. indicus as they are dark
pigmented, and P. indicus, in addition to being a very transparent species, is a faster
swimmer (author’s observation). Thus, in higher turbidity conditions, predation rates on P.
indicus juveniles were often lower than those of M. monoceros. Previous studies with other
species have actually shown that the increased visibility associated with pigmentation can
be more important than either size or density in prey detection and selection by fish (e.g.
Zaret and Kerfoot, 1975; Kislalioglu and Gibson, 1976; Kneib, 1987). Results also seem to
indicate that the increase in locomotor activity of M. monoceros juveniles with increasing
turbidity (decreasing light intensity) is much steeper than that of P. indicus, as even in
higher turbidity conditions the number of M. monoceros juveniles eaten was elevated.
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 47
In turbid waters, the number of P. indicus eaten is not affected by increasing prey
density, at least from 6 to 18 per tank (43 – 128/m2). As only four levels of turbidity and
three levels of prey density were analysed, it was not possible to assert which functional
response corresponds exactly to each turbidity condition analysed. Hence, additional
experiments with higher levels of replication and the analysis of more levels of prey
density are needed. These findings could also indicate that predation in turbid areas
(mangrove areas) only acts as a fine-tuner of abundance when shrimp densities exceed a
given threshold (as suggested by Strong, 1984), from which it does serve a useful purpose
in holding prey population sizes low, but quickly loses its impact on the prey population
when these grow rapidly.
Effects of crowding or gregariousness of these species in nature are not well known,
though individuals are found in close proximity to each other when swimming freely,
resting on the surface, or when burrowed in the ground (personal observations). In aquaria,
the most prominent anti-predator behaviour observed in P. indicus juveniles was the
formation of both aggregations and schools. Formation of aggregation was also observed
for M. monoceros juveniles, but not schools. School and aggregation formation are
actually known to be very important anti-predator processes (e.g. Cushing and Harden
Jones, 1968; Pitcher, 1973; Taylor, 1976; Sullivan and Atchinson, 1978; Turchin and
Kareiva, 1989), as by grouping together, prey can reduce the risk of being eaten. In clear
water conditions, the fish were able to pursue and capture shrimps very intensively and,
hence, the number of shrimps eaten increased with increasing prey density, though the
school effect was still present. In low turbidity conditions, on the other hand, the fish
detection capacity was lower than in clear water and, consequently, the protection effect of
schools was increased. In this manner, the number of shrimps eaten was almost constant
along the prey density spectrum analysed. These results seem to indicate that the
functional response is dependent on prey density and that it therefore contributes to
population regulation (see Oaten and Murdoch, 1975). In intermediate and high turbidity
conditions, predation rates varied proportionally to prey density, suggesting that the
schooling behaviour was not present. The same result was found for M. monoceros
juveniles in low and high turbidity, confirming that this species does not school.
Demographic factors are thus important variables determining predator– prey interactions.
The constancy of P. indicus mortality rate with increasing turbidity from 0.16 to 0.64 g
lÀ 1 in experiments with intermediate and higher prey density suggests that as the turbidity
increases there is a compensation of the decrease in predator’s visual reactive distance with
the increase in shrimps’ locomotor activity. Consequently, the probability that a fish might
encounter the shrimps remains constant, and turbidity has no apparent effect on predation.
With 6 shrimps per tank, for instance, the mortality decreased with increasing turbidity
along the entire turbidity spectrum analysed because prey density was low and even
though the activity level of shrimps increased with increasing turbidity, encounter rates
could not increase significantly in order to maintain the plateau level.
4.3. Substrate type
Substrate type influences predation on shrimp juveniles, and hence these should prefer
substrates that offer maximum protection. Actually, several authors have identified
48 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
correlations between sediment characteristics and shrimp abundance for several species
(e.g. Williams, 1958; Moller and Jones, 1975; Branford, 1981a,b).
¨
Because P. indicus juveniles do not bury in the laboratory (Hughes, 1966; personal
observations), only sandy-shell substrate provided a deterrent effect on predation in clear
water conditions. This was probably a result of the cryptic coloration, which combined
with immobilization permitted blending in of the shrimps with their background, avoiding
detection by visual predators (see Clements and Livingston, 1984; Russo, 1987). In fact,
during preliminary trials, the animals were observed to remain immobile during the
majority of the time, except when the predator approximated. Several studies have shown
that movement of prey appears to elicit predator strikes and to evoke feeding behaviour in
a variety of visual predators (e.g. Zaret and Kerfoot, 1975; Stein and Magnuson, 1976;
Main, 1987). These results are not in accordance with P. indicus juvenile’s distribution in
nature, as according to Hughes (1966), Branford (1981a) and De Freitas (1986), they seem
to prefer very muddy areas within mangrove swamps, both in the primary channels or the
smaller creeks of the upper reaches. According to Ronnback et al. (2002) (author’s
¨ ¨
unpublished data) at Inhaca Island, P. indicus in the mangrove forest does not show any
preference to muddy substrate.
However, it is important to note that this species buries in natural conditions, though
not completely (Hughes, 1966; author’s observation). The presence of a light yellow sandy
substrate had no effect on P. indicus vulnerability to predation, probably because this
species does not bury and the substrate colour did not allow shrimps to blend with the
background. The presence of sandy-shell substrate, however, decreased shrimps’ vulner-
ability to thorn fish predation, probably as a result of a significant camouflage provided to
the semi-transparent body, which made prey detection by fish more difficult. In this
manner, the presence of this type of substrate may only affect predation by modifying the
shrimps’ activity levels. However, higher turbidity conditions already provided predation
protection. In Maputo Bay, P. indicus is mostly captured in areas of turbid waters
(commercial fisheries fleet) close to Maputo River estuary.
Whether shrimps bury, rest, perch or swim was found previously to depend on species,
size, presence or absence of vegetation or other structures and time of the day, among other
parameters (e.g. Moller and Jones, 1975; Minello et al., 1987). M. monoceros juveniles
¨
avoid detection by visual fish predators by burrowing during daylight (personal observa-
tions) and thus decreasing their apparent availability to visual feeding predators (see Minello
and Zimmerman, 1984). This anti-predation behaviour has previously been observed in
several penaeid species (see Ruello, 1973; Moller and Jones, 1975; Minello and Zimmer-
¨
man, 1983; Minello et al., 1987; Dall et al., 1990). Moller and Jones (1975), working with
¨
Penaeus semisulcatus, suggested that burrowing is triggered by increasing light intensity,
while emergence from the substratum at dusk is mainly governed by endogenous control.
The thorn fish were not able to detect or remove buried shrimps from the substratum
and only attacked shrimps lying on the bottom or swimming in the water column,
confirming again the Whitfield and Blaber’s (1978) assessment that this species is a
strictly visual feeder. However, some fish species that also rely on chemosensory abilities
can locate and capture buried shrimps by moving the sand with their snout and pelvic fins.
M. monoceros juveniles can burrow completely both in muddy and sandy substrate, but
in sandy-shell substrate, the burying is not complete as part of the dorsal area, eyes and
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 49
rostrum remain visible above the surface. The ability of brown shrimp to bury is thus
affected by the substrate characteristics just as happens with several other burrowing
penaeids (e.g. Penaeus setiferus, P. aztecus and Penaeus dourarum: Williams, 1958;
Penaeus japonicus: Egusa and Yamamoto, 1961). In clear water, the presence of any
substrate type provided deterrent effect on predation. Accordingly, the selection by
juvenile M. monoceros of a substratum suitable for burrowing reduces significantly the
predation success of fish that rely especially in visual cues for prey detection, and a soft
substratum especially protects this species in clear water as predation rates in all
substratum types were significantly different from those found in control experiments.
These findings agree with the distribution of juveniles and sub-adults in Maputo Bay and
Saco da Inhaca, where M. monoceros is more widespread than other species and can be
found in a diverse number of habitats, from areas with submerged macrophytes to the
deeper reaches of the mangrove swamps (Hughes, 1966; De Freitas, 1986; author’s
unpublished data).
In low turbidity conditions, the presence of substratum did not especially protect M.
monoceros juveniles, confirming that in the absence of sediment the shrimps are not very
active due to the high light intensity and that the fish’ visual detection capacity is lowered
due to the turbidity effect. As turbidity increased further, the M. monoceros’ vulnerability
to predation was minimum in the presence of muddy sediment, probably as a result of the
shrimps’ cryptic coloration that permits blending in with the dark background and avoid
detection. Hence, in muddy substrate conditions, predation rates were not affected by
increasing turbidity. In high turbidity conditions, the locomotor activity of shrimps was
already high due to the low light intensity and, consequently, predation rates were higher
in tanks with no sediment and with sandy-shell sediment when compared to low turbidity
conditions. Actually, Macnae and Kalk (1962) and Joshi et al. (1979) have demonstrated
that this species favours a muddy substratum, though that result was not confirmed by De
Freitas (1986) work, in which it is shown that this species can be found in a wide variety of
substrates.
In the presence of muddy and sandy sediment, turbidity had no effect on thorn fish’s
predation success, but in sandy-shell sediment, where the M. monoceros’ burrowing is
not complete, the effect of turbid water on predation rate was apparently increased by a
reduction in shrimps’ burrowing. In fact, burrowing by M. monoceros juveniles was
reduced in turbid water, as happens with P. aztecus (Minello et al., 1987) and other
penaeids. In this way, substrate type, as light intensity, also regulates the locomotor
activity rhythms of shrimps. Actually, several studies have shown that the activity
rhythms of many penaeid species are modified in the absence of substrate (e.g.
Metapenaeus bennettae, Metapenaeus macleayi and Penaeus plebejus: Racek, 1959;
P. dourarum: Fuss and Ogren, 1966; Crangon crangon: Hagerman, 1970; P. semi-
sulcatus and P. monodon: Moller and Jones, 1975). These activity rhythms that depend
¨
upon the presence and type of substrate and upon the negative phototaxy are probably
an efficient protection against visual predation, and the interaction between turbidity and
substrate type on predation rates can thus influence distribution and abundance patterns
of prey.
So, fish predation on the burying shrimps can be affected by turbidity, substrate type, as
well as by prey and predator species and their interactions.
50 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
4.4. Pneumatophore density
The use of mangrove structures (pneumatophores) by juvenile penaeid shrimps as refuge
from predation was documented for the first time by Primavera (1997). Results of the
present study indicate that, as assessed in her work, the presence of pneumatophores does
provide shrimp a significant amount of protection from fish predators. However, the quality
of refuge provided by pneumatophores in a certain microhabitat seems to depend upon its
complexity, the prey species and its density.
In clear water conditions, increased structural complexity reduced the consumption of
M. monoceros juveniles linearly, a result that is well documented by other experimental
predator –prey studies (e.g. Crowder and Cooper, 1982; Coull and Wells, 1983; Anderson,
1984; Nelson and Bonsdorff, 1990). However, this result still lacks support from the field
studies since at Saco da Inhaca M. monoceros appears in higher densities at the intertidal
flats than within mangrove areas (Ronnback et al., 2002). A thorn fish can see much
¨ ¨
further in low structure conditions, and in high structural complexity, it probably goes
through lengthy periods of search during which no shrimp prey is visible. Hence, the
structures of the environment have an important role in reducing visual contact with prey
and prey vulnerability is lower in these conditions simply because random visual
encounters between predator and prey are reduced. On the other hand, as pneumatophore
density increases, predator activity declines due to a decrease in behaviour associated with
visual contact with prey. In fact, at low structural complexities, the thorn fish were
observed to be active searchers whereas at high densities they became ambush predators,
because in these conditions the pursuit of prey by the predator seems to be inhibited. Thus,
the thorn fish also modifies its foraging tactics with changes in structural complexity,
besides turbidity. These behavioural shifts were observed both in clear and turbid water
and are probably a result of differential energy costs of the different foraging tactics in
differentially structured habitats.
According to Anderson (1984), largemouth bass Micropterus salmoides in an environ-
ment with moderate density of vegetation had higher prey encounter rates than largemouth
bass in a highly structured environment, and that the optimal behaviour in high structure
was more complex than the best foraging mode in low structure. Some authors have also
shown that the diet breadths of fish predators increase with increasing structural complex-
ity (Vince et al., 1976; Anderson, 1984). In nature, structure should then mitigate the
effect of predation on fish’s preferred prey species, possibly resulting in changes in the
overall composition of the prey community (e.g. Crowder and Cooper, 1982; Anderson,
1984).
In low turbidity conditions, increasing pneumatophore density up to 30 per tank seems
to provide predators with increasing levels of cover, enabling them to catch more M.
monoceros juveniles, as the locomotor activity of shrimps is already high due to the
turbidity conditions and so they probably ‘‘feel safe’’ to search for food and/or a suitable
substrate to burrow. In higher turbidity conditions, predation rates initially decreased with
increasing pneumatophore density up to 15 per tank because in these conditions the prey
detection and pursuing activity was more difficult. As the structural complexity increased
up to 30 pneumatophores per tank, the predation rates increased significantly, suggesting
that in those conditions the shrimps’ locomotor activity increased significantly and,
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 51
consequently, predation rates also increased. Besides, the foraging behaviour of fish has
also changed to ambush, allowing them to capture more prey. In turbid waters, as
happened in clear water, there was a break in capture rates at higher pneumatophore
density as a result of the lowered reactive distance of thorn fish and the consequent
behavioural shift.
In low structural complexity conditions, the predation rates were not affected by
increasing turbidity up to 0.16 g lÀ 1, suggesting that the shrimps’ locomotor activity
increases with increasing turbidity, as it is known that the reactive distance of fish
decreased. In intermediate and higher structural complexity the turbidity level had no
significant effect on predation rates, suggesting that the locomotor activity of shrimps is
highly increased with increasing structural complexity, and consequently as it increases,
the locomotor activity increase caused by the increasing turbidity becomes negligible.
Unlike M. monoceros, P. indicus vulnerability to predation in clear water conditions
was not a linear function of increasing habitat complexity. Actually, the results indicate
that small juvenile white shrimps would suffer higher rates of predation in conditions of
null and 30 pneumatophores density than in conditions of low and high densities, which is
apparently related to changes in prey and predator behaviour. Low structural complexity
provides shrimps with cover, allowing the predation rates to decrease, but intermediate
structure density probably provides predators with cover enabling them to catch more prey.
According to Anderson (1984), fish learn as juveniles to forage on a certain group of prey
species, or to apply different strategies in different structural complexity conditions, and
thus the observed predation response to habitat complexity can be explained by
modifications of thorn fish predator behaviour. As the survival in intermediate density
was not significantly different from that on control tanks, shrimps selecting areas with
intermediate pneumatophore density may not survive at greater rates than those found in
areas without any type of structures. Thus, it may be expected that predators would be
attracted to those intermediate structured areas, which provide them with cover, as long as
the structure density is insufficient to reduce foraging efficiency to less than it would be in
other habitats.
In high turbidity water (0.32 g lÀ 1), the cover provided to fish in intermediate and higher
levels of structural complexity seems to be highly significant. Thus, because P. indicus
juveniles are not able to detect the predator that furthermore is immobile due to the
behavioural shift, their locomotor activity increases, resulting in an increase in predation
rates. However, it is not known if P. indicus schooling behaviour changes with structural
complexity variation.
Since P. indicus juveniles do not burrow completely, in natural conditions the need for
adequate cover is great, especially since many mangrove as well as seagrass predators
consume primarily epifaunal species (Nelson, 1979; Stoner, 1979; Coen et al., 1981). In
fact, according to Ronnback et al. (2002), this species exhibits a stronger preference for
¨ ¨
mangrove forest than M. monoceros. The pneumatophores provide three dimensions
within which P. indicus juveniles may hide and space themselves. These shrimps were
observed to attach their bodies parallel to the pneumatophores when they perceived the
stalking behaviour of thorn fish and to remain immobile, avoiding detection. This strategy
would protect P. indicus juveniles from many ‘‘active chase’’ predators that require a
visual cue for prey detection. However, prey species that are protected from one search and
52 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
attack strategy (e.g. active chase or pursuit) may be extremely vulnerable to other mode of
prey capture (e.g. stalking or ambush) (e.g. Neill and Cullen, 1974; Primavera, 1997) and,
consequently, that anti-predator behaviour is not successful against all fish predation
strategies.
In low and intermediate structural complexities, the P. indicus mortality rates’ decrease
with increasing turbidity is much smoother than in no pneumatophores conditions, as a
result of an increase in shrimps’ locomotor activity due to increasing turbidity combined
with the cover provided to predator. These two factors seem to have an additive effect in
increasing prey activity and in decreasing predators’ prey visual detection and capture
capacity, so that in higher structure density the turbidity effect was not detected. Thus, the
effective provision of shelter of different habitats is highly variable and depends not only
upon structure, density, but also on the behaviour of predator and prey as well (e.g. Minello
and Zimmerman, 1983; Main, 1987; Primavera, 1997) and the way these factors interact.
5. Conclusions
Turbidity provides protection from thorn fish T. jarbua predation for both species of
shrimps studied.
The borrowing behaviour of shrimps influences the predation rate of Thorn fish.
Accordingly, the presence of a suitable substrate for burying decreases vulnerability to
predation of M. monoceros, a borrowing shrimp, but not of the non-burying P. indicus.
The presence of pneumatophores does provide shrimp a significant amount of
protection from fish predation. Thus the density and distribution of juvenile shrimps in
the field can be partly explained on basis of predation pressure (Crowder and Cooper,
1982).
Our results confirms previous findings that comparing areas with regard to their
protective capacity for juvenile shrimp are complicated as stated by Minello et al. (1987),
due to the highly significant interactions among habitats structures, predators and prey
species (different levels of each factor alter others). This study provides some information
for the assessment of the ecological value of mangroves and adjacent intertidal substrate as
well as estuaries as suitable shelter for juvenile penaeid shrimps (Minello et al., 1987;
Primavera, 1997).
Validation of these conclusions is the subject of conjectures since the laboratory con-
ditions are artificial and therefore making difficult extrapolations to natural environment.
Acknowledgements
´
We thank the Head of Marine Biological Station at Inhaca, Dr. Tomas Muacanhia, for
logistic support. In particular, Prof. Nils Kautsky, Dr. Steve Telford and two anonymous
reviewers are acknowledged for their helpful comments and English corrections on the
early version of this work. This work is funded by SIDA/SAREC and is part of the long-
term research programme ‘‘The development of biological research capacity at the
Department of Biological Sciences-Eduardo Mondlane University.’’ [AU]
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 53
References
Anderson, O., 1984. Optimal foraging by largemouth bass in structured environments. Ecology 65, 851 – 861.
Bailey, K.M., Houde, E.D., 1989. Predation on eggs and larvae of marine fishes and the recruitment problem.
Adv. Mar. Biol. 25, 1 – 83.
Branford, J.R., 1981a. Sediment preferences and morphometric equations for Penaeus monodon and Penaeus
indicus from creeks of the Red Sea. Estuar. Coast. Shelf Sci. 13, 473 – 476.
Branford, J.R., 1981b. Sediment and the distribution of penaeid shrimp in the sudanese Red Sea. Estuar. Coast.
Shelf Sci. 13, 349 – 354.
Brewer, D.T., Blaber, S.J.M., Salini, J.P., 1989. Feeding biology of Caranx bucculentus Alleyne and Macleay
(Teleostei: Carangidae) in Albatross Bay, Gulf of Carpentaria, with special reference to predation on penaeid
prawns. Aust. J. Mar. Freshw. Res. 40, 657 – 668.
Chong, V.C., 1995. The prawn – mangrove connection—fact or fallacy? Seminar on Sustainable Utilization of
Coastal Ecosystems for Agriculture, Forestry and Fisheries in Developing Regions, held in Kuala Lumpur,
Malaysia. Japan International Research Center for Agricultural Science (JIRCAS).
Chong, V.C., Sasekumar, A., Leh, M.U.C., D’Cruz, R., 1990. The fish and prawn communities of a Malaysian
coastal mangrove system, with comparisons to adjacent mud flats and inshore waters. Estuar. Coast. Shelf Sci.
31, 703 – 722.
Clements, W.H., Livingston, R.J., 1984. Prey selectivity of fringed filefish Monacanthus ciliatus (Piscis: Mon-
acantidae): role of prey accessibility. Mar. Ecol. Prog. Ser. 16, 291 – 295.
Coen, L.D., Heck Jr., K.L., Abele, L.G., 1981. Experiments on competition and predation among shrimps of
seagrass meadows. Ecology 62 (6), 1484 – 1493.
Coles, R.G., Lee Long, W.J., Squire, B.A., Squire, L.C., Bibby, J.M., 1987. Distribution of seagrasses and
Associated Juvenile Commercial Penaeid Shrimps in North-eastern Quensland Waters. Aust. J. Mar. Freshw.
Res. 38, 103 – 119.
Coull, B.C., Wells, J.B.J., 1983. Refuges from predation: experiments with phytal meiofauna from the New
Zealand rocky intertidal. Ecology 64, 1599 – 1609.
Crowder, L.B., Cooper, W.E., 1982. Habitat structural complexity and the interaction between bluegills and their
prey. Ecology 63, 1802 – 1813.
Cushing, D.H., Harden Jones, F.R., 1968. Why do fish school? Nature 218, 918 – 920.
Dall, W., Hill, D.J., Rothlisberg, P.C., Staples, D.J., 1990. The biology of the penaeidae. Adv. Mar. Biol. 27, 1 – 489
(Academic Press, London).
De Freitas, A.J., 1986. Selection of nursery areas by six southeast African Peaeidae. Estuar. Coast. Shelf Sci. 23,
901 – 908.
Egusa, S., Yamamoto, T., 1961. Studies on the respiration of the ‘kuruma’ prawn, Penaeus japonicus Bate.
Burrowing behaviour with special reference to its relation to environmental oxygen concentration. Bull. Jpn.
Soc. Sci. Fish. 27, 22 – 26.
Fuss, C.M., Ogren, L.H., 1966. Factors affecting activity and burrowing habits of the pink shrimp, Penaeus
dourarum. Biol. Bull. Mar. Biol. Lab., Woods Hole 130, 170 – 191.
Hagerman, L., 1970. Locomotory activity patterns of Crangon vulgaris (Crustacea, Natantia). Ophelia 8,
255 – 266.
Holling, C.S., 1959a. The components of predation as revealed by a study of small mammal predation of the
European pine sawfly. Can. Entomol. 91, 293 – 320.
Holling, C.S., 1959b. Some characteristics of simple types of predation and parasitism. Can. Entomol. 91,
385 – 398.
Hughes, D.A., 1966. Investigations of the nursery areas and habitat preferences of juvenile shrimps in Mozam-
bique. J. Appl. Ecol. 3, 349 – 354.
Hughes, D.A., 1968. Factors controlling emergence of pink shrimp, Penaeus dourarum, from the substrate. Biol.
Bull. Mar. Biol. Lab., Woods Hole 134, 48 – 59.
´ ´ ´
INE, 1999. Anuario Estatıstico. Instituto Nacional de Estatıstica, Mocambique. 134 pp.
¸
Joshi, P.K., Kulkarni, G.K., Nagabhushanam, R., 1979. Studies on the behaviour and substratum prefer-
ence in juveniles of the marine prawn, Metapenaeus monoceros (Fabricius). Hydrobiologia 65,
195 – 198.
54 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
Kislalioglu, M., Gibson, R.N., 1976. Some factors governing prey selection by the 15-spined stickleback,
Spinachia spinachia (L.). J. Exp. Mar. Biol. Ecol. 25, 159 – 169.
Kneib, R.T., 1987. Predation risk and use of intertidal habitats by young fishes and shrimp. Ecology 68, 379 – 386.
Krumbein, W.C., Sloss, L.L., 1963. Stratigraphy and Sedimentation. Freeman, San Francisco, CA. 497 pp.
Laprise, R., Blaber, S.J.M., 1992. Predation by Moses perch, Lutjanus russelli, and blue-spotted trevally, Caranx
bucculentus, on juenile brown tiger prawn, Penaeus esculentus: effects of habitat structure and time of day.
J. Fish Biol. 40, 627 – 635.
Loneragan, N.R., Kenyon, R.A., Staples, D.J., Poiner, I.R., Conacher, C.A., 1998. The influence of seagrass type
on the distribution and abundance of postlarval and juvenile tiger prawns (Penaeus esculentus and P. semi-
sulcatus) in the western Gulf of Carpentaria, Australia. J. Exp. Mar. Biol. Ecol. 228 (2), 175 – 195.
´ ˜
Macia Jr., A.A., 1990. Estudo de duas especies de camarao (Penaeus indicus e Metapenaeus monoceros)
´
capturadas nas pescarias artesanal e semi-industrial da Baıa de Maputo. Tese de Licenciatura. Universidade
Eduardo Mondlane, Maputo. 53 pp.
Macnae, W., 1974. Mangrove Forests and Fisheries. FAO, Rome. IOFC/DEV/74/34, 35 pp.
Macnae, W., Kalk, M., 1962. The ecology of the mangrove swamps at Inhaca Island, Mocambique. J. Ecol. 50,
¸
19 – 34.
Main, K.L., 1987. Predator avoidance in seagrass meadows: prey behaviour, microhabitat selection and cryptic
coloration. Ecology 68, 170 – 180.
Minello, T.J., Zimmerman, R.J., 1983. Fish predation on juvenile brown shrimp, Penaeus aztecus Ives: the effect
of simulated Spartina structure on predation rates. J. Exp. Mar. Biol. Ecol. 72, 211 – 231.
Minello, T.J., Zimmerman, R.J., 1984. Selection for brown shrimp, Penaeus aztecus, as prey by the spotted
seatrout, Cynoscion nebulosus. Contrib. Mar. Sci. 27, 159 – 167.
Minello, T.J., Zimmerman, R.J., 1991. The role of estuarine habitats in regulating growth and survival of juvenile
penaeid shrimps. In: De Loach, P.F., Dougherty, W.J., Davidson, M.A. (Eds.), Frontiers of Shrimp Research.
Elsevier, Amsterdam, pp. 1 – 16.
Minello, T.J., Zimmerman, R.J., Martinez, E.X., 1987. Fish predation on juvenile brown shrimp Penaeus
aztecus Ives: effects of turbidity and substratum on predation rates. US Natl. Mar. Fish Serv. Fish.
Bull. 85, 59 – 70.
Minello, T.J., Zimmerman, R.J., Martinez, E.X., 1989. Mortality of young brown shrimp Penaeus aztecus in
estuarine nurseries. Trans. Am. Fish. Soc. 118, 693 – 708.
Moore, J.W., Moore, I.A., 1976. The basis of food selection in flounders, Platichthys flesus (L.), in the Seven
Estuary. J. Fish Biol. 9, 139 – 156.
Moller, T.H., Jones, D.A., 1975. Locomotory rhythms and burrowing habits of Penaeus semisulcatus (de Hann)
¨
and P. monodon (Fabricius). J. Exp. Mar. Biol. Ecol. 18, 61 – 77.
Neill, S.R., Cullen, J.M., 1974. Experiments on whether schooling by their prey affects the hunting behaviour of
cephalopods and fish predators. J. Zool., Lond. 172, 549 – 569.
Nelson, W.G., 1979. Experimental studies of selective predation on amphipods: consequences for amphipod
distribution and abundance. J. Exp. Mar. Biol. Ecol. 3, 225 – 245.
Nelson, W.G., Bonsdorff, E., 1990. Fish predation and habitat complexity: are complexity thresholds real? J. Exp.
Mar. Biol. Ecol. 141, 183 – 194.
Oaten, A., Murdoch, W.W., 1975. Functional response and stability in predator – prey systems. Am. Nat. 109,
19 – 30.
Palha de Sousa, L., 1996. Estado actual de conhecimento dos recursos pesqueiros de Mocambique, prespectivas
¸
para o futuro. In: Dias, D., Scarlet, P., Hatton, J., Macia, A. (Eds.), Proceedings do Workshop sobre: O papel
ˆ
¸˜ ˜
da Investigacao na gestao da zona costeira, Maputo, 24 e 25 de Abril 1996. Departamento de Ciencias
´
Biologicas, Universidade Eduardo Mondlane, Mozambique, pp. 50 – 58. Funded by: IDRC/NORAD.
Pitcher, T.J., 1973. The three dimensional structure of schools in the minnow, Phoxinus phoxinus. Anim. Behav.
21, 673 – 686.
Primavera, J.H., 1997. Fish predation on mangrove-associated penaeids—the role of structures and substrate.
J. Exp. Mar. Biol. Ecol. 215 (2), 205 – 216.
Primavera, J.H., Lebata, J., 1995. Diel activity patterns in Metapenaeus and Penaeus juveniles. Hydrobiologia
295, 295 – 302.
Racek, A.A., 1959. Prawn investigations in Eastern Australia. Res. Bull. St. Fish. N. S. W., No. 6, 57 pp.
A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56 55
Robertson, A.I., Blaber, S.J.M., 1992. Plankton, epibenthos and fish communities. In: Robertson, A.I., Alongi,
D.M. (Eds.), Tropical Mangrove Ecosystems. Coastal and Estuarine Studies, vol. 41. American Geophysical
Union, Washington, DC, pp. 63 – 100.
Robertson, A.I., Duke, N.C., 1987. Mangroves as nursery sites: comparison of abundance and species compo-
sition of fish and crustaceans in mangroves and other near shore habitats in tropical Australia. Mar. Biol. 96,
193 – 205.
Ronnback, P., Troell, M., Kautsky, N., Primavera, J.H., 1999. Distribution pattern of shrimps and fish among
¨ ¨
Avicennia and Rizophora microhabitats in the Pagbilao mangroves, Philippines. Estuar. Coast. Shelf Sci. 38,
223 – 234.
Ronnback, P., Macia, A., Amqvist, G., Schultz, L., Troell, M., 2002. Do penaeid shrimps have a preference for
¨ ¨
mangrove habitats? Distribution pattern analysis on Inhaca Island, Southern Mozambique. Estuar. Coast.
Shelf Sci. 55, 427 – 436.
Ruello, N.V., 1973. Burrowing, feeding and spatial distribution of the school prawn, Metapenaeus maclearyi
(Haswell) in the Hunger River region, Australia. J. Exp. Mar. Biol. Ecol. 13, 189 – 206.
Russo, A.R., 1987. Role of habitat complexity in mediating predation by the gray damselfish Abudefduf sordidus
on epiphytal amphipods. Mar. Ecol. Prog. Ser. 36, 101 – 105.
Staples, D.J., Vance, D.J., Heales, D.S., 1985. Habitat requirements of juvenile penaeid shrimps and their
relationship to offshore fisheries. In: Rothlisberg, P.C., Hill, B.J., Staples, D.J. (Eds.), Second Australian
National Shrimp Seminar National Prawn Seminar. CSIRO, Cleveland, pp. 47 – 54.
Stein, R.A., 1977. Selective predation, optimal foraging, and the predator – prey interaction between fish and
crayfish. Ecology 58, 1237 – 1253.
Stein, R.A., Magnuson, J.J., 1976. Behavioral response of crayfish to a fish predator. Ecology 57,
751 – 761.
Stoner, A.W., 1979. Species specific predation on amphipod crustacea by the pinfish Lagodon rhomboides:
mediation by macrophyte standing crop. Mar. Biol. 55, 201 – 207.
Stoner, A.W., 1982. The influence of benthic macrophytes on the foraging behaviour on pinfish, Lagodon
rhomboides (Linnaeus). J. Exp. Mar. Biol. Ecol. 58, 271 – 284.
Strong, D.R., 1984. Density-vague ecology and liberal population regulation in insects. In: Price, P.W., Slobod-
chikoff Pand, C.N., Gaud, W.S. (Eds.), New Ecology. Wiley, New York, pp. 313 – 327.
Sullivan, J.F., Atchinson, G.J., 1978. Predator – prey behaviour of fathead minnows, Pimphales promelas and
largemouth bass, Micropterus salmoides in a model ecosystem. J. Fish Biol. 13, 249 – 253.
Taylor, R.J., 1976. Value of clumping to prey and the evolutionary response to ambush predators. Am. Nat. 110,
13 – 29.
Turchin, P., Kareiva, P., 1989. Aggregation in Aphis varians: an effective strategy for reducing predation risk.
Ecology 70, 1008 – 1016.
Vance, D.J., 1992. Activity patterns of juvenile penaeid prawns in response to artificial tidal and day – night
cycles: a comparison of three species. Mar. Ecol. Prog. Ser. 87, 215 – 226.
Vance, D.J., Haywood, M.D.E., Staples, D.J., 1990. Use of mangrove estuary as a nursery area by postlarval and
juvenile banana prawns, Penaeus merguiensis de Man, in northern Australia. Estuar. Coast. Shelf Sci. 31,
689 – 701.
Vance, D.J., Haywood, M.D.E., Staples, D.J., 1996. How far do prawns and fish move in mangrove? Distribution
of banana prawn P. merguiensis and fish in tropical mangrove forest in northern Australia. Mar. Ecol. Prog.
Ser. 131, 115 – 124.
Vince, S., Valiela, I., Backus, M., Teal, J.M., 1976. Predation by the salt marsh killing fish Fundulus heteroclitys
(L.) in relation to prey size and habitat structure: consequences for prey distribution and abundance. J. Exp.
Mar. Biol. Ecol. 23, 255 – 266.
˚
Whitfield, A.K., Blaber, J.M., 1978. Scale-eating habits of the marine teleost Terapon jarbua (Forskal). J. Fish
Biol. 12, 61 – 70.
Wickham, D.A., 1967. Observations on the activity patterns in juveniles of the pink shrimp Penaeus dourarum.
Bull. Mar. Sci. 17, 769 – 786.
Wickham, D.A., Minckler III, F.C., 1958. Substrates as a factor in shrimp distribution. Limnol. Oceanogr. 3,
283 – 290.
Wickham, D.A., Minckler III, F.C., 1975. Laboratory observations on daily patterns of burrowing and locomotor
56 A. Macia et al. / J. Exp. Mar. Biol. Ecol. 291 (2003) 29–56
activity of pink shrimp, Penaeus douradum, brown shrimp Penaeus aztecus, and white shrimp, Penaeus
setiferus. Contrib. Mar. Sci. 19, 21 – 35.
Williams, A.B., 1958. Substrates as factors in shrimp distribution. Limnol. Oceanogr. vol. 3, pp. 283 – 290.
Williams, A.B., Naylor, E., 1967. Spontaneously induced rhythm of tidal periodicity in the laboratory reared
Carcinus. J. Exp. Biol. 47, 229 – 234.
Zaret, T.M., Kerfoot, W.C., 1975. Fish predation on Bosmina longirostris: body-size selection versus visibility
selection. Ecology 56, 232 – 237.
Zimmerman, R.J., Zamora, G., 1984. Selection of vegetated habitat by brown shrimp, Penaeus aztecus, in a
Galveston Bay salt marsh. Fish. Bull., U.S. 82, 325 – 336.