An investigation into active and passive acoustic techniques to study aggregating fish species

Abstract

Techniques of single- and multi-beam active acoustics and the passive recording of fish vocalisations were employed to evaluate the benefits and limitations of each technique as a method for assessing and monitoring fish aggregations. Five species, Samson fish (Seriola hippos), mulloway (Argyrosomus japonicus), West Australian dhufish (Glaucosoma hebraicum), Bight redfish (Centroberyx gerrardi) and pink snapper (Pagrus auratus) were investigated on the basis of their abundance, ecological importance and differing behaviour. The primary focus was on S. hippos, a large nonvocal schooling fish, and A. japonicus, a large vocal fish, with each species forming aggregations for spawning purposes.Simrad EQ60 single-beam echosounder assessments of mid-water, S. hippos aggregations at seven sites west of Rottnest Island illustrated the relative biomass increase, stabilisation and decrease between the months of October and March each year from October 2004 to March 2007. Surveys highlighted the preferred sites for spawning, spatial extents of each aggregation, as well as a decline in aggregation stability at full moon and end of season periods. Regular Department of Fisheries surveys displayed the relative ease with which single-beam techniques could be deployed, used and data analysed to monitor large, comparatively stable, deep (>50 m) aggregations of large swimbladdered fish. Acquired acoustic data illustrated the limitations of single-beam surveys conducted on a mobile school of fish.RESON 8125 and 7125 multi-beam sonar (MBS) surveys of S. hippos at Rottnest Island locations, some conducted simultaneously with the Simrad EQ60 single-beam, illustrated the improved spatial resolution of midwater targets achievable with MBS systems. The identification of individual S. hippos targets in MBS data facilitated the confirmation of S. hippos undetected by single-beam transects, due to relative sampling volumes. The MBS surveys showed evidence of possible fishing effects on S. hippos aggregations with school structure varying after a two hour period of fishing and video tows. Relative decline in aggregation stability towards the end of the season and possible avoidance behaviour from approaching vessels was observed as successive MBS transects, over a short space of time, recorded school movement around the wreck above which S. hippos aggregations sit.P. auratus spawning in the Cockburn Sound, Fremantle illustrated the limitations of single-beam acoustics to monitor aggregations of mobile fish in shallow water, due to vessel avoidance behaviour. Similar sampling issues were observed in MBS surveys despite the inherent geometric advantages of the wide acoustic swath and increased sample volume. It was anticipated that adjusting the MBS mounting position, such that nadir beams were orientated laterally athwartships (rather then vertically downwards), increased the lateral distance at which the fish could be observed, thus reducing vessel avoidance implications. However, due to time constraints and equipment availability, remounting the MBS was not possible at the time of survey and the effects of MBS orientation could not be verified.Single-beam and passive acoustic surveys of G. hebraicum illustrated the complexity of acoustic investigation of comparatively sedentary, demersal fish which often spawn in small groups. Discrimination of individuals using single-beam techniques was often restricted by the fish proximity to the seafloor and the footprint of the single-beam. Single-beam species identification of small groups of fish is impractical without simultaneous visual confirmation, due to the stochastic nature of fish reflectance. However, single-beam acoustics could provide information on G. hebraicum spawning related essential fish habitat using seafloor classification. While biological assessment of G. hebraicum otoliths, swimbladder and related muscular structure imply a soniferous species, as yet no vocal behaviour of any of the Glaucosomatidae has been reported, despite attempts here to detect vocalisations. Thus the characteristics of this species presented the greatest limitations for study using active or passive acoustic techniques.Passive acoustic techniques were shown to be ideally suited for monitoring the low density, benthic aggregations of A. japonicus in the Swan River. Spawning related vocalisations of A. japonicus were recorded in situ and in aquaria (Mosman Bay, Swan River and TAFE, Fremantle aquaculture centre respectively) each spawning season between October and May for four spawning seasons. A. japonicus calls, produced by the contraction of bi-lateral paired sonic muscles around the posterior two thirds of a heavily damped swimbladder, were classified into three categories relating to differing spawning functions. Category 1 calls (‘Bup’) of 2-4 swimbladder pulses were believed to function to gather males together in temporary broadcasting territories and to announce readiness to spawn. Category 2 calls comprised 11-32 pulses in a single audible tone (‘Baarp’), which could also be broken into two or more parts (‘Ba-Baarp’) with a believed function as a call of attraction, predominantly from males to females. The third Category comprised calls produced in quick succession at increasing call rate to a point of cessation. Series of Category 3 calls (‘Thup’) were recorded only at times associated with spawning, in fewer numbers than other call categories and consisted of between 1 and 4 pulses.Pulse repetition and spectral peak frequencies of Category 3 calls were notably higher than those of Category 1 and 2, both in situ and in aquaria, despite the similar number of pulses. For example, in situ pulse repetition frequencies of up to 114 Hz for Category 3 calls compared with approximately 59 Hz for other categories. It is suggested that the increased pulse repetition frequencies of Category 3 calls require greater, unsustainable levels of energy (corroborated by the decreasing pulse rate as these calls progress) and such calls are therefore reserved for specific, uncommon events, possibly episodes of courtship. Ground truth in aquaria calls exhibited similar call structure to those recorded in situ, however, pulse repetition rates and occurrence were significantly lower (respective pulse repetition frequencies of 41.74 and 58.68 Hz for captive and in situ Category 2 calls).Season-long monitoring of sound production in Mosman Bay determined spawning commencement was correlated with a daytime water temperature threshold at, or above 18.5 °C, occurring between October and November. Generalized Additive Models showed sound pressure levels (SPLs) and, by proxy spawning throughout the season, were correlated with temperature, salinity, sunset and tidal effects with decreasing order of effect. Increases in short-term sound production were observed on a semi-lunar basis, occurring at the new and full moons. Local chorus level maxima were found to occur on a 3.97 day basis (s.d. = 1.8), similar to that found from egg collection in aquaria and previous in situ SPLs in local studies of A. japonicus. Comparisons between Mosman Bay tidal related afternoon/evening activity and nocturnal behaviour of alternative populations in captivity suggest that A. japonicus exhibits adaptive vocal behaviour, and by proxy spawning activities, dependent on environmental variables.Individual A. japonicus were localised during spawning within and close to an array of hydrophones by using vocalisation arrival-time differences, surface reflection and comparative energy level techniques to analyse vocalisations. Several individual A. japonicus were followed for periods ranging from seconds to several minutes as they called repetitively. Monitoring individual movement and separation distances between calling fish confirmed low mobility over long periods, indicative of lekking behaviour. The determination of call source levels employed calls of known range using data from the localisation study. Mean squared pressure source levels and 95 % confidence limits of the three call categories were measured as: 163 (147.7, 178.6), 172 (168.4, 176.0) and 157 (154.0, 160.3) dB re 1μPa for Categories 1, 2 and 3, respectively.During periods of low density calling in the 2006-7 spawning season, techniques of call counting produced absolute abundance estimates for A. japonicus present within the hydrophone detection range of approximately 500 m, observing a maximum of 15 calling individuals. Assuming a 1.3:1 sex ratio this implies a detectable spawning population of 26 fish within approximately 100, 000 m[superscript]2 (range restricted across stream by depth) equivalent to approximately 3, 850 m[superscript]2 per fish (assuming a random distribution of callers and recipients). However, during high density ‘continuous chorus’ calling the maximum number of callers able to be discerned using call counting techniques was exceeded. The application of call counting techniques and call contributions to overall SPLs to estimate biomass during ‘chorus’ calling, where calls merge together, requires further investigation. Recorded chorus levels were not a simple function of animals calling within the receiver proximity, but were strongly influenced by source-receiver range. A preliminary model to estimate minimum numbers of callers within derived range boundaries has been laid out.Recording of A. japonicus vocalisations illustrated the developing capabilities of passive acoustics to monitor soniferous fish species. A suggested set of protocols has been laid out to standardise the reporting of fish calls together with supplementary data relating environmental variables to their subsequent effects on the acoustic characteristics of the call. Standardisation of reporting will facilitate future spatial and temporal comparison of inter- and intra-species sound production.This study has illustrated that the features of each acoustic technique endear them to particular species-specific characteristics. For example, although S. hippos did not vocalise they formed midwater aggregations of large fish (107 cm mean fork length) and were thus amenable to active acoustic monitoring. In contrast, A. japonicus form low density, benthic aggregations and hence are not suited to study by active acoustics, but vocalised profusely rendering them suitable for passive acoustic monitoring. In many cases a combination of techniques both acoustic and non-acoustic is required to monitor the particular species, in order to ground truth the data.