Spatiotemporal Variation and Predictors of the Purpleback Flying Squid (Sthenoteuthis oualaniensis) Distribution Surrounding the **sha and Zhongsha Islands during a Fishing Moratorium

Abstract

As an economic species widely distributed in the South China Sea (SCS), the purpleback flying squid (Sthenoteuthis oualaniensis) still has a large potential for exploitation, and the variations in its use as a resource are highly correlated with environmental and other factors. In this study, using a generalized additive model (GAM) and gradient forest analysis (GFA), in conjunction with environmental factors, the distribution of purpleback flying squid surrounding the **sha and Zhongsha islands during the fishing moratorium period was investigated. The results indicated that catch per unit effort (CPUE) had a gradual increase from May to July 2023 in the primary fishing area surrounded the **sha Islands during May to June, then moved southward towards 13–15° N after July. CPUE is used as an important indicator to reflect the abundance of the fishery, while the GFA results show that CPUE has a better fit than catch in this study. Therefore, the subsequent analysis focused on CPUE. Longitude and sea surface temperature (SST) were of relative higher importance, followed by sea surface salinity (SSS), latitude, chlorophyll a concentration (Chla), sea surface height (SSH), and mixed layer depth (MLD). Longitude and CPUE had a significant, positive correlation. The CPUE gradually increased with latitude within 14–16° N. The CPUE increased slowly as SST increased from 29.5 to 30.5 °C in the primary fishing area. The Chla in this fishing zone was 0–0.2 mg/m³ and displayed a significant positive association with CPUE. Conversely, SSS, SSH, and MLD had negative correlations with CPUE. These findings will promote the sustainable utilization of purpleback flying squid in the SCS.

1. Introduction

The South China Sea (SCS) is a closed marginal sea in the western Pacific Ocean that is characterized by an extensive continental shelf [1]. The **sha Islands are situated in the western part of the SCS, spanning an area of more than 500,000 km². The Zhongsha Islands are situated in the central region of the SCS, with the **sha Islands located an additional 100 km away. The SCS is situated in the north-central northwest Pacific, serving as a transitional zone between the continental shelf and the deep sea extension. In this area, water depth ranges from 100 to 1000 m in its western part and can reach up to 3000–4000 m in its eastern section. Additionally, the sea area surrounding the **sha and Zhongsha islands exhibits favorable climatic conditions for marine ecosystems and is predominantly influenced by the northeast monsoon during winter, which is characterized by elevated wind velocities and declining temperatures [2]. In the summer period, the dominant southerly air currents, coupled with elevated temperature and humidity levels, have a substantial influence on tropical cyclones and precipitation patterns [2]. This region also serves as the primary tropical fishing ground in China, with a rich diversity of coral reef and pelagic fish species. Additionally, it is recognized as one of the key migratory sites for purpleback flying squid (Sthenoteuthis oualaniensis) [3,4].

The purpleback flying squid is an annual warm-water oceanic cephalopod with a wide distribution throughout the SCS [5]. It has a complex population structure, a short life cycle, robust reproductive capacity, high natural mortality rates, rapid growth rate, and substantial catch [6]. It has previously been reported that the abundance and exploitable biomass of purpleback flying squid in the SCS exceed one million tons, and it is widely acknowledged that the deep-water region of the south-central SCS harbors a substantial population of purpleback flying squid with considerable developmental potential [7,8]. The stock structure of purpleback flying squid is intricate, and it is widely acknowledged that the purpleback flying squid found in the SCS can be categorized into medium and dwarf forms based on variations in dorsal luminescence [9,10]. The medium form represents the dominant stock, and is distributed extensively throughout the SCS, while the abundance of dwarf-form individuals remains relatively limited [11]. The purpleback flying squid lives in a habitat layer with a seawater temperature range of 14–31 °C and displays distinct vertical movement patterns [5]. It stays at depths below 200 m during the day but frequently ascends to the middle and upper sea regions for nocturnal feeding. Seasonal migrations of this squid are noticeable and are mainly characterized by movements from shallow to deep sea areas during winter and from deep to shallow sea areas for spawning [5,12].

Previous studies have demonstrated that migratory pelagic fish distributions exhibit a preference for feeding environments that are conducive to their growth while simultaneously seeking out optimal temperature conditions [13,14]. The distribution of purpleback flying squid in fishing grounds is significantly influenced by marine environmental factors, including sea surface temperature (SST), sea surface height (SSH), and sea surface chlorophyll a concentration (Chla) [15]. Previous studies have shown that variations in the marine environment within areas such as fishing grounds and spawning grounds can result in changes to the spatial habitat preferences of purpleback flying squid for spawning and feeding activities, consequently leading to fluctuations in its population abundance [16]. Studies of the spatiotemporal distribution of purpleback flying squid in the waters near the **sha and Zhongsha islands have revealed a strong correlation between its abundance and environmental factors such as SST, Chla, and SSH. However, previous studies of this species have primarily focused on a limited range of environmental factors when analyzing its habitat distribution [4,17], without considering multivariate environmental factors or employing statistical models to examine their characteristics in waters, such as those near the **sha and Zhongsha islands. It is generally believed that the waters around the islands and reefs in the SCS may be important spawning areas for purpleback flying squid. Consequently, alterations in the marine environment of these regions will likely have a substantial influence on the crucial early life history processes of the species, such as the spawning behavior, with the peak spawning period occurring from April to July [18,19]. Furthermore, the fishing moratorium period (refers to a system for the protection of fishery resources organized and implemented by the General Administration of Fisheries of China, which stipulates that certain operations are not allowed to engage in fishing operations at certain times of the year and in certain waters) in the SCS waters north of 12° N from May to August has restricted the capture of purpleback flying squid to the waters south of 12° N in the Nansha Islands. No fishing vessels are permitted in the vicinity of the **sha and Zhongsha islands during this period. Therefore, studies of the distribution of the purpleback flying squid fishing ground in this area during the period of the fishing moratorium are relatively scarce. In this study, we utilized catch data from the fishing moratorium period in 2023, focusing on the primary fishing locations situated around the **sha and Zhongsha islands. Using a generalized additive model (GAM) and gradient forest analysis (GFA), we analyzed the spatiotemporal variability of the purpleback flying squid resource. Additionally, we investigated the influence of various environmental factors such as SST, sea surface salinity (SSS), Chla, SSH, and mixed layer depth (MLD) on its use as a resource. This comprehensive investigation provides valuable theoretical support for sustainable utilization and management strategies to ensure the resource of purpleback flying squid in the SCS. Simultaneously, it can also offer scientific guidance for the specialized fishing operations targeting purpleback flying squid during this specific season and in this particular region.

2. Material and Methods

2.1. Data Collection

2.1.1. Fisheries Data

As it is a species with potential for exploitation, it is necessary to carry out rational and specialized fishing activities for the purpleback flying squid in order to rationally utilize this species. With the approval of the Ministry of Agriculture and Rural Affairs of the People’s Republic of China, Guangxi has initiated a dedicated fishing operation targeting purpleback flying squid in the South China Sea waters during the fishing moratorium in 2023. The data utilized in this study were obtained from this specialized fishing expedition. Fishery data were obtained from 33 fishing vessels operating in the waters around the **sha and Zhongsha islands from May to July in 2023. All fishing vessels in the study used mask nets (Figure 1). From 16 May to 23 July, a total of 33 fishing vessels sequentially departed for the capture area and subsequently returned within the boundary defined by four points: 111°00′ E, 17°00′ N; 115°00′ E, 17°00′ N; 115°00′ E, 12°00′ N; and 111°00′ E, 12°00′ N (Figure 2). The dataset constructed for the study comprised information regarding the duration of operation, longitude and latitude of operation, and catch data. The geographical coordinates of sites where fishing vessels were productive were gridded at a resolution of 0.25° (longitude × latitude). The timing of this productivity was recorded as late May, early June, middle June, late June, early July, and middle July.

2.1.2. Environmental Data

The selected marine environmental factors were SST, SSS, Chla, SSH, and MLD. Daily SST, Chla, and SSH data from 20 May to 20 July 2023 were obtained from the Ocean Color (http://oceanwatch.pifsc.noaa.gov/thredds/catalog.html; accessed on 1 September 2023). The daily SSS and MLD data from 20 May to 20 July 2023 were obtained from the Global ARMOR3D L4 dataset, which was downloaded from the Copernicus Marine Environment Monitoring Service (CMEMS) (https://marine.copernicus.eu/access-data; accessed on 1 September 2023). The environmental data covered the entire fishery area, and all environmental data were averaged over six time periods in late May, early June, middle June, late June, early July, and middle July and were calculated on a ten-day interval basis. The spatial resolution was converted to 0.25° (longitude × latitude) to match the fishery data.

2.2. Data Analyses

2.2.1. Calculations of the Catch per Unit Effort (CPUE) and Barycenter of the Fishing Ground

Fishery data statistics included fishing date (month and day), fishing location (longitude and latitude), daily catch (tons), and fishing effort (days). The timing of this catch was broken down into late May, early June, middle June, late June, early July, and middle July. Catch per unit effort (CPUE) can be used as an indicator to reflect the abundance of fishery resources [10,15]. In this study, CPUE was calculated by dividing the total catch of all vessels in different time periods by the total number of operating days of all fishing vessels in that time period, with the resulting value expressed in kg/day/boat. Equation (1) was calculated as follows. Purpleback flying squid catch locations were gridded separately at a resolution of 0.25° × 0.25°, with gridded CUPE data subsequently matched with environmental factors for a correlation analysis.

$${CPUE}_{mij}=\frac{\sum {Catch}_{mij}}{\sum {Times}_{mij}}$$

(1)

where $\sum {Catch}_{mij}$ is the total catch of all fishing vessels in a fishing area. $\sum {Times}_{mij}$ is the total number of operating days of all fishing vessels in a fishing area. i, j, and m are the longitude, latitude, and month, respectively.

Additionally, to reflect the variations in the purpleback flying squid fishery at different time periods, the barycenter of the fishing ground at different time periods was calculated. Longitude and latitude were calculated as follows:

$${Lon}_0=\frac{\sum _{i=1}^k{(C}_i\times {Lon}_i)}{\sum _{i=1}^k{C}_i}$$

(2)

$${Lat}_0=\frac{\sum _{i=1}^k{(C}_i\times {Lat}_i)}{\sum _{i=1}^k{C}_i}$$

(3)

where Lon₀ and Lat₀ are the longitude and latitude of the barycenter of the fishing ground, i is the number of grids, C_i is the catch of the ith grid, and Lon_i and Lat_i are the longitude and latitude of each of the ith grids, respectively.

2.2.2. Catch and CPUE Fitted in Relation to Environmental Factors

We used a GFA to analyze the relationships between the purpleback flying squid catch, CPUE, and multiple environmental factors and then derived the significance and gradient response of their effects on the catch and CPUE. Gradient forests are based on random forests and are used to determine the relationships between multivariate response variables and multivariate predictor variables [20]. In the multiple regression tree, the response variables are divided into two groups at specific predictor variable values to achieve maximum homogeneity, i.e., the sum of squares of deviation (i.e., impurity) within each group is minimized. The grou** method is recursive until the group contains the least number of response variables. In contrast, random forests are collections of multiple regression trees that aggregate the results of a large number of regression trees and use a self-help method to determine the optimal grou** method [21]. The GFA builds on the above methods by providing goodness-of-fit values for each response variable, as well as the importance of the predictor variables weighted according to the goodness of fit [20]. The gradient forest provides an R² value for each response variable and the standardized importance I_fp for each predictor variable. The standardized importance is the increase in the mean squared prediction error in the out-of-bag (OOB) samples when the predictor variables are randomly ordered or disrupted. Larger I_fp values indicate stronger predictor variables that have a more substantial impact on the model’s predictive performance. The importance of the split values along the predictor gradient reflects the relative change in the response variable. In the GFA analysis conducted in this study, the measure of resource change included catch and CPUE data, and the measure of environmental change included the environmental variable datasets (SST, SSS, Chla, SSH, and MLD). The GFA was performed for 1000 runs to obtain the mean and standard deviation (sd) of the R² value. The results with the highest R² were used to derive I_fp and other statistical parameters. The GFA was conducted using the “gradientForest” package in R4.1.0.

Additionally, we used a GAM to further analyze the CPUE data. The CPUE data of different periods were extracted and matched with the SST, SSS, Chla, SSH, and MLD of the corresponding time periods. The factors characterized by multicollinearity should be screened before model construction. The variance inflation factor (VIF) was applied to screen for multicollinearity characteristics among the explanatory variables added to the model. A value of VIF > 3 is generally considered to represent multicollinearity [22]. The selected factors based on the results of the VIF can all be added to the GAM for analysis (Table S1). The GAM was introduced to analyze the most influential factors affecting the distribution of the fishery with the purpleback flying squid CPUE as the response variable and SST, SSS, Chla, SSH, MLD, and season as the explanatory variables. Season was selected as the factor variable, with the final expression of the GAM model being as follows [23]:

$$Y=\alpha +{\sum }_{i=1}^k{f}_i\left(x_i\right)+\epsilon$$

(4)

where Y is the CPUE of the purpleback flying squid; x_i is the explanatory variable; and five environmental variables (SST, SSS, Chla, SSH, and MLD) were selected. Furthermore, seasonal variables were also added to the model as factor variables, with a being the function intercept. f_i(x_i) is the smoothing function, and ε denotes the residuals. The error functions for the model analysis were all normally distributed, and a natural logarithmic connection function was used in the model. CPUE was log-transformed by adding a constant term before being incorporated into the model (Figure S1). The fitting and validation of the above models were conducted using the R4.1.0 software, and the construction and testing of the GAM were completed using the “mgcv” package [24,25].

3. Results

3.1. Variations in Catch and CPUE

The total catch of purpleback flying squid amounted to 1345.37 tons, with the peak catch occurring in middle June at 382.02 tons, followed by 237.77 tons in early June. Subsequently, catches of 202.34, 191.64, and 176.50 tons were recorded in middle July, late May, and late June, respectively (Figure 3). The CPUE exhibited an increasing trend from late May to middle July, with an average of 1.22 t/day × boat. The highest CPUE was observed in middle July (3.82 t/day × boat), followed by 2.15 t/day × boat in early July. In late June, middle June, early June, and late May, the CPUE values were recorded as 1.44 t/day × boat, 1.22 t/day × boat, 0.98 t/day × boat, and 0.63 t/day × boat for these respective time periods (Figure 3). As shown in Figure 4 and Figure 5, distinct spatiotemporal variations were observed in catch and CPUE. In late May, a high CPUE was primarily concentrated in the western and southern waters of the ** methods are commonly used to capture chub mackerel, with higher catch rates expected when the thermocline is shallower and fish remain active above it. However, when the water temperature below the thermocline is suitable for fish, it becomes difficult for light to penetrate the thermocline and reach the upper water column. Moreover, a deeper MLD will result in an expanded range of depths that are conducive for fish habitats, consequently impacting the efficacy of light-induced fish trap**. The GAM results revealed that the MLD within the fishery area was relatively shallow, with a significant negative correlation observed between the CPUE of purpleback flying squid and MLD (Figure 9). Purpleback flying squid can be captured through light trap** methods, and its distribution may therefore be correlated with the MLD.

4.3. Future Work

Oceanographic conditions exert a direct influence on the survival, distribution, recruitment, and migration patterns of cephalopods. Understanding the impact of the marine environment on the resource of purpleback flying squid is essential for effective and sustainable management of this marine resource. Remote sensing of the marine environment is widely used in studies of sustainable fisheries development and fishing efficiency. Under appropriate conditions, satellite remote sensing can provide extensive and high-resolution information regarding the sea surface environment. This is conducive to studying fisheries oceanography, identifying central fishing grounds, and facilitating the work of scientific researchers, fisheries production workers, and fisheries administrators. Marine remote sensing inversion data for SST and Chla have been extensively applied in fisheries, while environmental conditions such as SSH anomalies, SSS, photosynthetically active radiation, and seawater surface wind speed have gradually gained prominence in recent years, providing a more comprehensive understanding of the impact of the marine environment on fishery resources. In this study, we exclusively utilized data on the distribution of purpleback flying squid during a fishing moratorium of one voyage, which entails certain limitations. Subsequent research should aim to continue the specialized fishing efforts in order to enhance our understanding of the purpleback flying squid’s distribution during this specific period. As it is a pelagic fish species, the distribution of the purpleback flying squid is closely related to the environmental factors in the upper ocean, however. It is a species with obvious vertical movement habits, and there is indeed a certain limitation in considering only the effects of surface environmental factors, and the effects of deeper environmental factors may be considered in further analysis. Further improvements are also needed in model selection for the application of catch data, perhaps taking into account the effect of fishing vessels or fishermen as a random effect on the distribution of yields. Additionally, it is crucial to continuously monitor the frequent occurrence of unusual weather events (e.g., the El Niño–Southern Oscillation (ENSO)). By making these improvements, it may be possible to explore the distribution of the purpleback flying squid more comprehensively, which may provide a reasonable theoretical basis for accurate prediction of the purpleback flying squid fishery.

5. Conclusions

In this study, we employed both GFA and GAM to investigate the distribution patterns of purpleback flying squid in the **sha and Zhongsha Islands during the fishing moratorium. The findings revealed distinct seasonal variations in the distribution of purpleback flying squid, with temperature and longitude identified as significant influencing factors. Over the past few decades, increased fishing pressure has been observed in northern waters of the SCS, resulting in a significant decline in traditional economic fish resources. However, despite an estimated resource potential exceeding one million tons for purpleback flying squid off the SCS, current catches remain far below sustainable levels. Given its short life cycle and vulnerability to climate and marine environmental changes, it is crucial to integrate multiple environmental variables for accurate prediction of purpleback flying squid fisheries’ distribution characteristics, which is a key consideration for sustainable exploitation of this valuable resource.