Paternal care behaviour of sand gobies is determined
by habitat related nest structure
Maria Järvi-Laturi, Topi K. Lehtonen1) , Christophe Pampoulie2) &
Kai Lindström3)
(Department of Biological and Environmental Sciences, University of Helsinki, P.O. Box
65, FI-00014, Helsinki, Finland)
(Accepted: 3 September 2007)
Summary
This study examines the effects of habitat-dependent nest structure on male parental behaviour in a small marine fish, the sand goby (Pomatoschistus minutus). In the northern Baltic,
the sand goby breeds in two different habitats: on soft sand bottoms and on rocky beaches.
We compared in laboratory conditions the behaviour of nest-guarding males occupying either a typical sand or rock habitat nest. We found that males with a rock habitat nest fanned
their eggs considerably less and with shorter bouts than males in the sand habitat treatment.
Hence, our study shows that nest structure can be an important factor determining parental
care behaviour. The differences in nest structure may result in divergent selection pressures
on male sand gobies occupying the two habitats.
Keywords: egg fanning, habitat choice, nest structure, parental care, sand goby.
Introduction
An individual may encounter different selection pressures in adjacent habitats due to differences in resource competition, pathogen load, predation
1) Corresponding author’s current address: Department of Biology, University of Konstanz,
D-78457 Konstanz, Germany, e-mail: [email protected]
2) Current address: Marine Research Institute, Skúlagata 4, P.O. Box 1390, IS-121 Reykjavik,
Iceland.
3) Current address: Environmental and Marine Biology, Åbo Akademi University, FI-20500
Turku, Finland.
© Koninklijke Brill NV, Leiden, 2008
Behaviour 145, 39-50
Also available online - www.brill.nl/beh
40
Järvi-Laturi, Lehtonen, Pampoulie & Lindström
pressure and many abiotic factors. An optimal balance in relation to these
factors should then be found by choosing a habitat where breeding success
is maximised (Morris, 1987). In several bird species, varying environmental
conditions, and hence nest and territory quality, have been found to affect not
only breeding success but also parental investment (e.g., Kilpi & Lindström,
1997; Alabrudzińska et al., 2003). Hence, these studies suggest that optimal
allocation to current versus future reproduction may vary in relation to the
environment (habitat) an individual occupies. Indeed, species that use various habitats for reproduction provide excellent opportunities for studying
the effects of varying environmental conditions on parental investment and
behaviour.
Empirical research has demonstrated that in fish with paternal care, factors that may affect care behaviour include temperature (Reebs et al., 1984),
the amount of dissolved oxygen in the water (Torricelli et al., 1984; Jones
& Reynolds, 1999; Takegaki & Nakazono, 1999; Lissåker et al., 2003) and
salinity (St Mary et al., 2001). For example, several species adjust their eggfanning rate to the level of dissolved oxygen (Torricelli et al., 1984; Jones &
Reynolds, 1999; Takegaki & Nakazono, 1999), and these changes in parental
effort may affect body condition of the egg-tending parent (Lissåker et al.,
2003). To maintain their body condition at an adequate level, parental males
of many fish species are known to use their own offspring as an additional energy reserve (FitzGerald & Whoriskey, 1992; Manica, 2002). This filial cannibalism might be needed to complete several brood cycles (Rohwer, 1978),
which illustrates the point that the males need to make context-dependent
decisions regarding the balance between future and present reproduction.
Sand gobies, Pomatoschistus minutus, have recently been found to use two
different types of breeding habitat in the northern Baltic (Lehtonen & Lindström, 2004). Soft sand bottoms are the commonly acknowledged breeding
habitats of this small marine fish. On sand, a male excavates a nest under
a suitable object, covers it with sand, and attracts females to spawn in his
nest. Sand habitat nests are scarce and males compete for access to them
(Lindström, 1988). In contrast, on rocky bottoms, nest sites are abundantly
available and no nest construction takes place. Instead, males utilise cavities amidst rocks. The nests in the two habitats are structurally different:
sand bottom nests have only one small opening whereas rocky bottom nests
are usually more exposed to water flow from more than one side. In both
habitats, egg-caring sand goby males clean and fan the eggs in the nest they
Habitat specific paternal care
41
defend. These activities are carried on until the eggs hatch at the age of 1-2
weeks, depending on the water temperature. In the study area, sand gobies
usually have only one breeding season in their lifetime, during which they
can breed several times (Lindström, 2001).
Despite a flourishing literature on potential environmental effects on paternal care (e.g., MacDonald et al., 1995a,b; Jones & Reynolds, 1999; St
Mary et al., 2001; Hale et al., 2003), little is known about how habitat, nest
structure or brooding substrate affects care behaviour. Here we apply laboratory experiments to assess care behaviour of males occupying either a
typical sand or rock habitat nest. If care behaviour is correlated with nest
structure, energy budget of the males may differ depending on the nest they
occupy. Therefore, we also assessed energy expenditure of the males over
one complete brood cycle.
Material and methods
This study was conducted during the sand goby breeding seasons (mid May–
July) of 2000 and 2001 at Tvärminne zoological station, southern Finland.
Sand gobies were caught in rock and sand habitats near the field station using
a hand trawl and hand nets. Fish were separated by sex and habitat, and
kept in large storage tanks. During the short period before the experiments,
the fish were fed ad libitum with live Neomysis integer shrimps and frozen
chironomid larvae. All stock and experimental tanks were supplied with
continuously renewed seawater.
Before the fish were moved to experimental tanks (60 × 35 × 40 cm or
70 × 40 × 40 cm), they were measured to the nearest mm and weighed to the
nearest 0.01 g on an electronic balance. Males were measured and weighed
again after the experiment. Males caught in rock habitat were used in the
rock habitat treatment whereas those caught in sand habitat were only ever
used in the sand treatment. Females used in experiment 2 (see below) were
caught in the same habitat as the males, but in experiment 1 (see below), only
sand habitat females were used because of a low capture rate of females in
the rock habitat.
During the experimental phase, males were fed daily with four frozen
chironomid larvae each. Most of the sand habitat males had built a nest, and
rock habitat males had settled into one, within 24 h of being introduced into
42
Järvi-Laturi, Lehtonen, Pampoulie & Lindström
the experimental tanks. At that point, a ripe female was added. The nests
were checked for spawning twice a day and the female was removed as
soon as spawning had occurred. Females used for the simultaneously run
replicates of the different treatments were chosen from the same size class
to reduce clutch size variation. Fish were released back into their original
environment after the experiments.
Experiment 1, nest size standardised
To investigate habitat specific differences in parental care, we measured fanning behaviour of egg-tending males over a complete brood cycle. We also
estimated body condition as W/TL3 , where W is weight and TL is total
length.
Half of the tanks were designated to the ‘sand habitat’ and had a 5 cm layer
of fine sand on the bottom. The tanks in the ‘rock habitat’ treatment had a
layer of grey tiles and small stones on the bottom. The standardised nests
used in both treatments were ceramic tiles that mimicked flat rocks that are
often used as nest sites also on sand bottoms (Lehtonen & Lindström, 2004).
The tiles were supported slightly above the bottom by four small stones
placed under each corner. This enabled sand goby males to use them as nest
sites irrespective of habitat type. Each nest had a piece of transparent film
that lined its inner surface. Females attached their eggs onto the transparency
during spawning.
Each male was given a gravid female to spawn with. If spawning did not
occur within three days, both the male and female were replaced. The experiment was run from May to July. At the beginning of the season (May),
the duration of a brood cycle was 12-14 days and later on, when water temperature was higher (late June, July), ca. eight days. The water temperature
varied between 8 and 18◦ C, depending on the starting date of the replicate.
Immediately after a spawning was observed, the transparent film with the
eggs was photographed (Minolta RD 175 fitted with a 50 mm macro lens),
and the eggs were then replaced back to the nest. The egg clutch was photographed again at the end of the brood cycle. The digital images were later
used for counting the number of eggs in the nest. Four sand habitat and ten
rock habitat males terminated their brood cycle by cannibalising all eggs during the first days, which implies no significant difference between habitats
(Fisher’s exact test, p = 0.31). These males were excluded from the data
Habitat specific paternal care
43
analyses. The number of successful replicates was 12 in both treatments. Of
these fish, the total length of the males caught in the sand habitat was on
average 53.6 mm (SD = 2.43, N = 12) and of rock habitat males 52.1 mm
(SD = 1.62, N = 12; two sample t-test, t = 1.78, df = 22, p = 0.089).
Data on the clutch size of one male in each treatment is not available. Each
male and female was used only once.
Male sand gobies fan their eggs by moving their pectoral and caudal fins.
This behaviour, and location of the male in the tank, was video-recorded for
a 15-min period on each observation day. The day of spawning is considered
as day 0 and the first recording of parental care activities was done on day
2. The second recording took place in the middle of the brood cycle (days
4-9, depending on the prevailing water temperature) and the last when the
eggs had clearly visible pigmentation and were soon to be hatched (days 612). The timing of the second and third observation was adjusted according
to the temperature and, thus, the expected development time of the eggs.
Recordings were analysed using an event recorder program. The proportion
of time engaged in the following activities was calculated from the total
observation time: (1) the time spent outside the nest and (2) fanning. In
addition, the average fanning bout length was calculated as the total time
spent fanning divided by the number of fanning bouts.
Experiment 2, natural nest sites
The aim of this experiment was to measure habitat specific differences in
paternal care in near to natural conditions. The tanks were decorated with
either real rocks or sand from a breeding area of sand gobies, and only natural
nest sites that had earlier been used by sand gobies in the field were provided.
Despite a low statistical power (see below for sample sizes), we consider the
semi-natural set up an important supplement to the more controlled design
of experiment 1. Experiment 2 was conducted in July, late in the sand goby
breeding season. The temperature of seawater stayed at ca. 18◦ C, resulting
in constant brood cycle duration of eight days.
The sand habitat aquaria had a shell of the mussel Mya arenaria as a nest
site. In the field, Mya shells account for the majority of available nests in
the sand habitat (Lehtonen & Lindström, 2004). The rock habitat aquaria
were provided with a flat rock (the nest site) on top of a layer of small rocks.
When the female had spawned, she was removed from the tank and the area
44
Järvi-Laturi, Lehtonen, Pampoulie & Lindström
covered by her egg mass was drawn onto a transparent sheet, which was
later photographed. The area of the egg mass was used as an estimate of
the size of her clutch. Care behaviour of the male was observed 2, 6 and 8
days after the spawning during 10-min observation sessions. Observers hid
behind plastic blinds to minimise disturbance. After 8 days, when the eggs
were ready to hatch, the remaining egg mass area was measured and the
replicate was terminated. The eggs often hatched soon after exposure to the
air and the hatchlings were then released to the sea.
The experiment was replicated with 8 rock and 9 sand habitat males.
However, 4 males did not survive until the end of the experiment (a common
phenomenon also in the field in the latter part of the breeding season) and
1 consumed all his eggs, leaving 6 males that completed the brood cycle in
both treatments. These males were of similar size (rock habitat: 51 mm, SD
= 4.7, N = 6; sand habitat: 50 mm, SD = 6.8, N = 6; two sample t-test,
t = 0.419, df = 10, p = 0.68). The final egg area is available only for 10 of
these 12 males; two clutches hatched just before being measured.
Statistical analyses
All variables were tested for normality. Fanning bout length had to be
ln(x + 1)-transformed, whereas the proportion of time fanning and time
spent outside the nest were transformed with arcsin square-root transformation. Because we had measured male behaviour on three different occasions
for each male, we used repeated measures ANOVA for analysing treatment
effects.
Results
Experiment 1
Males spent more time fanning in the sand habitat (repeated measures
ANOVA, between subjects habitat effect, F1,22 = 24.56, p < 0.001; Figure 1a). There was no clear-cut increase in total fanning time towards the
end of the care period (repeated measures ANOVA, within subjects time effect, F2,44 = 2.596, p = 0.086; habitat and time interaction: F2,44 = 0.15,
p = 0.86). The average fanning bout was shorter in the rock habitat (repeated measures ANOVA, between subjects habitat effect, F1,22 = 43.88,
Habitat specific paternal care
45
Figure 1. Proportion of time spent fanning in rock (filled bars) and sand habitat treatment
(open bars) (a) in size-matched artificial nests at the beginning, in the middle and at the end of
the brood cycle, and (b) in natural nest sites 2, 6 and 8 days after the brood cycle was started.
Figure 2.
Fanning bout length in rock (filled bars) and sand habitat nests (open bars). (a)
Size-matched artificial nests and (b) natural nest sites.
p < 0.001; Figure 2a). Males mostly fanned only 1-5 s at a time in rock
habitat nests, whereas in sand habitat nests a typical fanning bout was considerably longer. Males spent similar proportions of time outside the nest in
both treatments (repeated measures ANOVA, between subjects habitat effect,
F1,22 = 0.000, p = 0.99).
Males that completed their brood cycle started with 984 ± 389 (mean ±
SD) eggs in the rock habitat (N = 11) and 917 ± 252 eggs in the sand
habitat (N = 11). All males lost or ate some of their eggs, but there was
no difference in the number of eggs lost between habitats (two sample ttest, t = 1.185, df = 20, p = 0.25). The change in male body condition
was not associated with egg loss (potential filial cannibalism; ANCOVA,
F1,16 = 0.624, p = 0.44). There was a tendency for body condition to
show a larger decrease in sand habitat males than in rock habitat males
(F1,16 = 3.64, p = 0.075).
46
Järvi-Laturi, Lehtonen, Pampoulie & Lindström
Experiment 2
In the second experiment, males lost eggs at similar rates, irrespective of
habitat. Just before hatching, males with rock habitat nests still had 57%
(SD = 14, N = 4) of their original egg mass. In the sand habitat nests egg
survival was 62% (SD = 21, N = 6). However, we found differences in
care behaviour between the treatments. The proportion of time spent fanning
tended to be longer in sand habitat nests than in rock habitat nests (repeated
measures ANOVA, between subjects habitat effect: F1,10 = 4.313, p =
0.065; Figure 1b). Fanning time increased towards the end of the care period
(within subjects time effect: F2,20 = 48.15, p < 0.001), but the increase was
faster in the sand habitat (habitat and time interaction: F2,20 = 4.576, p =
0.023; Figure 1b). Males with a rock habitat nest had considerably shorter
fanning bouts than males occupying a sand habitat nest (repeated measures
ANOVA, between subjects habitat effect: F1,10 = 33.18, p < 0.001; Figure
2b) and as the brood cycle progressed, bout length increased (within subjects
time effect: F2,20 = 14.92, p < 0.001) similarly in both treatments (habitat
and time interaction: F2,20 = 2.261, p = 0.13).
Discussion
We found a pronounced difference in parental care behaviour between the
rock and sand habitats. Males spent more time fanning their eggs in sand
habitat nests than in rock habitat nests in the experiment with standardised
nest size. Additionally, males with a sand habitat nest fanned without interruption for longer periods of time, whereas males in rock habitat nests used
very short fanning bouts. Fanning bout length and frequency have been suggested to have even larger impact on hatching success than the total time
spent fanning (Östlund & Ahnesjö, 1998). Hence, our results suggest that
males make adjustments of potentially high impact according to the type of
nest they occupy.
Previous studies by Lissåker et al. (2003) and Jones & Reynolds (1999)
have proposed that the efficiency of fanning enhances with increasing size
of the entrance of the nest. In addition, sand goby males facing conditions
of low dissolved oxygen were found to have a higher fanning expenditure
and parental effort (Lissåker & Kvarnemo, 2006). We propose that the differences between rock and sand environments parallel the regulated oxygen
Habitat specific paternal care
47
treatments of the above-mentioned experiment. Hence, more free water flow
in rock nests with several openings than in sand nests with one small point
of entrance may explain why males fanned less and with shorter bouts in
our rock habitat treatment. Despite the different patterns of fanning behaviour, the survival rate of the eggs did not differ between our two treatments,
suggesting that males with a rocky habitat nest were indeed in a position to
decrease their fanning effort without compromising the success of the current brood. Not only do rock habitat nests seem better ventilated but on average rocky sites are also more exposed to wave action and water movements.
This per se might provide the eggs with a more oxygenated environment. At
some hard bottom localities in the Baltic, males of the three-spined stickleback, Gasterosteus aculeatus, spent exceptionally little time fanning and
used only short fanning bouts (Borg, 1985). In addition, at least one population of sticklebacks is known to disperse their progeny among rocks and
then to leave them without any parental care (MacDonald et al., 1995a,b).
MacDonald et al. (1995a,b) suggested that this novel but stable behaviour
has evolved secondarily to the evolution of emancipation from parental care;
fanning and other care behaviours seem to be unnecessary in certain circumstances.
Oxygen deficiency is harmful for egg development and can result in increased egg mortality (Padilla & Roth, 2001; Hale et al., 2003). Males facing
a decrease in oxygen level usually increase their fanning effort (e.g., Reebs
et al., 1984; Torricelli et al., 1984; Jones & Reynolds, 1999; Takegaki &
Nakazono, 1999) and may consequently lose their body fat reserves at an
increased rate (Lissåker et al., 2003). Thus, we expected that a decrease in
body condition during the parental phase would be more pronounced for
males that fanned more actively, i.e., males with a sand habitat nest. However, we found only a tendency towards the expected pattern. It is possible
that there are costs related to rock habitat nesting that remained unidentified in this study. Nevertheless, we suggest that one possible explanation for
the large size of male sand gobies in rock habitat (Lehtonen & Lindström,
2004) is that these males, instead of intense fanning, are able to invest more
resources in growth than males with a sand habitat nest.
Fanning frequency and the time spent fanning are also likely to increase
with the increasing oxygen consumption of the older eggs (Torricelli et al.,
1984). We found an increase in the fanning rate as the eggs became older
in the experiment with natural nest sites, but only a tendency towards that
48
Järvi-Laturi, Lehtonen, Pampoulie & Lindström
direction when we controlled for nest size. In the sand goby, the increase
in fanning effort is more pronounced towards the end of the breeding season (Pampoulie, Lehtonen & Lindström, unpubl.), which might explain the
difference between our two experiments. The observed fanning effort was exceptionally high in both treatments in the experiment with natural nest sites,
and much higher than when we controlled for nest size. The explanation is
likely to be the high water temperature with a consequent oxygen deficiency,
and possibly the proximity to the end of the breeding season (with a very
low potential for future reproductive success). The season-related conditions
could also explain the smaller treatment difference in the proportion of time
spent fanning in experiment 2 (late season) as compared to experiment 1.
We wanted to ensure as natural setting as possible by only giving males
the type of nest that is available in the habitat they were caught. This design
does not, however, separate between the effects of the nest structure and possible phenotypic differences between the males occupying the two habitats.
The previous study of Lehtonen & Lindström (2004) suggests that such fixed
behavioural differences between the males from the two habitats are small
or non-existent; both habitats are apparently occupied by the members of the
same population. Even if some phenotypic differences existed, our results indicate that males can afford to allocate less in egg care in the rocky than sand
habitat even when measured in standardised aquarium conditions. Stronger
wave action and water movements in the more exposed rocky habitats have
a potential to pronounce this difference in the field.
Why do not all males of the Baltic sand goby population choose a rocky
habitat nest as these not only require a lower fanning effort but can also
harbour a larger number of eggs (Lehtonen & Lindström, 2004)? At this
point, we can only offer tentative candidates for the costs that seem to be
balancing the benefits of occupying a rocky habitat nest. One of the possible
constrains is the briefness of evolutionary time involved: the rocky habitats
that are suitable for breeding sand gobies may only have become available
when the Baltic Sea was formed after the latest glacial period. In addition,
a natural rock habitat nest usually has several possible points of entrance,
which should make the nest more difficult to defend than a nest in the sand
habitat. Specifically, the more open structure of the rock habitat nests may
provide more opportunities for parasitic fertilisations and predation attempts
upon both eggs and the egg-defending male. The increased need for nest
guarding is implied by the fact that the nest type did not affect the time
Habitat specific paternal care
49
the males left their nests although less time was spent fanning eggs in a
rocky habitat nest. Furthermore, it is possible that more debris is churned
among the eggs in the rocky nests in which ventilation is only partly under
the male’s control. Studies addressing these possible costs and constrains
would be highly interesting as they would shed light on the evolutionary
processes that are shaping an important aspect of life history of these fish.
In conclusion, sand goby males seem to exhibit habitat-specific behavioural adaptations: they fan less and with shorter bouts when they are occupying a rock habitat nest, but do not pay any obvious costs from this decrease
in parental effort in terms of hatching success of the eggs they are tending.
Acknowledgements
We thank the Tvärminne zoological station for working facilities. Hanna Kokko, Bob Wong
and anonymous reviewers are acknowledged for their comments on the manuscript. The work
was funded by the Academy of Finland, Walter and Andreé de Nottbeck’s foundation (M.J.-L.
and T.L.) and CIMO (C.P.).
References
Alabrudzińska, J., Kaliński, A., Słomczyński, R., Wawrzyniak, J., Zieliński, P. & Bańbura, J.
(2003). Effects of nest characteristics on breeding success of Great Tits Parus major. —
Acta Ornithol. 38: 151-154.
Borg, B. (1985). Field studies on three-spined sticklebacks in the Baltic. — Behaviour 93:
153-157.
FitzGerald, G.J. & Whoriskey, F.G. (1992). Empirical studies of cannibalism in fish. — In:
Cannibalism: ecology and evolution among diverse taxa (Elgar, M.A. & Crespi, B.J.,
eds). Oxford University Press, New York, NY, p. 238-255.
Hale, R.E., St Mary, C.M. & Lindström, K. (2003). Parental responses to changes in costs
and benefits along an environmental gradient. — Env. Biol. Fishes 67: 107-116.
Jones, J.C. & Reynolds, J.D. (1999). Costs of egg ventilation for male common gobies
breeding in conditions of low dissolved oxygen. — Anim. Behav. 57: 181-188.
Kilpi, M. & Lindström, K. (1997). Habitat-specific clutch size and cost of incubation in
common eiders, Somateria mollissima. — Oecologia 111: 297-301.
Lehtonen, T. & Lindström, K. (2004). Changes in sexual selection resulting from novel
habitat use in the sand goby. — Oikos 104: 327-335.
Lindström, K. (1988). Male–male competition for nest sites in the sand goby Pomatoschistus
minutus. — Oikos 53: 67-73.
Lindström, K. (2001). Effects of resource distribution on sexual selection and the cost of
reproduction in sand gobies. — Am. Nat. 158: 64-74.
Lissåker, M. & Kvarnemo C. (2006). Ventilation or nest defense – parental care trade-offs in
a fish with male care. — Behav. Ecol. Sociobiol. 60: 864-873.
50
Järvi-Laturi, Lehtonen, Pampoulie & Lindström
Lissåker, M., Kvarnemo, C. & Svensson, O. (2003). Effects of low oxygen environment on
parental effort and filial cannibalism in the male sand goby, Pomatoschistus minutus. —
Behav. Ecol. 14: 374-381.
MacDonald, J.F., Bekkers, J., Macisaac, S.M. & Blouw, D.M. (1995a). Intertidal breeding
and aerial development of embryos of a stickleback fish (Gasterosteus). — Behaviour
132: 1183-1206.
MacDonald, J.F., Bekkers, J., Macisaac, S.M. & Blouw, D.M. (1995b). Experiments on
embryo survivorship, habitat selection, and competitive ability of a stickleback fish
(Gasterosteus) which nests in rocky intertidal zone. — Behaviour 132: 1207-1221.
Manica, A. (2002). Filial cannibalism in teleost fish. — Biol. Rev. 77: 261-277.
Morris, D.W. (1987). Tests of density-dependent habitat selection in a patchy environment.
— Ecol. Monogr. 57: 269-281.
Östlund, S. & Ahnesjö, I. (1998). Female fifteen-spined sticklebacks prefer better fathers. —
Anim. Behav. 56: 177-183.
Padilla, P.A. & Roth, M.B. (2001). Oxygen deprivation causes suspended animation in the
zebrafish embryo. — Proc. Natl. Acad. Sci. USA. 13: 7331-7335.
Reebs, S.G., Whoriskey, F.G. & FitzGerald Jr., G.J. (1984). Diel patterns of fanning activity,
egg respiration and the nocturnal behavior of male three-spined sticklebacks, Gasterosteus aculeatus L. (f. Trachurus). — Can. J. Zool. 62: 329-334.
Rohwer, S. (1978). Parent cannibalism of offspring and egg raiding as a courtship strategy.
— Am. Nat. 112: 429-440.
St Mary, C.M., Noureddine, C.G. & Lindström, K. (2001). Environmental effects on male
reproductive success and parental care in the Florida flagfish Jordanella floridae. —
Ethology 107: 1035-1052.
Takegaki, T. & Nakazono, A. (1999). Responses of the egg-tending gobiid fish Valenciennea
longipinnis to the fluctuation of dissolved oxygen in the burrow. — Bull. Mar. Sci. 65:
815-823.
Torricelli, P., Lugli, M. & Gandolfi, G. (1984). A quantitative analysis of the fanning activity
in the male Padogobius martensi (Pisces: Gobiidae). — Behaviour 92: 288-301.