Acoustic Estimation of Blue Mackerel (Scomber australasicus) Spawning Biomass in Yilan Bay, Taiwan: Integrating Depth Compensation and Fishery Data (2021–2024)

Abstract

The mackerel fishery is Taiwan’s most productive coastal fishery sector, with the blue mackerel (Scomber australasicus) being its primary target species. Given the economic and ecological significance of this fishery, considerable attention has been devoted to assessing stock status and promoting sustainable use. Between 2021 and 2024, acoustic transect surveys were conducted in Yilan Bay during the blue mackerel spawning season, supplemented by hook-and-line sampling to confirm the identity of single-target acoustic signals. Acoustic detections within ±10 m of capture depth and ±10 min of capture time were used to establish a depth-compensated regression model linking target strength (TS) to fork length (FL). Validation revealed that over 80% of the hook-and-line samples were blue mackerel. After careful noise filtering, a depth-compensated regression model was established to relate TS to FL and sampling depth. The model incorporated both logarithmic body length and depth terms, effectively accounting for vertical variations in TS. The model improved alignment with biological sampling data by effectively accounting for depth-related variations in TS, thereby enhancing biomass estimation accuracy. Cross-validation with auction records from Nan-Fang-Ao Fishing Harbor confirmed that the acoustic biomass estimates closely mirrored commercial catch trends. These findings highlight the effectiveness of depth-compensated acoustic methodologies for obtaining reliable, fishery-independent spawning biomass estimates, supporting their continued application in long-term monitoring and spatial resource management.

1. Introduction

The Taiwanese purse-seine fishery is one of the most productive coastal and offshore fisheries in the region, primarily targeting blue mackerel (Scomber australasicus), chub mackerel (Scomber japonicus), and jack mackerel (Trachurus japonicus) [1,2]. The northeastern waters of Taiwan, particularly the area between Pengjia Islet and Nan-Fang-Ao, are critical spawning and feeding grounds for these pelagic species. These habitats are shaped by dynamic oceanographic features, including the Kuroshio Current and coastal upwelling [3]. Among the targeted species, blue mackerel is the most abundant in Taiwanese waters and typically occupies surface to midwater depths, extending to approximately 200 m [4]. In the East China Sea, blue mackerel migrate southward to the waters around Taiwan to spawn between January and May and then gradually return northward as sea surface temperatures rise in summer [5].

In recent decades, concerns have been raised over the declining abundance of mackerel stocks, driven by a combination of intensified fishing pressure and shifts in oceanographic conditions. These concerns are exacerbated by the ecological importance of mackerel and the observed decline in catches [6]. In response, the Taiwanese government introduced the Mackerel and Carangid Fishery Management Measures in February 2013 through the Fisheries Agency of the Council of Agriculture—now the Fisheries Agency of the Ministry of Agriculture (FA, MOA)—to promote sustainable use of key pelagic resources. These measures include an annual ban from June 1 to June 30, during which the use of net gears targeting mackerel and carangid species is prohibited north of 24° N latitude. In 2018, additional restrictions were proposed, extending the closure to a 20-day period during the Lunar New Year (from the 29th day of the 12th lunar month to the 18th day of the 1st lunar month), with flexibility for future adjustments based on resource status and management outcomes.

However, environmental variability may be altering the reproductive dynamics of mackerel populations [7]. For instance, Sinaga et al. [8] reported shifts in the spawning season and frequency of blue mackerel, possibly linked to rising sea temperatures. In addition, the effectiveness of local management measures may be undermined by external fishing pressures and climate-induced habitat changes [2]. Most importantly, many studies rely on fishery-dependent data—such as landings statistics and biological sampling—raising concerns about their ability to capture real-time fluctuations in spawning biomass.

Over the past decade, numerous investigations have attempted to characterize the abundance and distribution of mackerel and carangid resources [9,10,11]. However, most of these studies relied on fishery-dependent indicators. In the northeastern waters of Taiwan, critical spawning habitats have been identified around the Northern Three Islets (Pengjia, Mianhua, and Huaping), Diaoyutai Island, and Yilan Bay (Figure 1), based mainly on commercial fishing activities and plankton-based indices [8,12]. Wei [13] described blue mackerel spawning grounds extending from the Mianhua Submarine Canyon to the offshore waters of Yilan, with peak spawning between March and May. Liao et al. [14] similarly documented chub mackerel spawning between Pengjia Islet and the offshore area near Su’ao during the same period. These findings suggest that Yilan Bay is the southernmost boundary of blue mackerel spawning activity.

However, dependence on fishery-based indicators for stock assessments introduces substantial uncertainty. Estimates derived from such data are prone to biases from variable fishing effort, environmental fluctuations, and market-driven dynamics [15,16,17]. In Taiwan, assessments of mackerel resources have primarily relied on fishery-dependent methods, including catch reports and port sampling [18,19]. While these approaches provide valuable information, they are limited by potential biases related to fishing effort and fisher behavior. Although prior studies offered valuable insights into the temporal and spatial distribution of spawning [13,14], the absence of fishery-independent assessments limits the reliable estimation of the true spawning biomass.

To address this gap, the present study applies fishery-independent acoustic surveys to estimate the spawning biomass of blue mackerel in Yilan Bay, located at the southern edge of their known spawning grounds. This approach provides spatially resolved insights into spawning habitat use and complements existing fishery-dependent knowledge of blue mackerel reproductive ecology. Acoustic surveys—widely applied in fisheries science for estimating pelagic species distribution and abundance—are recognized for their non-destructive nature [20,21,22,23]. Given the seasonality and spatial aggregation of blue mackerel during spawning, acoustic methods are particularly well suited for monitoring these populations.

To enhance the robustness of acoustic-based biomass estimates, this study incorporates a depth-adjusted TS model to correct for vertical variability in echo intensity, which is known to introduce bias if unaccounted for [24,25]. Acoustic-derived biomass estimates are then compared with commercial fishery records to assess consistency and evaluate the applicability of acoustic methods in regional stock assessments. Drawing on four years of fishery-independent acoustic surveys, this study provides a more accurate, ecologically grounded framework for estimating spawning biomass and informing sustainable fishery management in Taiwanese waters.

2. Materials and Methods

2.1. Survey Area and Transect Design

The research area, Yilan Bay in northeastern Taiwan (Figure 1), is a key fishing ground for blue mackerel. Acoustic surveys were conducted during the spawning seasons from 2021 to 2024, with 8 transects in 2021 and 2022 and 6 in 2023 and 2024. Although acoustic surveys were conducted in all four years (2021–2024), biological sampling data (e.g., catch composition, TS–FL regression) were available only for 2021–2023, as logistical constraints prevented sampling in 2024. Nevertheless, acoustic and commercial data from 2024 were included to maintain a continuous four-year dataset. The survey design aimed to ensure a coverage coefficient greater than 2 whenever possible, in accordance with Aglen [26], who recommended this threshold to reduce random error in acoustic abundance estimates. The transect layout was designed to span various depths, crossing isobaths from the shallow coastal zone to approximately 200 m, thereby encompassing the known vertical distribution range of spawning blue mackerel in Yilan Bay. Previous studies report that blue mackerel in Taiwanese waters spawn primarily from January to May, with peak activity in April [27,28]. Therefore, all surveys in this study were scheduled within this period to coincide with peak reproductive aggregations.

Commercial fishing vessels were employed to conduct acoustic surveys in 2021, 2022, and 2024 with a portable Simrad EY60 echo sounder (Kongsberg Maritime, Kongsberg, Norway). In 2023, the survey was conducted using the R/V Fishery Researcher 2, equipped with a Simrad EK60 echo sounder system (Kongsberg Maritime, Kongsberg, Norway). Survey speeds were maintained between 4 and 6 knots to ensure data stability and minimize distortion from vessel motion.

Given that species such as mackerels exhibit positive phototaxis and are frequently targeted at night with light-based fishing techniques [8,29], all surveys were conducted during daylight hours. This protocol aimed to minimize behavioral disturbance and improve the accuracy of acoustic detections by capturing fish in their natural, undisturbed distribution. The phototactic behavior of mackerels has also been confirmed experimentally under controlled light conditions, particularly in Atlantic mackerel [30,31].

2.2. Echo Sounder Calibration

To ensure accurate acoustic measurements, the echo sounder system was calibrated prior to each survey following established protocols [32]. Calibration was performed using Simrad ER60 software (v2.1.0, Kongsberg Maritime, Kongsberg, Norway), with data recorded during the procedure. A standard 13.7 mm calibration sphere was suspended 15 to 20 m from the transducer.

Before each calibration, water temperature and salinity were measured with a thermometer and a salinometer to calculate the speed of sound in water. This step minimized potential deviations and ensured reliable calibration results under varying environmental conditions. The echo sounder system settings for each device are summarized in

Table 1. Acoustic parameters used in the echo sounder system.

Parameter	Setting
System	EY60	EK60
Frequency	200 kHz	200 kHz
Absorption coefficient	0.070 dB/m	0.083 dB/m
Pulse rate	1 pings/s	1 pings/s
Pulse length	1.024 ms	1.024 ms
Transmitted Power	1000 W	1000 W
Two-way beam angle	−20.70 dB	−20.70 dB
Transducer gain	26.54 dB	27.59 dB
Maximum depth	200 m	200 m

.

2.3. Experimental Operations

Biological sampling was conducted using hook-and-line methods [33], with capture time, depth, and species systematically recorded. To distinguish between blue mackerel (Scomber australasicus) and chub mackerel (Scomber japonicus), preliminary identification followed morphological criteria referenced from Taiwan FishBase [34]. Specifically, individuals with more than 11 first dorsal fin spines were classified as blue mackerel. Those with exactly 10 spines were further examined for spots below the lateral line extending to the abdomen; if present, they were also classified as blue mackerel. Corresponding echo sounder data were analyzed within a 20 m vertical window (±10 m from capture depth) and a 10 min interval preceding each catch. The mean TS within this window was assigned as the TS value for the individual fish [35,36]. This approach established the relationship between individual body length and acoustic backscatter intensity (TS), forming the basis for subsequent TS–FL model development [37]:

$$TS=m\times \log L-b$$

(1)

m: slope.

b: intercept.

Matched biological and acoustic data were grouped into 1 cm FL intervals and stratified into shallow (≤50 m) and deep (>50 m) layers. For each group, mean TS and mean capture depth were calculated. A simple linear regression was then applied to quantify the compensatory relationship between TS and depth. The resulting depth-corrected TS values were subsequently integrated into the original TS–FL model, yielding an adjusted TS–FL–depth model. This approach accounted for both morphological and environmental influences on acoustic backscatter, thereby improving length estimation accuracy under varying depth conditions [38,39].

Despite these methodological improvements, some limitations remain. Hook-and-line sampling, for instance, may introduce selectivity bias by favoring more active individuals. Additionally, environmental factors such as temperature and salinity can affect acoustic signal propagation. Nonetheless, careful instrument calibration and standardized survey protocols were rigorously applied to minimize these sources of uncertainty.

2.4. Data Processing and Analysis

Acoustic data were processed and analyzed using Echoview software (v12, Echoview Software Pty Ltd., Vancouver, Australia). Echoes were classified into biological and non-biological signals. Biological echoes formed the basis for biomass estimation, while non-biological signals—originating from surface turbulence, fishing gear, air bubbles, or electronic noise—were filtered out to ensure data quality [40]. For instance, signals detected within five meters of the surface were excluded, and non-biological echoes were labeled as “bad data.” During the data export, these echoes were assigned blank values and omitted from subsequent echo integration analyses.

The Elementary Sampling Distance Unit (ESDU) was defined as a horizontal interval of 500 m and a vertical interval of 50 m. Since blue mackerel typically inhabit depths shallower than 200 m, acoustic analyses were limited to a maximum depth of 200 m.

The abundance index S_v is defined as the mean scattering intensity per unit volume [37]:

$$s_v=\left({\displaystyle \frac{n\times \mathsf{\sigma }\mathrm{b}\mathrm{s}}{V}}\right),S_v=10{\log }_{10}{s}_v$$

where s_v is the linear volume backscattering coefficient (m⁻¹), n is the number of targets in the sampled volume, σ_bs is the backscattering cross-section of a single individual (m²), V is the sampled water volume (m³), and S_v is the mean volume backscattering strength (dB re 1 m⁻¹).

By substituting definitions and applying a logarithmic transformation, the volume backscattering strength S_v can be explicitly related to known acoustic and physical parameters, namely the target number density (n/V) and the TS:

$$S_v=10{\log }_{10}\left({\displaystyle \frac{n\times {\mathsf{\sigma }}_{\mathrm{b}\mathrm{s}}}{V}}\right)=10{\log }_{10}{\displaystyle \frac{n}{V}}+10{\log }_{10}{\mathsf{\sigma }}_{\mathrm{b}\mathrm{s}}=10{\log }_{10}{\displaystyle \frac{n}{V}}+TS$$

(2)

where S_v is the mean volume backscattering strength (dB re 1 m⁻¹), n is the number of targets, σ_bs is the backscattering cross-section of a single individual (m²), V is the sampled water volume (m³), and TS is the target strength of an individual fish (dB re 1 m²).

From this, the total number of individuals (n) in the sampled volume can be calculated using

$$n={10}^{{\displaystyle \frac{S_v-TS}{10}}}\times V$$

(3)

where n is the number of targets, S_v is the volume backscattering strength (dB re 1 m⁻¹), TS is the target strength (dB re 1 m²), and V is the sampled water volume (m³).

To facilitate comparison with other studies and to express abundance in standardized units, S_v was also converted to the Nautical Area Scattering Coefficient (NASC), the vertically integrated backscatter standardized to one nautical mile [37]. While the NASC was used for reporting standardized abundance indices, subsequent biomass estimates were calculated directly from S_v First, S_v was converted to its linear form:

$$s_v={10}^{{\displaystyle \frac{S_v}{10}}}$$

The area backscattering coefficient was then obtained by integrating s_v over the vertical thickness of the scattering layer (T):

$$s_a={\int }_{Z_1}^{Z_2}{s}_{v}dz\approx s_v\times T$$

Finally, the NASC was calculated by scaling s_a to standardized units in nautical miles:

$$s_A=4\mathsf{\pi }\left({1852}^2\right) \times s_a=4\mathsf{\pi } \times {1852}^2\times {10}^{{\displaystyle \frac{Sv}{10}}}\times T$$

(4)

s_A represents the NASC (m²/n.mi.²), 4π is the geometric factor used in the conversion of S_v (dB) to scattering cross-section per unit area, S_v is the volume backscattering strength (dB), and T is the thickness of the scattering layer (m).

The biomass estimation process can be summarized in a stepwise manner, illustrating how acoustic measurements are converted into target density and biomass:

1.

Filter echoes to remove non-biological signals.

2.

Compute S_v from the filtered echoes.

3.

Calculate the number of targets (n) using S_v and TS.

4.

Convert S_v to the NASC for standardized abundance reporting, while retaining S_v for standing-crop calculations.

5.

Estimate individual weight using length–weight relationships.

6.

Calculate the standing crop in the survey area.

Acoustic estimates were converted to biomass using length–weight relationships to facilitate comparison with commercial fishery data. To reduce inaccuracies in weight measurements from fishing experiments, TS values were first converted to fork length (FL) and then to body weight (BW) using the following equation [6]:

$$BW=0.005 \times {FL}^{3.276}$$

(5)

where BW is the body weight (g) and FL is the fork length (cm)

After obtaining individual weight estimates, and assuming a uniform density of blue mackerel within and outside the surveyed transects, total standing biomass in Yilan Bay was calculated in two steps. First, the total number of individuals (n) in the sampled volume (V) was derived from the measured S_v and the known TS using

$$\mathrm{n}={10}^{{\displaystyle \frac{S_v-TS}{10}}}\times \mathrm{V}$$

where n is the number of blue mackerel, S_v is the volume backscattering strength (dB re 1 m⁻¹), TS is the target strength of an individual fish (dB re 1 m²), and V is the sampled water volume (m³).

Second, the density of blue mackerel (D) was calculated using

$$\mathrm{D}={\displaystyle \frac{n}{V}}={10}^{{\displaystyle \frac{S_v-TS}{10}}}$$

Finally, the total standing crop in the survey area was obtained by multiplying the density by the total survey volume and the mean individual weight (W):

$$Standing crop =\mathrm{V} \times \mathrm{D}\times W$$

(6)

where Standing crop is the total biomass of blue mackerel within the surveyed area (kg), V is the water volume of the surveyed area (m³), D is the density of blue mackerel (ind./m³), and W is the mean weight of blue mackerel (kg).

These calculations enabled the derivation of both individual density and biomass estimates, forming the basis for subsequent comparison with commercial landings.

2.5. Commercial Fishing Data

Commercial catch data for blue mackerel were obtained from Nan-Fang-Ao Fishery Harbor, the primary landing site for vessels operating in Yilan Bay. These data included total landings recorded throughout the spawning season, as well as short-term landings collected within a 7-day window before and after each acoustic survey. Following the approach of Alglave et al. [41], who combined scientific surveys with commercial fishery data, the present study analyzed commercial catch data alongside acoustic biomass estimates to assess both long-term seasonal trends and short-term variability in blue mackerel abundance (

Table 2. Commercial fishing sampling periods corresponding to acoustic surveys.

Year	Period Type	Date Range	Total Weight (Tons)
2021	Entire spawning period	4 January to 31 May	13,609
	±7 days of survey	21 March to 7 April	571
2022	Entire spawning period	1 January to 31 May	7170
	±7 days of survey	8 March to 26 March	273
2023	Entire spawning period	4 January to 24 May	21,917
	±7 days of survey	13 March to 30 March	5116
2024	Entire spawning period	1 January to 31 May	21,977
	±7 days of survey	17 March to 3 April	4290

).

3. Results

3.1. Catch Composition and Sampling Overview

During the three acoustic surveys, hook-and-line sampling showed that blue mackerel consistently dominated the catch composition in Yilan Bay throughout the spawning season. In 2021, they accounted for 87% of the total catch. Although this proportion declined slightly to 80% in 2022, other species—such as chub mackerel, jack mackerel, skipjack tuna, and hairtail—were also recorded. By 2023, all individuals caught were blue mackerel (Figure 2).

This analysis focuses on the period from 2021 to 2023, when catch composition data were systematically collected using consistent fishing methods across the spawning season. The sustained dominance of blue mackerel, consistently exceeding 80% of the total catch, strengthens the reliability of subsequent species-specific analyses.

Although no catch composition data were collected in 2024 due to logistical constraints, both acoustic survey data and commercial fishery records for that year were still analyzed.

3.2. Depth and TS Distribution

From 2021 to 2023, interannual variation in the primary catch depth was observed during the spawning season in Yilan Bay. In 2021, catches predominantly occurred at depths exceeding 50 m, with most individuals concentrated around 100 m. In contrast, during 2022 and 2023, the main catch depth shifted to shallower waters, centering around 50 m (Figure 3, right). These shifts may reflect environmental changes or behavioral adaptations in spawning aggregations.

TS distributions correspondingly exhibited clear vertical stratification around 50 m in 2021 and 2022, but no stratification in 2023. Additionally, the mean TS value in 2023 was higher than in the previous two years (Figure 3, left).

This analysis focuses on the years 2021 to 2023, when both acoustic and catch-depth data were available. No such data were collected in 2024.

3.3. Frequency Distributions of Fork Length and TS

In 2021, the dominant fork length of blue mackerel in the catch was approximately 36 cm. In 2022, the dominant length decreased to around 34 cm, accompanied by a broader size distribution and a lower average length. By 2023, the dominant catch length returned to approximately 36 cm (Figure 4, right), suggesting interannual variability in cohort composition or growth conditions.

Acoustic measurements over the same period revealed notable differences in TS values. In 2021, a total of 121 TS measurements were recorded, primarily ranging from −50 to −58 dB. In 2022, the number increased to 241, with values clustering between −38 and −42 dB, indicating a substantial rise compared with the previous year. In 2023, 510 measurements were collected, with values more broadly distributed between −36 and −48 dB (Figure 4, left).

Although the annual TS distributions differed, the catch compositions in all years were dominated by the target species, with capture rates exceeding 80% (Figure 2). Therefore, it is reasonable to assume that the TS values used in subsequent analyses primarily represent the target fish, supporting development of a species-specific TS–FL model.

3.4. TS–FL Regression and Depth-Compensation Model

Given that the primary fishing depth was approximately 50 m, this depth was used as a threshold to examine vertical variation in TS. FL data were grouped into 1 cm intervals, and for each interval, the mean TS and mean capture depth were calculated within the shallow (≤50 m) and deep (>50 m) strata. A linear regression was then performed on the averaged TS and depth values per FL class, yielding a depth-correction function (Figure 5). Although Figure 5 is based on only five FL intervals per depth stratum and serves as a preliminary assessment of vertical TS variation, the depth-corrected TS–FL model presented in Figure 6 incorporates all sampled individuals and provides a robust basis for TS estimation.

To ensure analytical consistency, only FL intervals represented in both depth strata across the 2021–2023 surveys were included. Although only five FL classes satisfied this criterion, each regression data point was based on multiple acoustic detections, providing a robust basis for quantifying depth-related variation in TS.

This depth-correction function was subsequently incorporated into the original TS–FL model to generate an adjusted regression equation. Incorporating depth as an explanatory variable substantially improved explanatory power, with the coefficient of determination (R²) increasing from 0.5585 to 0.8139 (Figure 6). The enhanced TS–FL–Depth model effectively reduces bias introduced by vertical variability in acoustic backscatter and enables more accurate size estimation under stratified environmental conditions.

3.5. Estimated Length and Spatial Distribution of Blue Mackerel

Figure 7 compares FL distributions estimated from acoustic data with those derived from hook-and-line sampling for 2021–2023. In 2021, the acoustic-estimated FL distribution was centered around 36 cm but slightly deviated from the measured catch data. In 2022 and 2023, the estimated distributions aligned more closely with observed values, although they tended to overestimate the proportion of larger individuals (>36 cm). This general agreement supports the validity of the depth-compensated TS–FL model for accurately representing the size structure of blue mackerel during the spawning season.

In 2021, relatively high TS values were concentrated near Guishan Island, suggesting localized aggregations of larger individuals. In contrast, TS distributions in 2022 and 2024 were more dispersed, with elevated TS values observed outside Nan-Fang-Ao Fishing Port and at several stations along the northern margin of Yilan Bay. No clear spatial pattern was evident in 2023.

Across the four-year survey period (2021–2024), consistently low TS values were recorded on the southeastern side of Yilan Bay, indicating a lower probability of encountering larger individuals there (Figure 8). The localized aggregation of high TS values observed near Guishan Island in 2021 did not recur in subsequent years, suggesting interannual variation in the spatial distribution of larger blue mackerel.

3.6. Biomass Estimation and Comparison with Commercial Data

In 2021, blue mackerel density was highest among the first three years, with a pronounced concentration south of Guishan Island. In 2022, density declined markedly across the entire survey area. By 2023, intermediate density levels were observed, particularly near Nan-Fang-Ao Fishing Port.

Figure 9 presents the spatial distribution of NASC values, used as a proxy for blue mackerel density, across the four survey years. In 2021, high NASC values were concentrated in the southwestern part of the bay, especially south of Guishan Island. In 2022, overall NASC values dropped significantly, and no clear density hotspots were observed. In 2023, moderate values were observed near Nan-Fang-Ao and along the central coast. By 2024, high-density zones reemerged, particularly in the southeastern and central-eastern regions of Yilan Bay. These results underscore marked interannual variation in the spatial distribution of spawning aggregations.

Commercial catch data collected during the spawning seasons broadly corroborated the acoustic biomass estimates. Seasonal peaks occurred in 2023 and 2024, followed by 2021, with the lowest levels recorded in 2022 (Figure 10). This pattern generally reflects the interannual fluctuations observed in acoustic-based biomass assessments.

However, short-term commercial catch data (within ±7 days of each acoustic survey) differed notably from acoustic results, particularly in 2021 and 2024. These inconsistencies may reflect temporal mismatches between survey timing and fishing activity, environmental variability, or changes in fishing effort.

Despite these short-term deviations, the overall seasonal trends in commercial landings strongly support the reliability of acoustic-based assessments for reflecting spawning stock abundance.

4. Discussion

4.1. Acoustic Estimation and Commercial Fishing

Sunarti et al. [8] reported fork lengths at 50% sexual maturity for blue mackerel as 32.02 cm, 32.14 cm, and 29.64 cm in 2017, 2018, and 2019, respectively. In the present study, most individuals detected acoustically—and partially validated through hook-and-line sampling—exceeded 32 cm FL, indicating that the surveyed population predominantly comprised mature, reproductively active individuals. As shown in Figure 11, the observed size distributions generally surpassed the reported maturity thresholds, further supporting the reproductive relevance of the acoustic detections.

During the study period, acoustically estimated spawning biomass of blue mackerel in Yilan Bay was 27,398, 8071, 28,880, and 29,211 metric tons in 2021, 2022, 2023, and 2024, respectively, while corresponding commercial landings were 13,609, 7170, 21,914, and 21,977 metric tons. These results show a generally positive seasonal relationship between acoustic biomass estimates and commercial catch, particularly in 2023 and 2024, when both were relatively high. Nevertheless, discrepancies in some years—such as the relatively high acoustic biomass estimate but low adjacent-week catch in 2021 and 2024—highlight the influence of factors such as fishing effort distribution, environmental variability, and fish availability. Acoustic surveys therefore reflect seasonal trends rather than provide exact short-term catch predictions. Integrating longer-term acoustic time series with complementary fishery-dependent or alternative assessment methods remains important for capturing interannual dynamics more comprehensively.

Due to the absence of biological sampling in 2024, acoustic-based size and biomass estimates for that year could not be directly validated against measured length distributions. Partial validation is nevertheless supported by the overall congruence between acoustic biomass estimates and commercial catch trends, underscoring the utility of the acoustic approach while acknowledging its limitations.

4.2. Fishing Grounds and Survey Coverage

The primary spawning and fishing grounds for blue mackerel around Taiwan include the Northern Three Islets, the Diaoyu Islands, and Yilan Bay. Yilan Bay was selected as the focal area due to its relatively simpler fishery composition and stricter spatial regulations. Under current government policies, fishing vessels exceeding 100 gross tons are restricted from operating within 6–12 nautical miles of the coastline, limiting industrial fishing pressure in this region.

In most survey years, the coverage coefficient exceeded the recommended value of 2 [26,42], with values of 3.82 in 2021 and 2022, 3.49 in 2023, and 1.8 in 2024. These values indicate that survey resolution was generally sufficient to ensure spatial representativeness and to support reliable estimates of standing stock.

4.3. Factors Influencing Detection Results

Although FL distributions derived from acoustic data generally matched those obtained from hook-and-line sampling, a slight tendency toward overestimation was noted (Figure 7). This discrepancy may reflect gear selectivity, species-specific aggregation, or vertical distribution within the water column.

Furthermore, differences between acoustically estimated spawning biomass and reported commercial landings (Figure 10) suggest that a substantial portion of the spawning biomass remains unharvested during the peak season. This partial exploitation may have implications for stock sustainability and management.

The limited availability of shallow-layer samples constrained the development of a fully depth-stratified TS–FL model (Figure 5), necessitating extrapolation from deeper strata. This underscores the importance of improving vertical sampling resolution in future surveys to enhance model accuracy and reduce potential estimation bias. Sustained long-term monitoring, combined with consistent acoustic methodologies, is essential for tracking interannual changes in spawning dynamics and supporting robust fishery-independent assessments.

4.4. Implications of the Depth-Compensated TS–FL Model

The depth-compensated TS–FL model demonstrates that vertical position significantly affects TS in blue mackerel (Scomber australasicus), highlighting the importance of accounting for depth in acoustic-based size and biomass estimation. While depth was the primary factor identified in this study, other influences such as swimbladder compression, body tilt, and schooling behavior may also contribute to TS variation [20,25]. Incorporating depth into the TS–FL regression improved the accuracy of density and size estimates. Future applications could further enhance reliability by combining depth-stratified acoustic surveys with multi-frequency data and underwater video to capture fish orientation and behavior, providing a more comprehensive understanding of factors influencing TS.

5. Conclusions

This study demonstrates that integrating depth-compensated acoustic methods with fishery sampling can reliably reflect the trends of blue mackerel spawning biomass across an entire spawning season in Yilan Bay. The TS–FL–depth regression model effectively corrected vertical variability in TS and aligned well with biological sampling data. Nevertheless, short-term estimates may deviate from catch data and should therefore be interpreted in the context of fishery operation dynamics and environmental conditions.

These results support the use of hydroacoustic surveys as a valuable complement to traditional fishery-dependent data, particularly in coastal regions with spawning aggregations. The improved temporal and spatial resolution of this approach can contribute to more science-based and adaptive management of pelagic fisheries in Taiwan. Future applications may extend this methodology to adjacent waters and refine the model by incorporating multi-frequency acoustic data and real-time environmental observations.