Introduction
Marine and fresh water resources have long been recognized as an important food source, and more generally, as means of economic activity in both industrial and artisanal societies. In fisheries areas, these resources have been exploited for decades, raising further concerns about fish stock sustainability in the future (Beverton, 1990). Hence, various methods have been used to determine fish abundance in the water column. Hydroacoustics, a popular method for effectively gathering information on temporal and spatial fish distribution and abundance, has become increasingly sophisticated over the years (MacLennan and Simmonds, 2005). Field application of acoustic methods to estimate animal abundance requires information on the acoustic size, and the target strength (TS) or backscattering cross-section of individual organisms (MacLennan, 1990; Thiebaux et al., 1991). Moreover, fish TS is a key variable in the acoustic assessment of fish abundance (Foote, 1987).
Empirical measurements of backscatter as a function of fish length have been extensively studied and used throughout the fisheries acoustics research community.
Additionally, modeling of acoustic backscatter provides a quantitative tool to examine the variability in backscatter measurements, increases the accuracy of the estimation of TS, and aids discrimination among types of acoustic targets (Horne and Jech, 1999).
Fish TS is influenced by biological factors, which include fish length (Love, 1971; Nakken and Olsen, 1977; Foote and Traynor, 1988), presence of a swim bladder (Clay and Horne, 1994), and the tilt of the fish relative to the incident acoustic wave (Love, 1971; Nakken and Olsen, 1977; Blaxter and Batty, 1990). In general, the three techniques used to measure the TS of fish with swim bladders are in situ (i.e., in- habitat/the natural environment of the fish), ex situ (i.e., in controlled experiments), and numerical or theoretical models based on organism morphology.
Development of backscatter models has provided the ability to manipulate variables over broader ranges than feasible with ex situ experiments (Foote, 1985; Stanton, 1989; Clay and Horne, 1994). Backscatter models, which have been used in conjunction with ex situ and in situ measurements, have demonstrated fair to excellent agreement between model predictions and empirical measurements (Foote, 1985; Foote and Traynor, 1988; Clay and Horne, 1994; Sawada, 1999; Hazen and Horne, 2004
Therefore, if a study is conducted using a backscatter model, it should consider those factors that potentially affect the shape, volume, and density of the swim bladder, and thus the backscatter measurements (Blaxter and Batty, 1990; Koumoundouros et al., 2000). The culmination of several backscatter modeling efforts are represented by the Kirchhoff-ray mode (KRM) model, which can be used to investigate acoustic backscatter from individual and aggregations of aquatic organisms. Then, backscatter models can be used to calculate echo amplitudes for individuals or groups with known size distributions. The model predictions can be compared to measurements in the laboratory and in the survey field (i.e., in the water), and new uses for backscatter models continue to be discovered.
The objectives of the present study were to estimate the TS of bambooleaf wrasse (Pseudolabrus japonicus) using a theoretical model, examine the TS values produced by the model by comparison with the measured values, and examine the characteristics of the backscatter model in its application to the estimation of the fisheries abundance.
Materials and methods
The bambooleaf wrasse is a marine resource developed as a food resource from fish living in Korean waters. This reef-associated species is commonly distributed in the coastal waters of the northwest Pacific. Its distribution in Korean waters covers the southern Korea Peninsula (i.e., Jeju Island) and the southern Japan Sea (Kim et al., 2001), but scant information is available on the TS for this species.
The culmination of several backscatter modeling efforts is represented by the KRM model, and the Helmholtz– Kirchhoff integral was used to develop an accurate and elaborate method to estimate the backscattered sound from fish (Foote, 1985; Foote and Traynor, 1988). This approach was simplified by Clay (1991; 1992), who incorporated Stanton (1989) finite bent cylinder equation in fluid- or gas-filled cylinders to model fish backscatter, and has been validated for length and tilt (Jech et al., 1995; Horne et al., 2000).
The KRM backscatter model (Clay and Horne, 1994) combines the breathing mode and Kirchoff approximation to estimate the intensity of sound backscattered by an object based on the speed of sound and density of the fish body and swim bladder. The acoustic scattering length of the fish body (Lb) and swim bladder (Lsb) can be estimated as a fluid-filled half cylinder and gas-filled cylinder. It can be expressed simply by the following equations:
where fr is acoustic frequency, θ is the tilt angle, ρ is density, c is the speed of sound, and the subscripts b, w, and sb indicate fish body, water, and swim bladder, respectively. The scattering length from the whole body is calculated by summing the scattering amplitudes from the fish body and swim bladder.
Then, the TS of the fish can be calculated by the following equation:
A model of the backscatter of the acoustic characteristics is obtained from the KRM model, including the cylinder shape, density contrast (g), sound speed contrast (h), and frequency.
In total, 20 live fish ranging in fork length from 13.7 to 21.3 cm with an average length of 17.7 cm were collected from the waters around Jeju Island in July 2010. After capture, the fish were kept in a tank with running seawater for 5~6 hours to maintain their condition. To not to affect the condition of the swim bladder by application of, for example, pressure, is important so that it maintains its natural shape in water. The fish were frozen rapidly using a mixture of dry ice chunks and alcohol after removal from the tanks and stored in a freezer at -40°C prior to the imaging process.
The selected fish were radiographed dorsally and laterally using a digital soft X-ray imaging system and rare earth film applying a proportional scale. Radiograph images of the fish body and swim bladder are presented in Fig. 1. The radiographs were used to measure the swim bladder and body morphology for use in the backscatter model, and the dark-colored column is easily distinguished as a result of differential X-ray absorption by the air-filled swim bladder in comparison to the rest of the fish body (Clay and Horne, 1994).
The soft X-ray images were traced and then projected to a standard length using the vertebral column as a ruler between the ventral and lateral views. The lateral and dorsal images of the fish body and swim bladder were traced and then digitized at 3-mm intervals relative to the fish axis; the fins and tail were not included in the trace (Fig. 2). Trace lines were smoothed and rotated so that the sagittal axis of the fish body was horizontal, and the resulting dorsal and lateral images were elliptically interpolated into 3-mm-thick cylinders to give a three-dimensional representation of the fish body and swim bladder (Clay and Horne, 1994). These digitized data were then used to calculate the TS from the tilt angle and frequency using an acoustic scattering model. The series of morphometric descriptors (including swim bladder volume and area) were estimated using these three-dimensional fish representations. The ratio of the major to minor axes of the body and swim bladder (maximum length and width) was measured on the lateral traces as an index of elongation.
TS measurements were performed in a 5 m×5 m×3 m indoor tank filled with seawater (Fig. 3). A split-beam transducer (EK-60, Simrad, Horten, Norway) at 120 kHz was mounted vertically at the top of the tank, and an underwater camera was prepared for monitoring horizontal fish movement at a depth of ~3 m. Calibration was conducted prior to the TS measurements using a 23 mm copper calibration sphere (120 kHz; Foote et al., 1987).
Eighteen samples of bambooleaf wrasse with body lengths ranging from 13.0 to 19.5 cm were prepared for ex situ measurement of TS. To acclimatize with the surface pressure, the fish were kept in a tank for at least 2 hours prior to the TS measurements; only fish that were swimming normally after 2~12 hours of acclimation were measured. The fish were removed from the tank with a dip net and placed in a bucket filled with water to be hooked in a suspended frame. They were anesthetized with 98% ethyl 3-aminobenzoate methanesulfonate (MS- 222) to reduce struggling during transfer and while TS measurements were recorded. The frame was lowered slowly to ensure that the fish remained outside the near field of the transducers (2 m under both transducers).
By convention, positive tilt angles represent a fish with head-up orientation, and negative tilt angles indicate a fish with head-down orientation. In our study, however, the tilt angle of the fish was recorded for only ~0°. The system specifications and setup parameters are shown in Table 1.
Results
In total, 20 samples of bambooleaf wrasse were selected for sampling with the backscatter model. In some samples, however, extracting a normal swim bladder shape proved difficult due to the presence of unexpected foreign objects such as water. The X-ray photographs of two samples are shown in Fig. 4. In (a), the swim bladder is clearly defined, but (b) shows a case in which the swim bladder is not well-defined, and for this sample, extraction of a swim bladder of normal shape is difficult and leads to estimation errors.
During sample treatment, the fish sample was deep frozen within 30~50 s using the shock freezing method and kept at –50°C until processing (Farrant et al., 1977; Ona, 1990). The swim bladder, however, is easily damaged by external shocks and can be percolated by water.
Fig. 5 shows the acoustic scattering pattern at 120 kHz digitizing interval of the fish and swim bladder forms. The change in scattering pattern can be observed by setting the interval when slicing. The main lobe at ~10° tilt angle of the swim bladder has almost the same shape. The patterns at broadside, however, were significantly changed by varying the digitizing interval, and the higher frequency produced more complicated acoustic scattering patterns depending on the shape of the target. The differences were generated from the construction and destruction of the acoustic scattering of each cylinder during coherent summation, and a general scattering pattern was generated when applied to fish and swim bladder cylinder slices with intervals of 1.25 and 3.75 mm.
Acoustic scattering models have focused on the geometric aspects of the scattering model and assumed known material properties, i.e., the sound speed contrast (h) and density contrast (g). In many cases, the values of g and h are adjusted within reasonable limits to fit the directly measured acoustic data. The g and h values used in our backscatter model calculations were not the measured values for bambooleaf wrasse. Medwin (2005) found that fish flesh had g values from ranging from ~1.03 to 1.06 and h values from ~1.03 to 1.08.
Fig. 6 shows the acoustic scattering pattern at 38 kHz with varying g and h. The scattering pattern was not significantly affected by the g and h values of the fish flesh, and the differences were <1 dB because almost all of the acoustic scattering from fish is generated from the swim bladder. Insignificant differences in the g and h values through other organs or bones within the fish may influence the scattering intensities, and differences in g and h between species can result in TS differences of up to 10 dB (Horne and Jech, 2005). For most coastal fish of typical body length, however, the scattering results are not significantly affected by using the g and h values generally reported for fish.
Fig. 7 shows a comparison of the results for the TS at ~0° tilt angle estimated using the theoretical model and by measurement. The measured values were slightly different from the model values at a 0° tilt angle (bold line) because the fish tilt angle was not controlled precisely by the tethered method. The dotted lines indicate TS values from –5° to 5° of tilt angle. The figure shows that the tilt angles of the fish were controlled within ±4°, with close agreement between the theoretical and measured values.
To establish a general formula with the theoretical model, the TS patterns for 20 samples were estimated, and the swimming angle range was assumed to be (–5, 15) to calculate the average TS. The regression between TS and TL was expressed by the following equation (Fig. 8):
In addition to the sample conditions and digitizing errors, the r2 value was also influenced by the biological characteristics. The results, however, indicated that thetheoretical estimation was sufficiently accurate to establish the relationship between TS and BL.
Discussion
X-ray photography shows that the sample quality and treatment are important during sample preparation. Fish wriggle when removed from the water, and even when they are placed immediately in the dry ice chunks, a delay of several seconds takes place until they cease movement. To better maintain the natural shape for X-ray analysis, the dry ice chunks need to be small and smooth so as not to press on the fish body, and the fish samples should be handled carefully until freezing.
The regression between TS and BL to estimate fish abundance can generally be established for ex situ TS measurements for a wide range of fish sample size. In this study, however, the regression was established from the results of theoretical estimations of 20 frozen samples. Our results demonstrated that the theoretical model can be used to estimate the regression such as an ex situ measurement, and with appropriate sample control, has potential use for increasing the accuracy of acoustical fish abundance surveys.
This study was conducted to estimate and examine the TS of bambooleaf wrasse (P. japonicus) using the KRM model and to examine the TS values produced by the model by comparison with ex situ measurements. Twenty frozen fish samples were prepared for X-ray photography to determine the dimensions of the fish body and swim bladder shape, and the TS patterns at a digitizing interval of 120 kHz. TS values at ~0° fish tilt angle for 18 live fish samples were also measured using a tethered method to verify the values estimated by the model.
The swim bladder was easily covered and damaged by unexpected objects during preparation of the frozen sample, and problems with the determination of the swim bladder morphology using X-ray photography and the digitizing interval of the fish body and swim bladder shape significantly affected the calculation of TS; however, the ranges in the density and speed of sound ratio for general fish flesh are relatively small (<1 dB) and do not affect the results.
Close agreement was observed between the measured and theoretical values, and the general formula using the theoretical method could be given as = 20logTL (cm) – 71.6 (r2=0.69). Our results demonstrate the potential of the theoretical model for acoustical fish abundance surveys using appropriate sample control to increase the accuracy.
