© 2004 by ICES/CIEM International Council for the Exploration of the Sea/Conseil International pour l'Exploration de la Mer
The biological validation of ADCP acoustic backscatter through direct comparison with net samples and model predictions based on acoustic-scattering models
Southampton Oceanography Centre European Way, Southampton SO14 3ZH, UK
*Correspondence to S. Fielding: tel: +44 2380 596016; fax: +44 2380 596247. e-mail: sof{at}soc.soton.ac.uk.
Mean volume-backscattering strength (MVBS) data collected using a 153-kHz, narrowband Acoustic Doppler Current Profiler (ADCP) were compared with zooplankton abundance and biovolume data collected using a Longhurst–Hardy Plankton Recorder (LHPR). A direct comparison showed that there was a linear relationship between MVBS and log-transformed zooplankton dry weight. This linear relationship, determined from a mixed zooplankton-species population, was then compared with that reported in previous work from a region dominated by a single species of copepod and found to be significantly different. The scatter around the linear relationships determined between MVBS and log-transformed dry weights in regions of complex, mixed zooplankton populations results in our inability to distinguish different relationships that could be expected from different populations in varying oceanic regions. It is suggested that, without further manipulation of the data, ADCP MVBS cannot be used to determine quantitative estimates of zooplankton abundance and biomass in mixed populations.
"Observed MVBS" was compared with model-predicted backscattering, calculated using acoustic models and abundance and size measurements of zooplankton from net samples. The results show that at high backscattering intensities (>–80 dB) the observed MVBS from an ADCP was generally consistent with the model predictions. Abundance, biovolume, and model-predicted backscattering contributions of six "significant acoustic-scattering" groups (amphipods, chaetognaths, copepods, euphausiids, fish, and pteropods) are shown to vary disproportionately. In particular, a rare and small but strong acoustic scatterer such as a pteropod can contribute as little as 0.1% to the total sample abundance and 0.1% to the biovolume but represent 69.5% of the total model-predicted backscattering.
Model, instrumental, and methodological artefacts are identified as potential sources of inconsistencies between the observed and model-predicted backscattering. These include the effect of the orientation of zooplankton, inadequate knowledge of model parameters such as the sound-speed and density contrasts, the mismatch between sampling volumes of the net and acoustic instrument, and net avoidance by the more mobile scatterers.
Keywords: acoustic models, Arabian Sea, bioacoustics, Doppler sonar, MVBS, Northwest Indian Ocean, sound scattering, zooplankton
Received 7 May 2003; accepted 15 October 2003.
 |
1 Introduction |
|---|
 Top 1 Introduction 2 Methodology3 Results4 Discussion5 ConclusionsReferences |
|---|
Biological distributions in the open ocean are patchy, resulting from either behavioural or environmental controls or a combination of both factors. To understand the resultant patterns and their causal processes, it is necessary to sample biological data at high resolution and on similar spatial and temporal scales as other environmental data. Acoustic remote sensing is one method of obtaining biological data that fulfils these criteria. The use of commercially available systems, such as an Acoustic Doppler Current Profiler (ADCP) and a SIMRAD EK500 multi-frequency echosounder, within multidisciplinary and biological surveys is now routine (Roe et al., 1996; Brierley et al., 1998; Herring et al., 1998; Fielding et al., 2001; Wade and Heywood, 2001) and purpose-built systems are also frequently used (Holliday and Pieper, 1995; Wiebe et al., 1996; Crisp and Harris, 2000). However, whilst the use of acoustic data to qualitatively describe zooplankton distributions and behaviour is common, a quantitative relationship between acoustic backscattering and zooplankton abundance, size, and taxonomy in a complex, mixed zooplankton-community environment remains an elusive "bioacoustic" goal.
Acoustic scattering from zooplankton is related to the acoustic frequency used and the size, orientation (Chu et al., 1993; McGehee et al., 1998), morphology, and physiology (Stanton et al., 1994) of the target. Thus, the average backscattering cross-section of one animal may vary dramatically from another of similar or greater size and biomass (Stanton et al., 1994). In a mixed zooplankton-species community (as typically found in the open ocean), it is therefore very difficult to directly relate mean volume-backscattering strength to zooplankton biomass.
Acoustic-scattering models have been developed to describe and predict the acoustic backscattering from single zooplankton of different taxonomic groups. Empirical measurements of zooplankton in enclosures (Greene et al., 1989; Foote, 1990) and individual, tethered animals under controlled laboratory conditions (Greene et al., 1991; Stanton et al., 1994) have furthered their development and refinement (Stanton et al., 1998a, b; Chu and Stanton, 1998). These models require validation in field studies by combining them with taxonomic, abundance, and size data from net samples in order to predict acoustic backscatter, and then relating the result with the observed MVBS collected concurrently (i.e. solving the acoustic "forward problem", Holliday and Pieper, 1995). Interpreting acoustic-backscattering data from the field is recognized to be complex (Stanton et al., 1996) and these scattering models have been applied in zooplankton field studies only recently (Wiebe et al., 1996; Greene et al., 1998). Both these papers compared predicted backscatter, estimated from length and abundance of different zooplankton classes captured using a MOCNESS and acoustic-scattering models, to high-frequency (420 kHz) observed data and found that they were highly correlated. Their comparisons used data from Georges Bank and the Gulf of Maine, respectively. This study presents data from the Indian Ocean (Arabian Sea), where the distribution of zooplankton, and potentially their morphology and physiology, is dominated by an intense oxygen-minimum layer (Childress and Thuesen, 1992; Herring et al., 1998). It compares the predicted acoustic backscatter, calculated from Longhurst–Hardy Plankton Recorder (LHPR) net samples and acoustic-scattering models, with the observed MVBS data from a hull-mounted ADCP.
The ADCP can provide qualitative information about zooplankton (Flagg and Smith, 1989a, b; Plueddemann and Pinkel, 1989) and has been used to describe diel migrations (Smith et al., 1989; Heywood, 1996) and the distribution of zooplankton at mesoscale features (Ashijan et al., 1994; Roe et al., 1996). There is an ongoing debate about the use of narrowband ADCPs as quantitative echosounders. This is primarily for two reasons. First, the instrument does not directly measure the received-signal amplitude, rather it measures a proxy – the value of the automatic gain control (AGC) needed to provide a constant signal level to the signal processor (Flagg and Smith, 1989a). Second, the inclination of the four beams to the vertical (usually 30°) makes in situ calibration with standard targets (Foote, 1983) very difficult. Griffiths and Diaz (1996) compared underway measurements by a 153-kHz, shipboard ADCP with those by a calibrated 200-kHz, EK500 sounder over a range of MVBS from –88 to –68 dB. The data from the two instruments were highly correlated (r2=0.989), but with an offset that amounted to 7 dB; with the ADCP at –75 dB, the EK500 indicated –82 dB. After applying the slope and offset corrections to the ADCP, the standard error of the difference between the beam-average ADCP data and EK500 was 0.7 dB.
The measurement by the AGC circuit is that of signal plus noise and, as the signal level approaches the noise level, it becomes an increasingly poorer analogue of the true signal. In this study, data with a low signal-to-noise ratio were omitted by following the practice of ignoring data from depth bins with suspect current velocities (Roe et al., 1996). This also partially corrects for the non-linearity of the ADCP, which is additionally corrected for by only accepting raw AGC values at least 15 counts (6.3 dB) above the noise level (identified from the deepest bin-depth). Scattering was never as high in this study as to take the instrument into the known region of non-linearity at AGC counts beyond 200 (RDI, 1990). In summary, errors from the use of AGC as a proxy for signal amplitude are likely to result in offset errors from the true MVBS. Although of concern in an absolute sense, these offset errors do not detract from the correlation analysis performed in this paper, especially given the uncertainty in model parameters, such as the reflection coefficient, that affect the absolute values of model-derived MVBS.
These issues with the ADCP are in addition to the more general question of whether single-frequency acoustics can provide meaningful size and abundance data on zooplankton populations. However, ADCP mean volume-backscattering strength has been compared directly with net-sampled zooplankton biovolume (Flagg and Smith, 1989a, b; Heywood et al., 1991; Batchelder et al., 1995) and been shown to be related. More recently, acoustic-model predictions derived from estimates of the size composition of dominant scatterers from net samples taken in the Gulf of Mexico were compared with MVBS (Ressler, 2002). A substantial difference (
19 dB) between predicted and observed scattering was found, keeping in mind that 3 dB is a doubling of the signal.
|
2 Methodology |
|---|
|
Top1 Introduction2 Methodology3 Results4 Discussion5 ConclusionsReferences |
|---|
As part of the multidisciplinary Discovery Cruise 209, August 1994, in the Arabian Sea (Herring et al., 1998), concurrent net and acoustic sampling was undertaken to map the distribution of zooplankton in relation to the oxygen-minimum zone and to examine the relationship between acoustic backscatter and zooplankton.
2.1 Net data collection
Zooplankton samples were collected using a Longhurst–Hardy Plankton Recorder (LHPR) (Longhurst et al., 1966; Williams et al., 1983). The LHPR, fitted with a 280-µm mesh, was towed at a constant speed of 4 knots (2 m s–1) in a V-shaped profile to a depth of 250 m, providing a series of sequential 2-min samples. Data presented in this paper are from samples within the downcast of LHPR Station 12670#4, the upcast being complicated by clogging of the net. Each 2-min sample was analysed in terms of taxonomic composition, total abundance, and total biovolume and the abundance, biovolume (where possible, as a result of the small sample size inherently collected by the LHPR), and size of six "significant acoustic-scattering" groups (amphipods, chaetognaths, copepods, euphausiids and decapods, fish, and pteropods) identified by Stanton et al. (1994).
2.2 Acoustic data collection
An ADCP was used to map the vertical patterns of acoustic backscatter along the ship's track. The hull-mounted, 153-kHz, RD Instruments, narrowband ADCP recorded backscattered signal strength from each of four acoustic beams. The vertical resolution was 128x4-m bin-depths, giving a range between 10 m and >500 m: 10 m being the minimum distance allowed by the instrument given its hull depth, transmit pulse, and blank after transmit. Acoustic backscatter is measured by the ADCP as a function of the automatic gain control (AGC). These data, from all four of the ADCP beams, were averaged into 2-min intervals. Mean volume-backscattering strength (MVBS, dB) was calculated for each bin-depth following the manufacturer's equation (RDI, 1990). Although the ADCP could theoretically measure MVBS to depths greater than 500 m, the signal-to-noise ratio was low at depth and consequently the data were rejected in the processing.
Since environmental data, temperature and salinity in this case, were not collected concurrently when towing the net, a constant sound-absorption coefficient (
) was used, in contrast to modern processing which allows to vary with measured variations in the temperature and salinity (as used by Roe et al., 1996; Fielding et al., 2001; Wade and Heywood, 2001). A mean for the whole of the ensonified water column was calculated from a CTD cast prior to the LHPR tow.
2.3 Data comparison
The observed MVBS/zooplankton relationship was investigated in two ways: (i) by comparing the observed MVBS to the zooplankton biovolume converted to the log-transformed dry weight and (ii) by comparing the observed MVBS with the model-predicted backscattering, calculated using acoustic-scattering models and measurements of size and abundance of several zooplankton groups.
Each variable (biovolume and model-predicted backscattering) was compared with the observed MVBS estimated by echo integration in the vicinity of the net-sampled volume. Hence, the first step was to identify, in time and space, acoustic data concurrent with the net samples.
All the acoustic data were averaged into 2-min intervals so both acoustic and net data were available on the same sampling interval. The configuration of the equipment, the acoustic data being collected from immediately below the ship and the LHPR-net samples from behind the ship, resulted in the net system always trailing some variable distance (L) behind the acoustic transducers. Therefore, the LHPR actually passed through the depth horizon (d) at a time (td) that differed by a variable period (
t) after the ADCP had sampled the same depth.
To match the data from the net with the acoustic information, the corrected time (tc) was calculated following the method presented by Zhou et al. (1994), where
|
| (1) |
t, was calculated from:
|
| (2) |
LHPR tows were reconstructed through the ADCP data using the variable time correction given by Equations (1) and (2) above. Although the sampling interval of the acoustic data could be matched to that of the LHPR, the depth interval sampled by the LHPR was more variable. Therefore, the acoustic data were integrated, in linear form, over the vertical range that the LHPR travelled during any 2-min sampling period, where the LHPR travelled further, in depth, than that of one acoustic bin.
2.4 Direct comparison
The biovolume of each LHPR sample was quantified using the normalized displacement volume (Beers, 1976; Postel et al., 2000). To compare the results with those of Flagg and Smith (1989b), henceforth denoted FS2, Heywood et al. (1991), henceforth denoted HSB, and Batchelder et al. (1995), henceforth denoted BVVS, biovolume was converted to dry weight (mg m–3) using the Wiebe et al. (1975) conversion tables where:
|
| (3) |
|
| (4) |
The dry-weight data were divided by 4
and log-transformed following the theory suggested by FS2: namely, given that target strength, and therefore the backscattered signal, is equal to log(
s/4), where s is the acoustic cross-section, and dry weight is approximately proportional to cross-sectional area, which should be proportional to acoustic cross-section, it is logical to plot log(DW/4) against MVBS, the log-transformed variable of backscattering cross-section.
Student's t-test was used to examine whether the slope of the regression-line fit to the log-transformed data (DW/MVBS) varied between this and previous studies. A predictive regression, as used by FS2 and HSB, was used to calculate DW/MVBS. However, it should be noted that a functional regression is more appropriate (Ricker, 1973; e.g. Wiebe, 1988) and both methods are presented here. A functional regression line was used to examine the relationship between the dry weight and the observed MVBS, and the total model-predicted backscatter and the observed MVBS (see below) because these are all independent measurements of the same variable, zooplankton biomass, and are therefore all subject to error in the measurement. We have assumed that the error in the measurements is of similar magnitude.
2.5 Model predictions
Model-predicted backscattering was calculated by solving the forward problem (Holliday and Pieper, 1995) and was determined for each 2-min LHPR sample by combining the taxonomic size and abundance data of zooplankton with the appropriate scattering models. In this case, the scattering models were used to calculate model-predicted backscattering from six groups: large (>2 mm) and small (<2 mm) copepods, chaetognaths, amphipods, fish, pteropods, and euphausiids including decapods because of their similarity in shape. The total model-predicted backscattering (TMPB) for each 2-min LHPR sample was assumed to be a linear combination of the backscattering from all the individual sound scatterers in the ensonified volume (following Clay and Medwin, 1977). Thus, the forward problem can be described by the following equation (Greene et al., 1998):
|
| (5) |
bsij
is a representative backscattering cross-section for size class j of taxon i, and s and t are the numbers of relevant size classes and taxa, respectively. The acoustic-scattering models used in computing the forward problem were developed by Stanton et al. (1994). The fluid-filled, bent-cylinder model was used for the amphipods, chaetognaths, copepods, euphausiids, and fish, and the dense fluid-filled, sphere model was used for the pteropods. All representative backscattering coefficients derived from the models were calculated for the 153-kHz frequency.
The output for each sample is a value of the predicted acoustic backscatter for each group, that is, the sum of all contributions from that acoustic-scattering group, and a total model-predicted backscattering. Through these data, the contribution to the TMPB of each zooplankton group can be defined. TMPB was compared with the observed MVBS using a functional regression, the slope of the regression line (v) being compared to the ideal of one-to-one using Student's t-test (H0, v=1, p>0.05).
|
3 Results |
|---|
|
Top1 Introduction2 Methodology3 Results4 Discussion5 ConclusionsReferences |
|---|
LHPR Station 12670#4 was a daytime tow (09:07–11:59 LT) taken in the vicinity of the Arabesque Reference Site (ARS, Herring et al., 1998), 19°N 59°E. During the downcast of the tow, the LHPR travelled a distance of approximately 14 km. A CTD cast, immediately prior to the LHPR, indicated that, whilst surface waters were well oxygenated, there was a distinct oxygen-minimum layer existing between 70- and 300-m depth (Figure 1a): in fact, the oxygen-minimum layer extended to depths >1000 m (Herring et al., 1998). ADCP MVBS data during the period of the downcast of LHPR Station 12670#4 showed the presence of two main sound-scattering layers (Figure 1b). Sound-scattering layer one (SSL1) occurred between the surface and 70-m depth with MVBS from –64 to –76 dB, with the greatest intensities occurring near the surface above 50 m. The second sound-scattering layer (SSL2) occurred below 250 m and with MVBS of approximately –70 dB. Between these two scattering layers was a region of low MVBS, ranging between –94 and –80 dB, with the lowest values recorded between 100 and 175 m. In addition, for the first hour a "weak" scattering layer (–80 to –72 dB) was present between 180 and 225 m.
|
A summary of the acoustic volume-backscattering statistics and the standardized biovolume and abundance at the mean depth for each sample is shown in Table 1. Where the LHPR sampled over the distance of one bin-depth, the mean MVBS was calculated. The final column indicates whether the sample was collected from water with high- or low-oxygen concentration as identified from the CTD prior to the LHPR. Samples collected from the surface to 50 m were in oxic conditions (45–200 µmol l–1), those from 50 to 100 m were in dysoxic conditions (4.4–45 µmol l–1), and those below 100 m were in anoxic conditions (<4.4 µmol l–1) – following Bernhard and Sen Gupta (1999).
|
LHPR-sampled total zooplankton biovolume and abundance maxima occurred in the surface 70 m, above the oxycline, concurrent with the highest recorded values of MVBS (Figure 2a, b). Below the oxycline, biovolume and abundance were low compared with surface values, with no discernible increase below 150 m where MVBS was beginning to increase with increasing depth. The distribution of the biovolume and abundance of the significant acoustic-scattering groups (Figure 2c) was similar to the total values, except for the surface samples. Near the surface, the maxima in both biovolume and abundance for the significant acoustic-scattering groups coincided with a minimum in total biovolume and a maximum in total abundance. Since copepods were the dominant taxonomic group, and are classified as a significant acoustic scatterer, the values of total abundance and significant acoustic-group abundance are similar. The significant-group biovolume is, however, a fraction of the total biovolume within the oxygenated surface waters, since the dominant fraction of the surface total biovolume results from medusae and a residue of dead organic matter and phytoplankton.
|
3.1 Direct comparison
Despite there being high biovolume, and thus high log(DW/4
), in the surface samples concurrent with high MVBS, the mismatch between high MVBS and low log(DW/4) in the deeper samples resulted in a low, although significant, correlation between MVBS and LHPR log(DW4) (Figure 3a). The predictive linear-regression line fitted to the data yielded the relationship:
|
| (6) |
DW/MVBS. FS2 comment that the intercept (2.992) will be, at the least, instrument-specific.
|
In addition, the more appropriate functional regression was applied to the log-transformed, dry-weight data that had been calculated using the more recent conversion tables given in Wiebe (1988), yielding the relationship:
|
| (7) |
These calculations were also applied to data referring only to the significant acoustic-scattering zooplankton groups (DWag) identified by Stanton et al. (1994), where:
|
| (8) |
|
| (9) |
) and MVBS was improved compared with the total sample dry weight (Figure 3b), and the resulting regression lines had a correlation coefficient of 0.56 (n=27, r2=0.32, p<0.05).
DW/MVBS was compared with previous results (0.115, FS2; 0.056, HSB; and 0.055, BVVS) and 0.1 (after FS2's theory of log(DW/4)
0.1xMVBS) using Student's t-test. The null hypothesis (H0) was that DW/MVBS, in this study, was not significantly different from the slope of previous regression lines (i.e. 0.115, 0.056, and 0.055 and the hypothesized 0.1). The null hypothesis was accepted when DW/MVBS was compared with HSB and BVVS, and rejected when compared with FS2 (Table 2).
|
3.2 Model-predicted backscattering
The predicted backscattering of six "significant acoustic-scattering" zooplankton groups, identified in the taxon-specific model equations given by Stanton et al. (1994), were calculated and compared with the observed MVBS in the vicinity of the LHPR-net sample (Figure 4a–f). The predicted backscattering from copepods, amphipods, chaetognaths, and fish was less than the observed MVBS, with the lowest values of predicted backscattering originating from the chaetognath and amphipod groups (
–135 dB). The predicted backscattering from the pteropod group was close to the observed MVBS. In fact, the slope of the functional regression equation fitted to solely the prediction from the pteropod contribution and observed MVBS is not significantly different from the expected relationship of one-to-one from the whole sample (v=1.13, r2=0.84, t-test, n=5, p<0.05) and is very positively correlated (r=0.92). Euphausiids were the only group where the model predictions were, in some cases, greater than the observed values of MVBS.
|
The percentage contribution of each group to the total model-predicted backscattering (TMPB) was examined (Figure 5). In surface waters, above 70 m, pteropods had the largest model-predicted backscattering, comprising up to 69.5%. Amphipods contributed the least, comprising only 0.3% of the TMPB. Below 70 m, euphausiids had the greatest model-predicted backscattering, dominating the TMPB with levels up to 99.5%. Copepods were the second-largest contributors, except below 230 m where they were replaced by fish. The contribution of each group to the TMPB was not necessarily related to its abundance or biovolume. Three samples, from 17.5-, 159.5-, and 242-m depth, emphasize the discrepancies between percentage contribution of each group to biovolume, abundance, and total model-predicted backscattering (Figure 6). Whilst pteropods were numerically (0.1%) and volumetrically (2.2%) insignificant at 17.5 m, they dominate the model-predicted acoustic signal (69.5%). In deeper samples, where pteropods were not present, the abundance was dominated by copepods and the biovolume and TMPB were dominated by euphausiids.
|
|
The TMPB for each sample was compared to the MVBS observed in the vicinity of the net-sampled volume (Figure 7). The functional regression-line fit to the log-transformed data was significantly different from the expected slope of 1 (v= 0.67, r2=0.29, t-test, n=25, p<0.05). Coding the data according to the dominant acoustic-scattering group indicates that when pteropods were present in the samples, the TMPB was comparable with the observed values. In deeper samples, where the euphausiid group was the largest contributor to TMPB, it typically exceeded the observed value. This implies that, with the assumptions we have made here, the euphausiid model fails to reproduce acoustic backscattering consistent with our observations. A possible cause of inaccuracy may be an orientation effect, which has both a biological and an instrumental origin (Griffiths et al., 2002). Acoustic backscattering from a euphausiid-shaped animal is directional. Its dependence on angle has been studied by Kristensen and Dalen (1986), Stanton et al. (1993a), and Macaulay (1994) and more recently in the laboratory by McGehee et al. (1998), when variations of up to 25 dB were found. The models we used assume broadside incidence of the acoustic signal to the zooplankton. However, the orientation of individual euphausiids in the water column is known only poorly and may vary during a diurnal migration period (biological origin). In addition, the ADCP beams are inclined 30° from the vertical (instrumental origin).
|
For angular variations away from broadside of 30°, Kristensen and Dalen (1986) observed and Macaulay (1994) modelled a 5-dB reduction in the target strength of a single euphausiid. Therefore, a –5 dB correction, not inconsistent with the angular variation of backscatter at 30°, was applied to the euphausiid and decapod contribution. The adjustment of the model-predicted data, whilst forcing agreement of the slope of the functional regression line with the expected slope of 1 (v=0.87, r2=0.44, t-test, n=27, p<0.05), did not remove the significant amount of scatter and hence the fit of the data to the regression line was not substantially improved (Figure 8).
|
Coding the data according to the dominant scattering group also identified that the scattered data points were those that had the greatest contribution from the euphausiid group to the TMPB. Further investigation into the source of the zooplankton samples, for example, the oxygen concentration of the water the sample was drawn from, shows that the model and observed acoustic data are comparable for oxic and dysoxic waters and not for samples drawn from anoxic water (Figure 8). Since variations in morphology and physiology have been suggested for some zooplankton living within the oxygen-minimum layer, it is possible that the scatter is related to inadequate definition of the reflection coefficient within the acoustic-scattering model. This follows the premise that the euphausiids and decapods living within the oxygen-minimum layer could be expected to have a softer carapace than their other ocean counterparts (Herring et al., 1998). It may well be that their reflection coefficient is less than 0.058, the value used in this study as determined empirically by Stanton et al. (1994), hence resulting in an over-prediction of the Arabian Sea-euphausiid backscattering as observed here.
However, the oxygen content of the water column varies with depth, and the samples with the largest difference between the model and the observed backscattering come from depths between 100 and 200 m, which coincides with the layer of low observed MVBS. An alternative approach is to analyse only the acoustic data from either "evenly" distributed, or frequent targets. This follows the conclusion of Brierley et al. (1998) that ADCP acoustic backscatter compared favourably with a calibrated echosounder only under these conditions. To test if this conclusion was valid for our study, all samples where the observed MVBS was greater than –80 dB were analysed and everything below –80 dB was discarded. This dynamic range was comparable with the data sets analysed by Wiebe et al. (1996) and Greene et al. (1998). The resulting slope of the functional regression line (Figure 9) fit to the data was not significantly different from the expected slope of 1 (v=1.00, r2=0.62, t-test, n=12, p<0.05).
|
Rather than this difference being an instrumental inability to reliably detect infrequent targets, it may be related to the difference in probability of each instrument sampling the same animals. The ADCP has four conical beams (each with a beam width of 3° and a pulse length of 4 m) and the volume of water ensonified increases with depth, whereas the LHPR-sample volume remains constant. At the surface, the volumes sampled are within an order of magnitude, the ADCP measuring from
200 m3 of water, based on a sampling volume of 5 m3 per ping from a 4-m bin-depth at a depth of 10 m and a 2-min ensemble of 40 pings, and the LHPR from 40 m3. At 200 m, within the suboxic zone, the ADCP MVBS measurements are made from 56 300 m3 of water, based on a sampling volume of 1408 m3 per ping from a 2-min ensemble of 40 pings, nearly three orders of magnitude greater than the LHPR. This divergence in the volume of water sampled affects the probability of each instrument encountering an infrequently distributed zooplankter. For example, if we assume that the probability of encountering one zooplankter per cubic metre within the water column is 0.025, equivalent to catching one on average in a 2-min LHPR sample, we can show that, because of its larger sampling volume, the ADCP is more likely to sample more than one zooplankter than is the LHPR (Figure 10). This discrepancy in sampling volumes may cause a degree of scatter between the observed MVBS and the model-predicted backscattering arising from the probability of encounter. However, the assumption must be that it should bias the observed MVBS to be greater than the model predictions because of the infrequent catching of either rare or scarce, or a combination of both, zooplankton in the net samples compared to the acoustic system. This is contrary to the results presented here if we retain the values of g, h, and R that we have also assumed.
|
|
4 Discussion |
|---|
|
Top1 Introduction2 Methodology3 Results4 Discussion5 ConclusionsReferences |
|---|
Our results show that log-transformed zooplankton dry weight (DW) is linearly related to the observed ADCP MVBS. This is consistent with previous work and means that MVBS can be used as a general proxy for the distribution of zooplankton (after FS2, HSB).
This, and other work, involves the use of "relative" acoustic backscatter and for statistical comparison with previous work, zooplankton volume (DV) was converted to dry weight, divided by 4 and log-transformed, and then plotted against ADCP MVBS, with various controls on the method of conversion to conform with previous analyses. The resulting slope of the regression line (DW/MVBS) expresses the rate of change in biomass with a change in the MVBS and the intercept is instrument- and experiment-specific. FS2 expected this to provide a linear dependence with a proportionality constant of 1/10 (i.e. DW/MVBS=0.1), where the larger the slope the less sensitive the ADCP would be to changes in zooplankton concentration.
DW/MVBS in this study was 0.042 (2 significant figures) (n=27, r2=0.22, p<0.05), lower than in previous studies of direct comparisons between ADCP MVBS and net-sampled zooplankton. According to FS2, this implies that the ADCP used in this study was the most sensitive or, alternatively that the zooplankton analysed were larger. The latter statement is certainly valid, because the mesh size of the net used in this study (280 µm) was larger than that used by FS2 and HSB. Therefore, the mean length of the zooplankton caught varied between 1 and 15 mm, larger than the zooplankton sampled by HSB (where 50% of the population <1 mm).
The data presented here exhibit a poorer correlation between log(DW/4) and ADCP MVBS than found by previous authors (Flagg and Smith (1989a, b), HSB, BVVS). However, it should be noted that the study which had the greatest regression and correlation coefficient (by Flagg and Smith (1989a, b), r2=0.96) was carried out in an area dominated by a single species of copepod, Calanus finmarchicus (Smith and Lane, 1988), i.e. a simplified "acoustic situation". Batchelder et al. attributed their own and Heywood et al.'s "poorer" correlations to using depth-integrated net hauls compared with vertically stratified net hauls.
An alternative reason for the variations in DW/MVBS was proposed by HSB. They showed that the slope of the regression line (DW/MVBS), and the fit of the data to the line (r2), varied if the value of the sound-absorption coefficient (), used in calculating acoustic backscatter, was changed. The value of , which is dependent on the temperature and salinity structure of the water column, used to calculate MVBS in the present study was 0.44 dB m–1 compared with 0.46 and 0.40 used by HSB. The slope of their regression line varied from 0.056 to 0.052 with the change in , which is not sufficient to account for the variation of 0.0144 (25%) between HSB's study and the one reported here. HSB also considered that the error caused by changes in would be negligible compared with the errors in the displacement method for measuring biovolume and patchiness in the zooplankton.
However, should we actually be trying to relate zooplankton biovolume or dry weight with MVBS or acoustic backscatter from any single-frequency echosounder? DW/MVBS from this study was found to be significantly different from the monospecific example of FS2 and not significantly different from studies undertaken in more complex, mixed zooplankton environments (e.g. HSB, BVVS). These variations may be attributed to the large degree of scatter exhibited in studies undertaken in regions not dominated by a single zooplankton species. This can be explained by the, now known, variation in the relationship between cross-sectional area and acoustic cross-section between different types of zooplankton, which has resulted in the classification of strong and weak acoustic scatterers (Stanton et al., 1994, 1996; Griffiths et al., 2002). FS2's theory for using log(DW/4) was based on the assumption that dry weight is proportional to cross-sectional area that is, in turn, proportional to the acoustic cross-section, the conversion of acoustic cross-section to MVBS or Sv providing the 0.1 proportionality (based on Sv = lon (N
/ 4 
V), where N is the number of scatterers, V is the sampling volume, and is the acoustic cross-section). FS2 presented data from an environment dominated by a single species of zooplankton where the above assumptions may have been valid, whereas in a complex, mixed zooplankton community these assumptions cannot be met.
Therefore, a comparison of the relationship between MVBS and zooplankton biomass found in the present study to previous studies is more complex than just comparing the value of DW/MVBS. Each study occurred in a different area, the western North Atlantic, Gulf Stream (FS2), the southern Indian Ocean (HSB), the eastern North Atlantic (BVVS), and the northern Indian Ocean (this study), with different zooplankton populations and community structures (Van Der Spoel and Heyman, 1983). In addition, a different zooplankton net with different mesh sizes was used in each study. Flagg and Smith (1989a, b) used a MOCNESS with a 149-µm mesh, HSB used a WP2 with a 142-µm mesh, BVVS used a ring-net with a 153-µm mesh, and the present study used an LHPR with a 280-µm mesh. Since MVBS is a complex function of zooplankton size, similar morphology and physiology results should not be expected from different nets, as each study will be catching a different zooplankton population with a different size, morphology, and physiology. As zooplankton patchiness can occur at sub-kilometre spatial scales (Haury et al., 1978; Folt and Burns, 1999), the validity of using the predicted mean biomass from the linear relationships between cruises as well as within one cruise becomes doubtful.
The use of models to describe acoustic scattering has arisen because the echo return from a target is dependent not only on its size but also on its composition (Stanton et al., 1993b). The results presented here show that abundance, biovolume, and model-predicted backscattering contributions of six zooplankton groups vary disproportionately. In particular, a rare and small but strong acoustic scatterer such as a pteropod contributed as little as 0.1% to the total sample abundance and 0.1% to the biovolume but represented 69.5% of the TMPB in one of our samples. These results emphasize both the need for accurate acoustic-scattering models and the necessity for caution when describing zooplankton abundance and biovolume from acoustic backscatter without concurrent net data to indicate the types of scatterers.
They also give a convincing demonstration that at intensities greater than –80 dB, observed ADCP MVBS data were consistent with the forward-problem predictions. The relationship between observed ADCP MVBS and TMPB compared favourably with that found when using a well-calibrated biological echosounder (Wiebe et al., 1996; Greene et al., 1998), and as in the study by Wiebe et al., the contribution of pteropods was typically larger than that of the other taxa. However, the results of this study also provide examples of situations in which inconsistencies between the observed and model-predicted backscattering indicate potential model, instrumental, and methodological problems.
4.1 Model
When the euphausiid group was the dominant scattering group, the TMPB typically exceeded the observed values. Two potential sources of error affecting the model predictions for this group are first, an orientation effect and second, inadequate knowledge of the sound-speed and density contrasts of the zooplankton. The orientation effect was potentially allowed for in the results, although it should be noted that if the animals in the ocean have a preferred orientation, the echo levels could easily differ by several decibels from the level averaged over all orientations (Wiebe et al., 1996). The second error refers to the calculation of the reflection coefficient used within the model. Euphausiids and decapods, which exist within the oxygen-minimum layer of the Indian Ocean, have softer carapaces than their other ocean-environment counterparts (Herring et al., 1998). This would affect their density and hence their reflection coefficient, resulting in greater predicted backscattering than observed values. This agrees with the observation that the models were over-predicting backscattering from samples collected at low-oxygen concentrations. The value of the reflection coefficient used in Stanton et al. (1993b) and Greene et al. (1998) was R=0.058. Using an alternative value of R (R=0.031, given by Foote (1990)) would reduce the mean bias of the euphausiid-dominated samples by 4.6 dB. Inadequate or approximate measurements of the speed of sound and density contrasts, and the lack of knowledge regarding the orientation of euphausiids and decapods within the open ocean results in our inability to distinguish between these two errors. This identifies the need to consider the environment, and its influence on the zooplankton, as well as the importance of correct measurements of the speed of sound and density contrast parameters within the acoustic-scattering models.
4.2 Instrumental
Brierley et al. (1998) commented on the unreliability of acoustic measurements made with an ADCP in regions of low or irregularly distributed targets. The results presented in this study show that at high (>–80 dB) MVBS intensities, which are related to many or evenly distributed targets, model-predicted backscattering and observed MVBS are comparable, the ADCP behaving as well as a "biological" echosounder. At low MVBS intensities, the model predictions become increasingly poor. Brierley et al. (1998) cited the orientation of the ADCP beams as a possible source of error. However, the models used in this study assume the random orientation of all targets within the ensonified volume. Additional factors of inadequate resolution of model parameters as well as the discrepancy between sampling volumes result in several potential sources of poor correlation, of which the latter two are applicable to any echosounder.
4.3 Methodological
Comparing model-predicted backscattering with observed MVBS assumes that the two instruments, i.e. the ADCP and the LHPR zooplankton net, are sampling the same zooplankton population. The first basic condition to be met is that both instruments are sampling similar size ranges. In this case, the size of the zooplankton caught by the LHPR (>1 mm, Nichols and Thompson, 1991) falls into the category of ka>0.3 at the 153-kHz acoustic frequency used, where k=2/
, is the wavelength and a is the equivalent spherical radius, and is therefore relatively well matched. The second condition should be that the two instruments have comparable sampling volumes. There is clearly a large discrepancy between the sampling volumes of the ADCP and the LHPR at depth. Simple calculations of encounter-rate probability identify this discrepancy as a source of variation between net and acoustic measurements of zooplankton, especially when a rare but strong acoustic target such as a pteropod is likely to be sampled in greater numbers by the acoustic instrument than with the net (Figure 10). This source of error is applicable and must be accounted for in any comparison between modern zooplankton-surveying techniques.
An additional problem resulting from spreading is singular to the ADCP. An ADCP has four beams, each angled 30° from vertical. This results in the fore and aft beams being angled 60° from each other, such that at 10-m depth the centre of the beams is separated by 17 m and at 300-m depth by 520 m. If a ship travels at 2 m s–1 (4 knots), a sample taken at 10-m depth would contain information on zooplankton collected during a 2-min sample over a distance of 240 m, whilst a sample taken at 300-m depth would cover a distance of 750 m. Hence, zooplankton patches at depth could be smeared by the ADCP. A combination of these problems and the knowledge that when the LHPR is at 250-m depth it could be up to 750 m behind the ship, allowed for when identifying concurrent MVBS, creates a large margin for error. This error does not include any discrepancy between encounter rate or "catch efficiency" of each instrument. An ADCP, or any other hull-mounted or towed echosounder, is a non-obtrusive sampling system, whereas a towed net is subject to avoidance reactions by the zooplankton (e.g. Orr, 1981) that may additionally bias differences between remotely acoustically-assessed biomass and net-sampled biomass.
|
5 Conclusions |
|---|
|
Top1 Introduction2 Methodology3 Results4 Discussion5 ConclusionsReferences |
|---|
The results presented in this paper are part of the ongoing quest to define the relationship between acoustic backscatter and zooplankton using simple manipulation of the zooplankton data to resemble mean volume-backscattering strength (following FS2), through to the use of modern acoustic-scattering models (e.g. Stanton et al., 1998a, b).
By comparing zooplankton biovolume and MVBS directly, it is possible to conclude that they are related, although the relationship is more complex than simple increases in biovolume resulting in higher acoustic intensities. This study shows comprehensively that direct comparisons and the use of linear relationships between ADCP acoustic backscatter and net-sampled zooplankton are not reproducible, and that without further manipulation of the data, ADCP mean volume-backscattering strength should be used as a descriptive tool rather than for deriving estimates of zooplankton abundance and biomass (e.g. Velez-Belchi et al., 2002).
The above conclusion, inherent to any single-frequency acoustic system, has resulted in the investigation of what many authors have termed "the forward problem" (Holliday and Pieper, 1995; Griffiths et al., 2002). The forward-problem approach of comparing acoustic backscatter observed in situ with that predicted from a combination of net-sampled data and acoustic-scattering models has been applied in zooplankton studies rarely and only relatively recently (Zhou et al., 1994; Wiebe et al., 1996; Greene et al., 1998). This study presents the first results comparing ADCP MVBS with model predictions calculated directly from a net-sampled, open-ocean, mixed zooplankton population. It should be noted that solving the forward problem does not lead directly to the goal of the acoustic determination of animal populations. However, resolution of this problem can provide valuable diagnostic information such as determining the relative contributions of each "significant acoustic-scattering" group to the TMPB, the potential to detect when the acoustic-scattering models are inadequate, permitting the detection of inadequate acoustic or net sampling, and, finally, determining the consistencies between observed MVBS and the net-sampled data, assuming that the models are correct and the nets used are appropriate in terms of the zooplankton size sampled.
Our overall conclusion, therefore, is that whilst ADCPs, and any other single-frequency echosounder, are useful for some aspects of biophysical coupling and the determination of patterns and processes, they must be used with extreme caution if attempts are made to extract quantitative size and biomass data from them.
|
Acknowledgements |
|---|
We thank the master, crew, and scientists on board RRS Discovery Cruise 209. We would also like to thank Dr P. Wiebe and Dr T. Stanton for the provision and use of the acoustic-scattering models. This research was funded by NERC Special Topic programme PRIME.
|
References |
|---|
|
Top1 Introduction2 Methodology3 Results4 Discussion5 ConclusionsReferences |
|---|
-
Ashjian C.J., Smith S.L., Flagg C.N., Mariano A.J., Behrens W.J., Lane P.V.Z. (1994) The influence of a Gulf Stream meander on the distribution of zooplankton biomass. The Slope Water, the Gulf Stream, and the Sargasso Sea, described using a shipboard Acoustic Doppler Current Profiler. Deep-Sea Research I 41:1 pp. 23–50.
Batchelder H.P., Vankeuren J.R., Vaillancourt R., Swift E. (1995) Spatial and temporal distributions of acoustically estimated zooplankton biomass near the marine light-mixed layers station (59°30'N 21°00'W) in the North Atlantic in May 1991. Journal of Geophysical Research – Oceans 100:C46549–6563.[CrossRef][Web of Science]
Beers J.R. (1976) Determination of zooplankton biomass. In Steedman H.F. (Ed.). Zooplankton Fixation and Preservation(UNESCO Press, Paris) pp. 35–84.
Bernhard J.M. and Sen Gupta B.K. (1999) Foraminifera of oxygen-depleted environment. In Sen Gupta B.K. (Ed.). Modern Foraminifera(Kluwer Academic Publishers, Dordrecht) pp. 201–216.
Brierley A.S., Brandon M.A., Watkins J.L. (1998) An assessment of the utility of an Acoustic Doppler Current Profiler for biomass estimation. Deep-Sea Research I 45:91555–1573.[CrossRef]
Childress J.J. and Thuesen E.V. (1992) Metabolic potential of deep-sea animals: regional and global scales. In Rowe G.T. and Pariente V. (Eds.). Deep-Sea Food Chains and the Global Carbon Cycle(Kluwer Academic, Dordrecht) pp. 217–236.
Chu D.Z., Foote K.G., Stanton T.K. (1993) Further analysis of target-strength measurements of Antarctic krill at 38 and 120 kHz – comparison with deformed cylinder model and inference of orientation distribution. Journal of the Acoustical Society of America 93:52985–2988.[CrossRef][Web of Science]
Chu D.Z. and Stanton T.K. (1998) Application of pulse-compression techniques to broadband acoustic scattering by live individual zooplankton. Journal of the Acoustical Society of America 104:139–55.[CrossRef][Web of Science]
Clay C.S. and Medwin H. (1977) Acoustical Oceanography: Principles and Applications(Wiley-Interscience, New York) 544 pp.
Crisp N. and Harris A. (2000) TUBA II – a compact multifrequency sonar suited to use in autonomous or towed platforms for the study of upper-ocean zooplankton distribution and abundance(IEEE Underwater Technology, Tokyo, Japan).
Fielding S., Crisp N., Allen J.T., Hartman M.C., Rabe B., Roe H.S.J. (2001) Mesoscale subduction at the Almeria-Oran front. part 2. Biophysical interactions. Journal of Marine Systems 30:287–304.[CrossRef][Web of Science]
Flagg C.M. and Smith S.L. (1989) On the use of the Acoustic Doppler Current Profiler to measure zooplankton abundance. Deep-Sea Research I 36:455–474.
Flagg C. N. and Smith S. L. (1989b) Zooplankton-abundance measurements from Acoustic Doppler Current Profilers. OCEAN '89, Marine Technology Society and I.E.E.E.Seattle, WA.
Folt C.L. and Burns C.W. (1999) Biological drivers of zooplankton patchiness. Tree 14:8300–305.[Medline]
Foote K.G. (1983) Maintaining precision calibrations with optimal copper spheres. Journal of the Acoustical Society of America 73:31054–1063.[CrossRef][Web of Science]
Foote K.G. (1990) Speed of sound in Euphausia superba. Journal of the Acoustical Society of America 87:1405–1408.[CrossRef][Web of Science]
Greene C.H., Stanton T.K., Wiebe P.H., McClatchie S. (1991) Acoustic estimates of Antarctic krill. Nature 349:110.[CrossRef][Web of Science]
Greene C.H., Wiebe P.H., Burczynski J. (1989) Analyzing zooplankton size distributions using high-frequency sound. Limnology and Oceanography 34:1129–139.[Web of Science]
Greene C.H., Wiebe P.H., Pershing A.J., Gal G., Popp J.M., Copley N.J., Austin T.C., Bradley A.M., Goldsborough R.G., Dawson J., Hendershott R., Kaartvedt S. (1998) Assessing the distribution and abundance of zooplankton: a comparison of acoustic and net-sampling methods with D-BAD MOCNESS. Deep-Sea Research II 45:71219–1237.[CrossRef]
Griffiths G. and Diaz J.I. (1996) Comparison of acoustic-backscatter measurements from a ship-mounted Acoustic Doppler Current Profiler and an EK500 scientific echosounder. ICES Journal of Marine Science 53:2487–491.
Griffiths G., Fielding S., Roe H.S.J. (2002) Biological–physical–acoustical interactions. In Robinson A.R., McCarthy J.J., Rothschild B.J. (Eds.). Biological–Physical Interactions in the Sea(John Wiley & Sons, Inc, New York) pp. 441–474.
Haury L.R., McGowan J.A., Wiebe P.H. (1978) Patterns and processes in the timescales of plankton. In Steele J. (Ed.). NATO Conference Series IV: Marine SciencesPlenum Press pp. 277–327.
Herring P.J., Fasham M.J.R., Weeks A.R., Hemmings J.C.P., Roe H.S.J., Pugh P.R., Holley S., Crisp N.A., Angel M.V. (1998) Across-slope relations between the biological populations, the euphotic zone and the oxygen-minimum layer off the coast of Oman during the southwest monsoon (August, 1994). Progress in Oceanography 41:69–109.[CrossRef][Web of Science]
Heywood K.J. (1996) Diel vertical migration of zooplankton in the Northeast Atlantic. Journal of Plankton Research 18:2163–184.
Heywood K.J., Scropehowe S., Barton E.D. (1991) Estimation of zooplankton abundance from shipborne ADCP backscatter. Deep-Sea Research I 38:6677–691.
Holliday D.V. and Pieper R.E. (1995) Bioacoustical oceanography at high frequencies. ICES Journal of Marine Science 52:3–4279–296.[CrossRef][Web of Science]
Kristensen A. and Dalen J. (1986) Acoustic estimation of size distribution and abundance of zooplankton. Journal of the Acoustical Society of America 80:601–611.[CrossRef][Web of Science]
Longhurst A.R., Reith A.D., Bower R.E., Seibert D.L.R. (1966) A new system for the collection of multiple serial plankton samples. Deep-Sea Research 13:212–222.
Macaulay M.C. (1994) A generalized target-strength model for euphausiids, with applications to other zooplankton. Journal of the Acoustical Society of America 95:2452–2466.[CrossRef][Web of Science]
McGehee D., O'Driscoll R.L., Traykovski L.V.M. (1998) Effects of orientation on acoustic scattering from Antarctic krill at 120 kHz. Deep-Sea Research I 45:1273–1294.
Nichols J.H. and Thompson A.B. (1991) Mesh selection of copepodite and nauplius stages of four calanoid-copepod species. Journal of Plankton Research 13:3661–671.
Orr M.H. (1981) Remote acoustic detection of zooplankton response to fluid processes, oceanographic instrumentation and predators. Canadian Journal of Fisheries and Aquatic Science 38:1096–1105.
Plueddemann A.J. and Pinkel R. (1989) Characterization of the patterns of diel migration using a Doppler sonar. Deep-Sea Research I 36:509–530.
Postel L., Fock H., Hagen W. (2000) Biomass and abundance. In Harris R., Wiebe P.H., Lenz J., Skjoldal H.R., Huntley M. (Eds.). Zooplankton Methodology Manual(Academic Press, San Diego, CA) pp. 83–192.
RDI. (1990) Calculating absolute backscatter(RD Instruments, San Diego, California) 24 pp.
Ressler P.H. (2002) Acoustic-backscatter measurements with a 153 kHz ADCP in the northeastern Gulf of Mexico: determination of dominant zooplankton and micronekton scatterers. Deep-Sea Research I 49:2035–2051.[CrossRef]
Ricker W.E. (1973) Linear regressions in fishery research. Journal of the Fisheries Research Board of Canada 30:3409–435.[Web of Science]
Roe H.S.J., Griffiths G., Hartman M., Crisp N. (1996) Variability in biological distributions and hydrography from concurrent Acoustic Doppler Current Profiler and SeaSoar surveys. ICES Journal of Marine Science 53:131–138.
Smith S.L. and Lane P.V. (1988) Grazing of the spring diatom bloom in the New York Bight by the calanoid copepods Calanus finmarchicus, Metridia lucens, and Centropages typicus. Continental Shelf Research 8:485–510.[CrossRef][Web of Science]
Smith P.E., Ohman M.D., Eber L.E. (1989) Analysis of the patterns of distribution of zooplankton aggregations from an Acoustic Doppler Current Profiler. California Cooperative Oceanic Fisheries Investigations Reports 30:88–103.
Stanton T.K., Chu D., Wiebe P.H., Clay C.S. (1993) Average echoes from randomly orientated, random-length finite cylinders: zooplankton models. Journal of the Acoustical Society of America 94:3463–3472.[CrossRef][Web of Science]
Stanton T.K., Clay C.S., Chu D. (1993) Ray representation of sound scattering by weakly scattering, deformed, fluid cylinders: simple physics and application to zooplankton. Journal of the Acoustical Society of America 94:3454–3462.[CrossRef][Web of Science]
Stanton T.K., Wiebe P.H., Chu D.Z., Benfield M.C., Scanlon L., Martin L., Eastwood R.L. (1994) On acoustic estimates of zooplankton biomass. ICES Journal of Marine Science 51:4505–512.
Stanton T.K., Chu D.Z., Wiebe P.H. (1996) Acoustic-scattering characteristics of several zooplankton groups. ICES Journal of Marine Science 53:2289–295.
Stanton T.K., Chu D.Z., Wiebe P.H., Martin L.V., Eastwood R.L. (1998) Sound scattering by several zooplankton groups. I. Experimental determination of dominant scattering mechanisms. Journal of the Acoustical Society of America 103:1225–235.[CrossRef][Web of Science][Medline]
Stanton T.K., Chu D.Z., Wiebe P.H. (1998) Sound scattering by several zooplankton groups. II. Scattering models. Journal of the Acoustical Society of America 103:1236–253.[CrossRef][Web of Science][Medline]
Van Der Spoel S. and Heyman R.P. (1983) A comparative atlas of zooplankton: biological patterns in the oceans(Springer-Verlag, Berlin) 186 pp.
Velez-Belchi P., Allen J.T., Strass V.H. (2002) A new way to look at mesoscale zooplankton distributions: an application at the Antarctic Polar Front. Deep-Sea Research Part II – Topical Studies in Oceanography 49:183917–3929.[CrossRef]
Wade I.P. and Heywood K.J. (2001) Acoustic-backscatter observations of zooplankton abundance and behaviour and the influence of oceanic fronts in the Northeast Atlantic. Deep-Sea Research II 48:4–5899–924.[CrossRef]
Wiebe P.H. (1988) Functional regression equations for zooplankton displacement volume, wet weight, dry weight, and carbon: a correction. Fishery Bulletin 86:4833–835.[Web of Science]
Wiebe P.H., Boyd S., Cox J.L. (1975) Relationships between zooplankton-displacement volume, wet weight, dry weight, and carbon. Fishery Bulletin 73:4777–787.[Web of Science]
Wiebe P.H., Mountain D.G., Stanton T.K., Greene C.H., Lough G., Kaartvedt S., Dawson J., Copley N. (1996) Acoustical study of the spatial distribution of plankton on Georges Bank and the relationship between volume-backscattering strength and the taxonomic composition of the plankton. Deep-Sea Research II 43:7–81971–2001.[CrossRef]
Williams R., Collins N.R., Conway D.V.P. (1983) The double LHPR system, a high-speed microplankton and macroplankton sampler. Deep-Sea Research I 30:3331–342.
Zhou M., Nordhausen W., Huntley M. (1994) ADCP measurements of the distribution and abundance of euphausiids near the Antarctic peninsula in winter. Deep-Sea Research I 41:91425–1445.[CrossRef]
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Lara-Lopez and F. J. Neira Synchronicity between zooplankton biomass and larval fish concentrations along a highly flushed Tasmanian estuary: assessment using net and acoustic methods J. Plankton Res., September 1, 2008; 30(9): 1061 - 1073. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Zedel, R. Patro, and T. Knutsen Fish behaviour and orientation-dependent backscatter in acoustic Doppler profiler data ICES J. Mar. Sci., January 1, 2005; 62(6): 1191 - 1201. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||