Environmental Biology of Fishes

, Volume 92, Issue 2, pp 207–215

How boat noise affects an ecologically crucial behaviour: the case of territoriality in Gobius cruentatus (Gobiidae)

Authors

    • Cognitive Neuroscience SectorSISSA, International School for Advanced Studies
  • Marta Picciulin
    • Etho-ecology Marine LaboratoryNatural Marine Reserve of Miramare—WWF
    • Department of Life SciencesUniversity of Trieste, CSEE
  • Marco Costantini
    • WWF Italia
  • Enrico A Ferrero
    • Department of Life SciencesUniversity of Trieste, CSEE
Article

DOI: 10.1007/s10641-011-9834-y

Cite this article as:
Sebastianutto, L., Picciulin, M., Costantini, M. et al. Environ Biol Fish (2011) 92: 207. doi:10.1007/s10641-011-9834-y

Abstract

Gobius cruentatus emit sounds during agonistic interactions. In order to evaluate the effect of boat noise exposure on G. cruentatus territorial behaviour, we played a field-recorded diesel engine boat noise during aggressive encounters between an intruder and a resident fish in a laboratory-controlled tank. We tested two factors: role (resident vs. intruder) and condition (noisy vs. silent); the test animals underwent all the treatments in a round-robin design. Agonistic behavior of the residents was modified by boat noise: during the playback residents were more submissive and won less encounters than in the control (silent) condition. We suggest that sound production is an effective tool for territorial defense, since the impairment of acoustic communication due to the recreational boat noise diminished the ability of the resident to maintain its territory.

Keywords

Agonistic interactionNoise pollutionFish soundsBoat noise

Introduction

Anthropogenic noise is a form of pollution that is of increasing concern these decades, with the potential for causing significant threats on marine fish, including temporary hearing loss, impaired temporal resolution ability, damages to the sensory epithelia of the inner ear and endocrinological stress responses (Popper et al. 2004; Slabbekoorn et al. 2010). Several human activities generate noise pollution: offshore drilling, geological exploration, commercial shipping, and port activities. Among these, motorized vessels are the most common source of anthropogenic noise in coastal waters, due to their high number and mobility (Haviland-Howell et al. 2007). Although lower than high-intensity acute noise of airguns or military sonars, boat noise is a continuous, chronic noise source. Several studies address the short-term effects of boat noise on fish behaviour on both pelagic and benthic species. This kind of noise causes cessation of activities, provokes the so-called ‘startle response’, induces modification in the fish position in water column as well flight behaviour (Slabbekoorn et al. 2010) and has the potential to impair detection of relevant signals by reducing the extent of signal propagation and therefore signalling efficiency between individuals (Amoser et al. 2004; Vasconcelos et al. 2007, Codarin et al. 2009). Recently a time budget analysis, applied for the first time on a fish species, revealed less overt behavioural changes due to boat noise exposure (Picciulin et al. 2010). Despite these studies, little is known about the noise effect on biologically relevant behaviours such as reproduction and territory defence. The present study aims to describe the territorial behaviour of the red mouthed goby (Gobius cruentatus, Gmelin 1789) during playback of boat noise in laboratory conditions.

G. cruentatus is a benthic species, extremely common at low depths in the Mediterranean Sea and in West-Atlantic Ocean, which lives both on sandy and rocky bottoms and use holes and crevices as shelter (Wilkins and Myers 1991, 1993). The territorial behaviour and acoustic repertoire of G. cruentatus have been described in detail by Picciulin et al. (2006) and Sebastianutto et al. (2008). In laboratory conditions, a resident fish shows very aggressive visual displays towards conspecifics, accompanied by acoustic emissions used with a warning function. According to the bourgeois principle “fight if owner, retreat if intruder” (Maynard Smith 1982) the resident fish, independently of its size and sex, wins territorial encounters as resident whereas the same individual as intruder becomes a loser and exhibits submissive behavior (Picciulin et al. 2006).

During territorial encounters, G. cruentatus acoustic repertoire consists of four different types of sounds emitted by both sexes: a tonal sound, a noisy tonal sound, a train of pulses, and a complex sound, made up of a tonal part immediately followed by a train of pulses. The peak frequencies of these sounds range between 82 and 185 Hz (Sebastianutto et al. 2008), matching well with the best hearing range of the species i.e. 300 Hz for sound pressure and 200 Hz for particle velocity (Wysocki et al. 2009). The production of sounds during territory defense is a widespread trait of many teleost species (reviewed by Ladich and Myrberg 2006); vocalizations are produced as ‘keep-out’ and threatening signals (Valinksy and Ridgley 1981; Raffinger and Ladich 2009; Vasconcelos et al. 2010) and are used for mutual assessment (Ladich 1997). In G. cruentatus sounds are deterrent signals mainly emitted by the resident fish during non-escalated phases of the encounter (i.e. during rest and alert position, Sebastianutto et al. 2008) and during the circling, i.e. the climax of the agonistic contest (Picciulin et al. 2006; Sebastianutto et al. 2008 but also Enquist et al. 1990; Ota et al. 1999 for other fish species). This is similar to other goby species, that use multimodal communication-not only visual but also acoustic displays–for a positive resolution of agonistic interactions (Amorim and Neves 2008; Meunier et al. 2009).

Recently Codarin et al. (2009) showed that boat traffic, which generates low-frequency noise below 1 kHz (Richardson and Würsig 1997), has the potential to negatively affect G. cruentatus hearing sensitivity. In a physiological study the fishes exposure to the noise produced by a cabin cruiser was accompanied by a substantial reduction (up to 10 dB) of the species’ auditory threshold. Considering the role played by residents’ sound production in the conflict (Sebastianutto et al. 2008), masking the acoustic intraspecific communication may also affect the output of territorial interactions.

Methods

Animals

Five adults (Total Length 13.2 ± 1.5 cm (mean±SD) of G. cruentatus were captured with trap nets outside the breeding season. According to the shape of the genital papilla (Tortonese 1975), they were three females and two males. Specimens were kept individually in small tanks (36 × 20 × 22 cm) and provided with artificial shelters. To avoid visual contact between fish, an opaque board was positioned externally between the walls of adjoining tanks. Automatically regulated lighting, which followed the natural light-dark cycle, was provided. Tank temperatures throughout the whole observation period (see below) ranged between 18 and 20°C. Fish were always fed with small shrimps and pieces of Atherina sp. after each territorial encounter. At least 2 weeks of acclimation occurred before the experiment started.

Design

Contests were staged with a round-robin tournament, in which every animal competed against every other (Gill and Swartz 2001) as resident and as intruder, and contests occurred both in the noisy–hereafter NC-and in the silent condition–hereafter SC–, for a total of 40 encounters (20 in SC and 20 in NC). The order of treatments followed a pseudo-random order. The within-subject has some advantages over the most common between-subject design, since it allows reduction of the effect of individual variability, is more powerful, and requires a lower sample size (McBurney 1998). Two experimental factors were tested: fish role, with two levels (resident R and intruder I), and the background noise condition, with two levels (SC and NC).

Territorial encounters were run between February and November, with a pause during the G. cruentatus breeding season (late spring). In order to prevent the effect of the previous outcome, at least two days passed between two fights of the same couple (Chase et al. 1994). All the animals were returned to the site of capture at the end of experimental period. The experimental set-up was based on that described by Picciulin et al. (2006) and Sebastianutto et al. (2008) and is briefly summarized below.

During SC, a fish (the intruder, I) was introduced into the experimental tank (120x50x60 cm), provided with an artificial plastic shelter on one side, where another fish had been acclimated in isolation for a week, therefore becoming the resident (R). Fish were allowed to interact for 10 minutes. Intruders were removed as soon as the contest was settled (i.e. the winner, either R or I, swam inside the shelter and stayed at its entrance) to avoid persistent stress on the loser. The same set-up was applied during the NC, although a boat noise was played with an underwater speaker suspended at the opposite side respect to the artificial shelter (70 cm far from the shelter) during the whole encounter, starting from the moment when the intruder was introduced in the tank.

Once the resident had fought against all the other fish in the group, it was replaced in the experimental tank by another fish, which became the new resident.

All the encounters were video-recorded by a video camera SONY VIDEO 8 TR 805 (10X) connected to a VHS videotape recorder. Sounds from the experimental tank were collected with a pre-amplified hydrophone Reson TC 4032 (sensitivity−170 dB re 1 V/μPa; frequency range: 5 Hz–120 kHz ) suspended just above the artificial shelter, stored on a DAT recorder Pioneer DC-88 (sampling rate 44.1 KHz, 16-bit), recorded on the VHS videotape and monitored on headphones. During the NC, the speaker (AQ339 Clark Synthesis, frequency range 35 Hz–17 kHz) was connected to a power amplifier and a DAT Recorder Sony TCD-D100. Calibration gain chain was periodically controlled and kept constant throughout the study.

Playback

As noise stimulus we recorded the acoustic emission of a 5 m-fibreglass boat with a 40 HP inboard diesel engine operating at maximum speed (15 knots). The recording occurred during daytime in the coastal rocky reef of the Marine Reserve of Miramare (at 45°42′08″ N latitude and 13°42′42″ E longitude), where fish density of target species was high. For a detailed description of boat noise recordings see Picciulin et al. (2010).

Samples of 25 s were considered for the playback, the boat samples included the highest amplitude value of the noise. The 25 s boat noise was played-back in loop for the whole duration of each NC-encounter using the underwater speaker. The projected noise was recorded by the Reson TC4032 hydrophone placed above the goby shelter, connected to a Pioneer DC-88 DAT (sampling rate 44.1 kHz, 16-bit). The power spectra of 1 min-projected boat noise were calculated with Spectra RTA (Sound Technology) spectral analyser and superimposed on the original boat noise spectrum, in order to compare sound pressure levels (SPL, L-weighted, 20 Hz to 20 kHz, RMS fast) and frequency distributions. The volume of the playback was adjusted in order to resemble the same SPL level recorded in the field. The signal was previously calibrated with a signal of 100 mV RMS @1 kHz recorded at the beginning of the tape. Although the tank walls caused reverberation, according to Akamatsu et al. (2002) the minimum estimated resonant frequency that could produce distortion in the experimental arena was 2050 Hz, therefore well above the upper limit of G. cruentatus hearing range (Wysocki et al. 2009).

Data analysis

All audio-video recorded experiments were analyzed frame-by-frame and every interaction was classified and logged on previously prepared ad hoc check-lists using Etholog 2.2 (Ottoni 2000). The behavioral units were scored according to the G. cruentatus’ ethogram (Picciulin et al. 2006), which consists of 26 behavioral units, grouped in 5 behavioral categories described in Table 1. We focused in particular on the behavioral unit ‘circling’, in terms of duration and frequency of occurrence. This very aggressive behavioral unit can be considered the climax of the encounter: during circling the two combatants assess each other with sounds, beats and bites and after this the contest is often settled (Picciulin et al. 2006; Sebastianutto et al. 2008).
Table 1

Description of the territorial behavior of G. cruentatus. For a more detailed description of each behavioral category and unit see Picciulin et al. (2006)

Behavioral category

Brief description

Behavioral units

Inferiority and retreat (INF)

Different types of submissive behavior

Backward swimming, flank position, fluttering, flee

Locomotion behavior (LB)

Different types of swimming

Directed swimming, scraping swimming, oscillating swimming, vertical position, changing position, wandering

Rest and alert position (RAP)

Different types of stationary positions

Stationary position, reared stationary position with spread fins, orthogonal display, lateral display

Confrontation behavior (CB)

Aggressive displays without physical contact

Reared lateral display, reared orthogonal display, reared stationary position with spread fins gaping, changing position gaping, reared orthogonal display gaping, reared lateral display gaping

Threatening and chasing behavior (TCB)

Very aggressive displays with physical contact

False thrust, thrust, beat, arched flank, circling, pursuit

Since the projected noise masked all the vocalizations in the NC, we also considered the occurrence of a new behavioral unit, i.e. Sound (S: the fish opens opercula, the ventral fin shivers and the body rapidly oscillates), in order to compare sound production in the two experimental conditions (SC vs. NC). In a pre-test on 20 encounters run by Rocca (2001) this behavior (n = 112) was associated with sound production 85% of the times; the co-occurring vocalizations were either train of pulses (64%) or complex sounds (36%).

The total duration (s) and the total number of acts (frequency of occurrence of behaviors in each trial) have been calculated per each of the 5 behavioral categories and per fish role (R or I). Once the normality of measured variables was tested using Kolgomorov-Smirnov test, the fish behavior (in terms of duration and frequency) was analyzed with Repeated Measures ANOVA, considering two within-subject factors (Role, i.e. resident and intruder; Condition, i.e. SC and NC) and a between-subject factor (Fish).

A Wilcoxon test was used to test (i) the number of encounters won by each fish according to Role and Condition (ii) the effect of Role and Condition on the duration and the frequency of occurrence of the ‘circling’ behaviour, (iii) the occurrence of behavioral unit “S” by each fish in the two conditions, (iiii) number of vocalizations emitted by the resident and the intruder only in the SC.

In addition to p-level values, effect sizes were provided in terms of partial eta squared, \( \eta_{\text{p}}^2 \). This informative quantity represents the proportion of the total variance that is attributable to the effect, and according to Cohen (1988) a \( 0.01 < {{\eta }}_{\text{p}}^2 < 0.06 \) is considered a small effect, \( 0.06 < {{\eta }}_{\text{p}}^2 < 0.14 \) is considered a medium effect and \( {{\eta }}_{\text{p}}^2 > 0.14 \) is considered a large effect. Statistical analyses were performed using the StatSoft 8.0 (StatSoft, Inc.) software.

Results

Comparison between the projected noise and the species’ pressure audiogram

The equivalent continuous SPL (Lleq, 1 min) of the played back boat was 161 dB re 1 μPa, with a maximum instantaneous SPL of 165.9 dB re 1 μPa. Within the hearing range of G. cruentatus measured in terms of pressure by Wysocki et al. (2009), the noise stimulus showed a peak of 126 dB re 1 μPa at 86 Hz and another peak of 129 dB re 1 μPa at 300 Hz. Comparisons between boat power spectra and the G. cruentatus pressure audiogram indicated that the played back noise was detectable by the fighting animals in the frequency range between 70 and 400 Hz (Fig. 1), and that—on average—its sound energies exceeded the hearing thresholds of 12–25 dB.
https://static-content.springer.com/image/art%3A10.1007%2Fs10641-011-9834-y/MediaObjects/10641_2011_9834_Fig1_HTML.gif
Fig. 1

Comparison between the 1/3 octave band spectrum of the played back stimulus and Gobius cruentatus audiogram (Wysocki et al. 2009) in terms of Sound Pressure Levels (SPL)

Outcomes of the encounters

During SC, the resident and the intruder won 65% and 25% of the twenty encounters respectively, while during the NC the resident won 40% and the intruder won 45% of the encounters (Fig. 2). In three encounters in the NC and in two encounters in the SC fish did not interact at all. Wilcoxon test showed that residents won significantly more than intruders only in the SC (n = 5; Z = 2.02, p = 0.04), while this was not the case in the NC.
https://static-content.springer.com/image/art%3A10.1007%2Fs10641-011-9834-y/MediaObjects/10641_2011_9834_Fig2_HTML.gif
Fig. 2

Comparison of the number of encounters won (mean ± SE) in the two roles either in noisy (n = 20) or in silent condition (n = 20). * indicates significant difference according to Wilcoxon test

Fish behaviour under different treatments

Results of Repeated Measures ANOVA on Duration and Frequency of acts were reported in Table 2. Fish spent more time swimming in the NC (79.23 s ± 75.05 SD) than in the SC (56.30 s ± 40.11 SD), irrespective from the role. A significant interaction Condition x Role was found for the inferiority behaviours (INF), the resident spending more time displaying submissive behaviors in the NC (4.20 s ± 5.32 SD) than in the SC (1.00 s ± 1.81 SD), while the intruder resulting more submissive in the SC (48.20 s ± 65.11 SD) than in the NC (24.60 s ± 49.36 SD) (Fig. 3). In addition, during the NC the circling lasted significantly more than in SC (Wilcoxon test: n = 5, Z = 2.02, p = 0.04; Fig. 4). The Role on the other hand had no effect neither on the duration nor on the frequency of circling (Wilcoxon test, all ps > 0.05).
Table 2

Results of Repeated Measures ANOVAs carried on Duration (i.e. total time spent) and Frequency of acts as dependent variables. ** and bold show significant results

Behavioral Category

Factor

Df

Duration

Frequency of acts

F

P

ηp2

F

p

ηp2

LB

FISH

4

1.85

0.17

0.33

3.72

0.03**

0.50

COND

1

4.79

0.04**

0.24

0.33

0.57

0.02

COND × FISH

4

0.60

0.67

0.14

0.55

0.71

0.13

ROLE

1

0.99

0.34

0.06

9.14

0.01**

0.38

ROLE × FISH

4

1.68

0.21

0.31

1.51

0.25

0.29

COND × ROLE

1

0.24

0.63

0.02

0.34

0.57

0.02

COND × ROLE × FISH

4

0.97

0.45

0.20

1.99

0.15

0.35

RAP

FISH

4

0.84

0.52

0.18

3.53

0.03**

0.48

COND

1

0.80

0.38

0.05

0.22

0.64

0.01

COND × FISH

4

4.09

0.02**

0.52

1.98

0.15

0.35

ROLE

1

3.24

0.09

0.18

16.15

0.00**

0.52

ROLE × FISH

4

0.70

0.60

0.16

2.17

0.12

0.37

COND × ROLE

1

0.50

0.49

0.03

0.14

0.71

0.01

COND × ROLE × FISH

4

1.25

0.33

0.25

1.81

0.18

0.33

CB

FISH

4

1.16

0.37

0.24

2.33

0.10

0.38

COND

1

0.30

0.59

0.02

0.45

0.51

0.03

COND × FISH

4

0.72

0.59

0.16

0.63

0.65

0.14

ROLE

1

23.51

0.00**

0.61

18.89

0.00**

0.56

ROLE × FISH

4

2.69

0.07

0.42

3.13

0.05**

0.46

COND ×ROLE

1

1.53

0.24

0.09

0.30

0.59

0.02

COND × ROLE × FISH

4

0.61

0.66

0.14

0.29

0.88

0.07

TCB

FISH

4

1.66

0.21

0.31

2.34

0.10

0.38

COND

1

1.22

0.29

0.08

0.64

0.43

0.04

COND × FISH

4

0.99

0.44

0.21

0.41

0.80

0.10

ROLE

1

0.83

0.38

0.05

26.49

0.00**

0.64

ROLE × FISH

4

4.16

0.02**

0.53

3.17

0.04**

0.46

COND × ROLE

1

0.78

0.39

0.05

0.44

0.52

0.03

COND × ROLE × FISH

4

0.66

0.63

0.15

0.43

0.79

0.10

INF

FISH

4

0.96

0.46

0.20

1.33

0.31

0.26

COND

1

3.07

0.10

0.10

0.10

0.76

0.01

COND × FISH

4

2.82

0.06

0.43

0.55

0.70

0.13

ROLE

1

10.75

0.01**

0.42

8.68

0.01**

0.37

ROLE × FISH

4

1.39

0.29

0.27

2.31

0.11

0.38

COND × ROLE

1

4.75

0.05**

0.24

2.71

0.12

0.15

COND × ROLE ×FISH

4

2.53

0.08

0.40

0.35

0.84

0.08

https://static-content.springer.com/image/art%3A10.1007%2Fs10641-011-9834-y/MediaObjects/10641_2011_9834_Fig3_HTML.gif
Fig. 3

Duration of submissive behavior INF (mean ± SE) in the two conditions for resident and intruder, respectively. * indicates significant differences according to LSD post-hoc test

https://static-content.springer.com/image/art%3A10.1007%2Fs10641-011-9834-y/MediaObjects/10641_2011_9834_Fig4_HTML.gif
Fig. 4

Duration (mean+SD) of behavioral unit circling in the SC (n = 20) and the NC (n = 20). * indicates significant difference according to Wilcoxon test

Sound production

In the SC and considering the whole acoustic repertoire (n = 150, with 63% of train of pulses, 2% of complex sounds, 14% of tonal sounds and 21% of noisy tonal sounds), the resident vocalized significantly more than the intruder (Wilcoxon test n = 5; Z = 2.02, p = 0.04). On the other hand, Wilcoxon tests did not reveal significant effects of the Role (resident vs. intruder) or the Condition (NC vs SC ) on ‘S’ occurrence, despite the fact that in the NC this parameter was higher in the resident than in the intruder (NC: resident 4.55 ± 7.47, intruder 0.30 ± 0.33; SC: resident 4.25 ± 7.06, intruder 0.50 ± 0.12).

Discussion

Anthropogenic disturbances in the environment can rapidly change many habitats. Studies on human-made environmental changes in the Baltic sea, for example, show that eutrophication and reduced visibility in the environment affect sexual selection on mating traits in sticklebacks, increasing courtship activity and potentially mediating negative effects on population (Candolin et al. 2007; Wong et al 2007). Despite that fact that enhancement of the low-frequency noise component in shallow waters can create a chronic situation of low-level stress for benthic species, few studies so far have investigated the detrimental behavioral effects of noise on fitness-related activities. One of the few has shown that Chromis chromis diminishes the time spent in the nest guarding and G. cruentatus reduces the time spent in territory patrolling during boat noise exposure (Picciulin et al. 2010).

The present paper is the first study to investigate territorial defense in fish subjected to boat noise, demonstrating that the resident fish becomes less successful in repelling intruders in noisy conditions. The reduced ability to maintain the territory cannot be related to a diminished aggressive behavior of the resident, since it is more aggressive than the intruder during both the SC—in accordance with Picciulin et al. (2006)—and the NC, but instead it might be caused by an impaired acoustic communication between opponents.

Boat noise does not modify G. cruentatus sound production, when the latter is defined as the occurrence of the behavioural unit Sound (S). Unfortunately S can be considered only a rough—although the only one possible—measure of the ‘real’ sound production. On the other hand, during the NC sound perception of the fish was likely affected by the projected noise, whose peak energy matches the frequency range of species-specific vocalizations (Sebastianutto et al. 2008), largely exceeding the best G. cruentatus pressure-hearing thresholds (Wysocki et al. 2009). This needs to be completed by particle motion measurements of both fish sounds and boat noises, because gobies are commonly considered to be hearing generalists (Lugli et al. 2003) and are probably primarily sensitive to the particle motion of sound (i.e. particle acceleration), but we can reasonably conclude that during the NC the fish communication was impaired.

A reduced vocal interaction between opponents may be thought to be responsible for the observed modifications in the resident and intruder submissive behaviour as well for the longer duration of the circling. This would ultimately lead to a diminished number of wins for the resident fish during the NC. This interpretation is in agreement with Valinksy and Ridgley (1981), who showed that experimentally muted territorial residents of the skunk loach Botia horae (Cobitidae) were unsuccessful in repelling intruders.

In conclusion, we showed a negative impact of boat noise on territorial behaviour of G. cruentatus, likely to be because of a masked acoustic communication. A diminished ability of the territorial fish to hold the territory has been proven to lead to an increase in aggressive contests, raise the levels of androgens in the resident fish to the detriment of courtship behaviour, reproductive activities and care of offspring (Oliveira et al. 2002; Amorim and Almada 2005). In addition, it may affect the spacing of territories in territorial species, or increase the energy spent for honest signaling during social interactions (either agonistic or reproductive), changing composition and viability in the long-term population. Therefore, further research needs to be addressed at monitoring long-term implications of boat noise disturbance on fish communities.

Acknowledgments

This study was part of a project to monitor human-made noise in Marine Protected Areas supported by the Italian Ministry for Environment, Territory and Sea. We would like to thank Maurizio Spoto and the Miramare Natural Marine Reserve staff for technical assistance, Elena Pangaro for her support in data collection, Antonio Codarin for help in acoustic analyses, MPC Amorim for critical reading and valuable comments, and Valerie Lesk for English proofreading.

Copyright information

© Springer Science+Business Media B.V. 2011