Journal of Oceanology and Limnology   2021, Vol. 39 issue(2): 770-783     PDF       
http://dx.doi.org/10.1007/s00343-020-9296-1
Institute of Oceanology, Chinese Academy of Sciences
0

Article Information

SUN Chunyang, WANG Yingbin
Impacts of sampling design on the abundance estimation of Portunus trituberculatus using crab pots
Journal of Oceanology and Limnology, 39(2): 770-783
http://dx.doi.org/10.1007/s00343-020-9296-1

Article History

Received Nov. 8, 2019
accepted in principle Dec. 23, 2019
accepted for publication Feb. 23, 2020
Impacts of sampling design on the abundance estimation of Portunus trituberculatus using crab pots
Chunyang SUN, Yingbin WANG     
School of Fisheries, Zhejiang Ocean University, Zhoushan 316022, China
Abstract: Fishery-independent surveys can provide high-quality data and support fishery assessment and management. Optimization of sampling design is crucial to increase the quality of fishery surveys. Crab pots are important fishing gears used to catch crabs. We analyzed the impacts of sampling design of crab pots on the abundance of Portunus trituberculatus in the Changjiang (Yangtze) River estuary to the Hangzhou Bay and its adjacent waters in East China Sea. The crab pots were cylindrical, 240 mm in height and 600 mm in diameter of the iron ring. Our sampling designs (including fixed-station sampling, simple random sampling, stratified fixed-station sampling, and stratified random sampling), three number of stations (9, 16, and 24), and three numbers of crab pots (500, 1 000, and 3 000) were simulated and compared with the "true" abundance that obtained from bottom trawl surveys in the study area in 2007. The scenarios with 16 stations were set in stratification as a control group for comparison with unstratified designs. Results show that simple random sampling can obtain more stable results than fixed-station sampling in the abundance estimation of P. trituberculatus. In addition, stratified sampling resulted in more accurate abundance than unstratified sampling. The accuracy of the simulated results improved with the increase of the number of stations. No remarkable differences in the results were found among the scenarios of different number of crab pots at each station. However, resource-intensive areas exerted great impacts on simulation results. Thus, prior information or pre-survey results about resource abundance and density distribution are necessary. This study may serve as a reference for future sampling designs of crab pots of P. trituberculatus and other species.
Keywords: sampling design    number of sampling station    sampling time    Portunus trituberculatus    abundance estimation    crab pot    
1 INTRODUCTION

Fishery survey data are the basis of stock assessment and management (Jardim and Ribeiro, 2007). The design of sampling schemes has received increasing attention because of the demand for high-quality data of fishery resources (Chen, 1996; Cao et al., 2014). However, limited funds, time, and complex marine geological conditions are frequently encountered problems in resource surveys (Barbraud et al., 2012; Dorner et al., 2013; Cao et al., 2014). Simulations using computers for sampling design can effectively avoid the above problems because they can save cost while ensuring the complete use of data, maximize the investigation benefit (Simmonds and Fryer, 1996; Liu et al., 2009), and give the optimal option by comparing different sampling designs.

Scientific survey design can provide information about the target population, such as resource abundance and distribution at a certain spatial-temporal scale (Jardim and Ribeiro, 2007). In 2017, the Ministry of Agriculture selected Zhejiang North Fishing Ground as a pilot area to carry out the catch quota management (CQM) of Portunustrituberculatus (Anonymous, 2018). CQM is currently an advanced measure in fishery management. Based on the scientific monitoring and assessment of fishery resources, a total allowable catch (TAC) of target species over a certain period is determined and assigned (Huang and Huang, 2002). Thus, survey data are the basic factor for determining the appropriate TAC.

Portunus trituberculatus is an important coastal economy species in China (Song et al., 2006). The total catch of P. trituberculatus in China exceeded 500 thousand tons in 2016 (Editorial Committee of China Fishery Statistical Yearbook, 2017). Zhejiang is the province with the highest yield of P. trituberculatus, which accounts for about 35% of the total catch in China of this species (Editorial Committee of China Fishery Statistical Yearbook, 2019). The catch of P. trituberculatus in Zhejiang Province increased about 30% during the past 10 years (Sun, 2018). The annual TAC of P. trituberculatus in the pilot area was determined by resource survey. Therefore, a scientific and reasonable sampling plan for crab pots is crucial.

Fixed-station sampling is often used in China's fishery resource survey. By investigating resources at fixed stations, this method can compare the dynamic changes of resources in different years (Zhao et al., 2014; Wang et al., 2018). Simple random sampling is the basis of other sampling method and is often used as the standard to compare with other sampling methods (Jin et al., 2008; Liu, 2012). Stratified random sampling is supported by the theory of classical statistics (Cochran, 1977; Manly et al., 2002; Miller et al., 2007), which is widely used in the scientific investigation of fishery resources because it can make selected samples more representative and improve the estimation accuracy of parameters (Smith and Gavaris, 1993; Gou, 2005; Xu et al., 2015).

Crab pots are passive fishing gears with the advantages of simple structure, low cost, convenient operation, high catch, and less damage to the ecological environment (Lin et al., 2013; Zhang et al., 2015). Therefore, crab pots are used worldwide and are very popular in Europe, America, Australia, New Zealand, South Korea, Japan, and other countries (Watanabe and Yamasaki, 1999; Stevens et al., 2000). For example, almost all commercial crab fishermen in Louisiana and the Gulf Coast of Mexico use crab pots (Yang, 1999). Crab pot fishery in the East China Sea began in 1980–1990 and was introduced from South Korea and Taiwan of China to Zhejiang, Jiangsu, and Fujian Provinces. The crab pot fishery flourished in the early 1990s, and has now become an important form of fishery in the East China Sea (Zhang et al., 2010b).

Despite the above advantages, the shortcomings of using crab pots in survey methods are obvious. The soak time of pots can produce biased simulation results for the catch of Dungeness crabs (Miller, 1979; Smith and Jamieson, 1989). The difference among seasons can also affect the estimation of crab abundance (Bennett, 1974). Previous studies lacked the research on the algorithm of crab pots capturing P. trituberculatus.

In the current study, we conducted a pre-experiment on soaking time. Then, we assessed four sampling designs (including fixed-station sampling (FS), simple random sampling (SR), stratified fixed-station sampling (SFS), and stratified random sampling (SRS) designs) in the Changjiang (Yangtze) River estuary to the Hangzhou Bay and its adjacent waters of East China Sea for four seasons using an independent algorithm that we established for the simulation of P. trituberculatus capture. The goals were to establish an algorithm to calculate the probability that each P. trituberculatus could be caught in the sampling area to estimate the abundance of P. trituberculatus in the study area; to compare the performances of different sampling designs (fixed-station sampling and simple random sampling, stratified sampling and unstratified sampling) in the study area; to compare the effectiveness of different numbers of stations when estimating the abundance of P. trituberculatus; and to compare the impact of different numbers of crab pots on sampling results.

2 MATERIAL AND METHOD 2.1 Study area and data collection period

The study area was the Changjiang River estuary to the Hangzhou Bay and adjacent area (29°N-32°N, 120°E-125°E), which is the pilot water of the CQM of P. trituberculatus (Fig. 1). The water depth of the study area was between 20 and 50 m (Su, 2005). Fishery resources are abundant in this area, which is a traditional fishing ground in Zhejiang, Jiangsu, Fujian, Shanghai, and Taiwan, China (Ding et al., 1987; Zheng et al., 2003).

Fig.1 Study area and original sampling stations in 2007

The distribution and density data of P.trituberculatu were collected quarterly (February, May, August, and November) in 2007 from 34 sampling stations, which were identified through fixed-station sampling (Fig. 1). A single otter trawl vessel with main engine power of 50 kw was used to conduct the bottom trawl surveys. The towing speed was 2 knots and hauled 1 h at each sampling station. The effective open width of the sampling net was 15 m.

2.2 Dynamic distribution base map of Portunus trituberculatus

The "true" abundance of P. trituberculatus was estimated using the sweep area method based on the data collected from 34 stations in the study area in the four seasons in 2007 (Han et al., 2019). Then, the survey area was divided into 850 sampling grids of about 6'×6', and the Kriging interpolation method was used to interpolate the unknown elements in the whole study area (Cressie, 1993; Rivoirard et al., 2008; Liu, 2012). The abundance of P. trituberculatus varied greatly among in season and in region for breeding, feeding, and overwintering (Wu et al., 2016). The base maps of resource distribution of the four seasons were produced and then used as the base of the sampling design for the next work (Petitgas, 2010; Pokhrel et al., 2013) (Fig. 2). P. trituberculatus individuals were allowed to move in the actual situation rather than fixed to allow predation, avoid predators, and adapt to the external environment (salinity, current, temperature, etc.) (Cheng et al., 2012; Ding et al., 2014; Sun, 2018). In addition, each sampling scenario was repeated 1 000 times. The fixed resource distribution base map not only interrupts the actual situation but also makes repeated simulation work meaningless. The principle of the individual distribution on the base map was that the P. trituberculatus abundance within each sampling unit was known. The corresponding coordinate values were randomly generated inside the unit area when the unit was selected, representing each location of P. trituberculatus. New coordinates were randomly generated in each sampling unit and each sampling scenario was repeated 1 000 times by computers.

Fig.2 Base map of dynamic distribution of P. trituberculatus resources in each season
2.3 Specification and placement of crab pots

The fishing gear and fishing method of crab pots were determined according to the investigation in Daishan County of Zhejiang Province (Lin et al., 2013). The crab pots were cylindrical with 240 mm height, and the diameter of the iron ring was about 600 mm. All crab pots had the same specifications. As a passive fishing gear (Tang and Wu, 1990; Zhang et al., 2010a), the crab pot has an upper limit of catching capacity. No relevant literature and independent experiments were available about the maximum catch of crab pots as support. Therefore, experts and fishermen were consulted, and 10 was considered as the maximum number of catch for each pot (personal communication).

In our simulation work, the crab pots were placed in the north-south direction (Lin et al., 2013). The fishing boat travelled at a constant speed (8–10 kn) in a constant direction, and a group of crab pots was placed in the water in turn (Lin et al., 2013). In total, 500, 1 000, and 3 000 crab pots were selected in the simulation work. Then, 500 and 1 000 crab pots were connected to one main rope. The interval between two crab pots was 8 m. Furthermore, 3 000 crab pots were divided into three parallel lines with 1 000 crab pots on each line, and an interval of 0.2 n mile was assumed between two strings.

Crabs approach the pots along the odor field gradient of the bait, and the bait in the pot plays an important role in the induction (Chapman and Smith, 1978). The trapping effect of bait is also the biological basis of the algorithm established in our research. In our simulation work, the dates when the pots were placed into the sea were consistent with the actual sampling work for the four seasons, which was performed at the beginning of February, May, August, and November.

The soaking time of crab pot in the sea has a great impact on the quantity of capture (Hong et al., 1999). Three time options (6, 10, and 16 h) with nine sampling stations and 1 000 crab pots were set for simulation analyses to determine the appropriate soaking time (Lin et al., 2013; Zhang et al., 2015). The optimal soaking time was chosen and then used in each simulation scenario.

2.4 Sampling design

Four sampling designs, including FS, SR, SFS, and SRS, were selected for the simulation study. Three numbers of station (9, 16, and 24) were set to evaluate the influence of station number on the estimated results (Fig. 3).

Fig.3 Layout of 9, 16, and 24 stations in fixed-station sampling and 16 stations in stratified fixed-station sampling (strata A, B, C)

1) FS (fixed-station sampling): 9, 16, and 24 stations out of the 850 potential sampling stations were selected (Li et al., 2015). One sampling station was set in the estuary to ensure the sampling can better represent the study area because of the special environment of hydrology and ecological characteristics in this area.

2) SR (simple random sampling): 9, 16, and 24 stations were randomly selected from all 850 potential stations without replacement in each simulation.

3) SFS (stratified fixed-station sampling): Based on the resource distribution and isobaths, three strata (A, B, and C) were divided and the number of sampling stations in each stratum was allocated according to the stratum's area and the variance of density of resource in each stratum. The scenario of 16 stations was selected as a control group of stratified sampling, and 10, 4, and 2 sampling stations were chosen for strata A, B, and C, respectively.

4) SRS (stratified random sampling): the situation of 16 stations was conducted, and the locations of sampling station (10, 4, and 2) in each stratum (A, B, and C) were selected randomly without replacement.

Three numbers of crab pots (500, 1 000, and 3 000) were assumed to put in each station to evaluate the impact of the number of crab pots on the estimated results.

2.5 Algorithm and performance indices

An algorithm calculating the capture probability of crab pots and the number of P. trituberculatus that can be caught by the crab pots was built. The schematic diagram of the crab pot catching P. trituberculatus shows the principle of fishing (Fig. 4). The formula (Eq.1) is as follows:

    (1)
Fig.4 Schematic diagram of the crab pot catching P. trituberculatus

where P is the probability that each P. trituberculatus could be caught; S is the furthest distance that P. trituberculatus can move during the crab pot sampling time; x and y represent the position of each P. trituberculatus that can be caught within the range of the crab pots. The crab pot capture area was divided into one rectangular area and two semicircular areas (Fig. 4). When the individual's location is in the rectangular area, equation (F) should be selected; when the individual's location is in the semicircular areas, then equation (M) should be selected. In a simulated sampling process, all the calculated P values were added to obtain the nominal probability of P. trituberculatus that can be caught by this sampling design.

The relative estimation error (REE) and relative bias (RB) were calculated to evaluate the results of abundance for different sampling scenarios (Eqs.2 & 3).

    (2)
    (3)

where Yiestimated is the ith estimated value of abundance, Ytrue is the "true" abundance value of each season used to make the base map of resource distribution, and R is the number of simulations. REE is used to measure the accuracy and precision of simulation results (Chen, 1996; Wang and Jiao, 2015), whereas RB is used to evaluate the accuracy of estimates (Paloheimo and Chen, 1996; Jiao et al., 2004). The RB values can also indicate whether a survey design tends to underestimate or overestimate the "true" values (Xu et al., 2015).

2.6 Simulation procedure

A simulation framework for sampling design was developed (Fig. 5). Four sampling designs (FS, SR, SFS, and SRS) with three numbers of sampling stations (9, 16, and 24) and crab pots (500, 1 000, and 3 000) composed 96 scenarios (Table 1). The sampling area of the crab pots is located inside the sampling unit of the dynamic resource distribution base map. Individuals of P. trituberculatus in the crab pots sampling area were selected, and the probability that each P. trituberculatus could be captured was calculated using the established algorithm. The sum of the values was considered the amount of P. trituberculatus resources that this sampling scheme could capture. Then, the sweep area method was used to simulate the abundance of P. trituberculatus in the study area. Each sampling scenario was repeated 1000 times. The performance indices (REE, RB) were calculated to compare the simulation effect of different sampling scenarios and then choose the reasonably sampling designs.

Fig.5 Flowchart of the simulation study summarizing the simulation framework for the reasonable sampling station designs for fishery-independent survey using crab pot
Table 1 Different scenario schemes for the simulation study
3 RESULT 3.1 Determination of the soaking time

When the soaking time of crab pots was prolonged from 6 h to 10 h, the REE and RB values gradually decreased. For example, the REE value in winter decreased from 35.07% to 30.50%, and the RB value decreased from 35.05% to 30.47% (Fig. 6). When the soaking time was prolonged from 10 h to 16 h, the REE and RB values increased. For example, the REE value in winter increased from 30.50% to 37.11%, and the RB value increased from 30.47% to 37.08% (Fig. 6). The best simulation results occurred in 10 h. Thus in the current study, 10 h was selected as the soaking time at each sampling station during the simulation studies.

Fig.6 Relative estimation error (REE) and relative bias (RB) of different sampling times a. 9 stations, 1 000 crab pots; b. 9 stations, 1 000 crab pots. 6 h, 10 h, and 16 h represent the sampling time.
3.2 Comparison of different sampling designs

The REE values of FS and SR varied from 4.21% to 113.91% and from 31.90% to 65.11%, respectively (Fig. 7). Overall, FS had a wider range of REE values. The REE values of SFS and SRS varied from 1.37% to 44.96% and from 31.61% to 56.71%, respectively (Fig. 7). The REEs of SFS and SRS were relatively more stable and less variable than those of FS and SR. Therefore, stratified sampling resulted in less deviation than unstratified sampling. The corresponding REE values of fixed-station sampling (FS and SFS) were smaller than those of random sampling (SR and SRS), except for the scenarios with nine sampling stations (Fig. 7). For the RB values, we obtained similar results as the REE values, i.e., stratified sampling worked better than unstratified sampling. The RB values of SFS and SRS varied from -44.96% to 2.72% and from -54.02% to -4.67%, respectively. Meanwhile, the RB values of FS and SR varied from -69.97% to 113.76% and from -60.63% to -8.20%, respectively (Fig. 8).

Fig.7 Relative estimation error (REE) of different sampling designs (totally 96 scenarios) a. FS, 500; b. FS, 1 000; c. FS, 3 000; d. SR, 500. e. SR, 1 000; f. SR, 3 000; g. SFS, 16 stations; h. SRS, 16 stations. 9, 16, and 24 represent the number of sampling stations; 500, 1 000, and 3 000 represent the number of crab pots.
Fig.8 Relative bias (RB) for different sampling designs (totally 96 scenarios) a. FS, 500; b. FS, 1 000; c. FS, 3 000; d. SR, 500. e. SR, 1 000; f. SR, 3 000; g. SFS, 16 stations; h. SRS, 16 stations. 9, 16, and 24 represent the number of sampling stations; 500, 1 000, and 3 000 represent the number of crab pots.
3.3 Comparison of the number of sampling stations

The number of sampling stations exerted a great impact on results. The REE values of FS decreased with the increase in station numbers from 9 to 24 except for the scenario of 24 sampling stations in winter (Fig. 7). However, the decrease rate was not uniform. In spring and autumn, for example, when the number of stations was increased from 9 to 16, REE decreased rapidly (spring: from 113.91% to 5.88%; Autumn: from 69.98% to 53.35). When the number was increased from 16 to 24, REE slightly decreased (spring: from 5.88% to 4.44%; Autumn: from 53.35% to 51.75%) (Fig. 7ac). The REE values of SR decreased with the increase in station number. The decrease rates were stable for all the cases (spring: 55.75% to 44.62% to 38.03%) (Fig. 7df). Overall, the RB values of FS and SR showed a similar trend with REE values (Fig. 8). The RB values of FS gradually approached 0 with the increase in station numbers from 9 to 24, except for the scenarios of 24 sampling stations in winter (Fig. 8ac). In SR, all RB values were less than 0. The RB values of autumn (-60.63%) were much lower than those of the other seasons, and those of winter (-8.20%) were closer to 0 (Fig. 8df).

3.4 Comparison of the number of crab pots

The REE values gradually decreased and RB values gradually increased as the number of crab pots was increased, but the effect of the number of crab pots on the results was not remarkable. The most obvious change trend occurred in SR. For the scenarios of nine stations in summer, the RB value increased from -24.43% to -18.83%. For the most cases, when the number of crab pots was increased from 500 to 3 000, the range of REE and RB values varied by less than 2%.

3.5 Effect of density variance of spatial distribution

Resource density and distribution exerted remarkable effects on the sampling results, especially in FS. In the current study, the density variances of the P. trituberculatus abundance in spring, summer, autumn, and winter were 38.06, 85.68, 441.11, and 8.85, respectively. The resource intensive areas appeared in spring, summer, and autumn. In winter, the resource distribution was relatively even comparing with the three other seasons. In spring, the estimated abundance values for the scenario of nine sampling stations were much higher than the "true" value (REE=113.91%) because one of the selected sampling stations was located in the center of resource-intensive areas. All the RB values in SR and SRS were negative because of the limitation of crab pot capture capacity (Fig. 8). The RB values in autumn were the worst, whereas those in winter were closer to 0, and the difference between spring and summer was not remarkable (Fig. 8).

4 DISCUSSION 4.1 Discussion of soaking time

The soaking time of crab pots in the sea plays an important role in P. trituberculatus resource survey. The current study results indicate that 10 h of soaking time is reasonable. Too long or too short soaking time reduces the accuracy and precision of the estimated results. In specific, a long time is needed for the bait to attract the crab into the crab pot when the pot is placed in the sea, whereas the catching efficiency of the crab pot cannot be fully utilized in a short period (Tang and Wu, 1990). The crabs in the pot reached a certain amount with time. The catch efficiency of the crab pot decreased (Wu, 1996).

The soaking time of crab pots is a factor affecting the catches of crab. Zhang et al. (2015) indicated that the soaking time of crab pots is more reasonable between 8 and 11 h. Hong et al. (1999) found that the catch of crabs increases rapidly after 1 h of crab pot soaking, peaks at 8–9 h, and then decreases with time. They deemed 9–10 h as the optimal choice. Their conclusions are consistent with our study. However, Robertson (1989) and Öndes et al. (2017) reported that the catch per unit effort of Scylla serrata and Cancer pagurus in the pot fishery does not increase remarkably with the soaking time. This phenomenon may be due to the differences in species, fishing gear, and fishing methods.

4.2 Sampling design selection

The REE and RB values of FS varied greatly between seasons and station number (Fig. 7ac; Fig. 8ac). Reasonable station selection can produce accurate simulation results. Therefore, the selection of sampling stations in FS is particularly important. If the resource distribution is stable, FS is reasonable because it can reflect the annual change of the resource (Wang et al., 2018; Sun and Wang, 2019). The distribution of living resources may be extremely uneven and constantly moving in different seasons. Thus, FS should be the final choice if other sampling designs can be made.

The REE values of SR were stable and gradually decreased as the number of stations was increased when compared with FS because SR can reduce the intervention of subjective factors and make the results more objective (Figs. 7 & 8). However, the simulation results of SR showed a large deviation because of the randomness of simple random sampling, and almost all the outliers appeared in SR (Fig. 9). The deviation of FS was small because FS only repeated the sampling procedure 1 000 times at the fixed stations (Fig. 8). A previous study obtained similar simulation tests for bottom trawl and obtained the same conclusion that the deviation of SR is large (Sun and Wang, 2019). In a word, SR is more suitable for areas with uniform distribution of resources or exploring resources in the study area.

Fig.9 Box plot of the P. trituberculatus abundance simulated 1 000 times for each season in all simulated sampling designs S1–S24 corresponding to scenario number in Table 1. "*" represent outlier. The Y-axis reference line represents the true value of the resources of P. trituberculatus in each season.

In this study, the stratified sampling (SFS, SRS) had advantage over unstratified sampling (FS, SR). Many relevant studies obtained the same conclusion that stratified sampling works better than unstratified sampling under the same conditions because it can obtain more accurate results (Yuan et al., 2011; Liu, 2012; Zhao et al., 2014). In stratified sampling, the definition of strata directly affects the quality of the estimated abundance, and proper strata can yield good results, which would otherwise be worse than those of unstratified sampling (Smith and Gavaris, 1993; Li, 2010; Yuan et al., 2011). In our simulation, the results of stratified sampling with 16 stations were not only better than those of unstratified sampling with 16 stations but also better than those of 24 stations, which was more obvious in the scenarios of FS (Fig. 7). Stratified sampling can reasonably allocate stations, make different sampling efficiencies among different strata, and thus improve sampling accuracy (Gavaris and Smith, 1987; Chen et al., 2006).

4.3 Determination of the station number

Many studies have also shown the same conclusion that more stations can obtain higher-quality survey data (Lai and Kimura, 2002; Sun and Wang, 2019). However, too many sampling stations not only increase the funding and time but also generate redundant information, which reduces the precision and accuracy of estimated results (Zhao et al., 2014). Moreover, oversampling damages the biological resources and ecological environment of the surveyed area (Conners and Schwager, 2002). The REE values of SR gradually decreased when the station number was increased from 9 to 24, and a slight difference was observed between seasons. However, large errors appeared in FS with nine sampling stations. The same results as above also appeared in our previous sampling design study on trawling (Sun and Wang, 2019). Therefore, 24 stations can obtain viable survey data in this study, and 16 stations are necessary if the fund and time are limited. Nine stations are feasible to SR, but 16 stations are necessary for FS. Abnormal results appeared in S9 in winter because the resources in winter were fewer than those in the three other seasons, and too many sampling stations generated redundant information.

4.4 Selection of the number of crab pots

Three different numbers of crab pots in each station (500, 1 000, and 3 000) were set to explore the impacts of the number of crab pots on the estimated abundance of P. trituberculatus and choose a reasonable number of crab pots. Other scholars, such as Zhang et al. (2015), Tang and Wu (1990), Ye (2001), and Lin et al. (2013), used 50, 300, 500, and 3 000 crab pots for resource surveys, respectively. According to the current study, the number of crab pots exerted no remarkable effect on the simulation results. Therefore, when using crab pots to investigate the resources of P. trituberculatus, a small amount of crab pots can still achieve similar results with a large number of crab pots. A small number of crab pots should be selected from the perspective of saving time and funding or protecting fishery resources and the environment. In addition, a small amount of crab pots can improve the safety of fishing boats (Zeng, 2016).

4.5 Density of resource

The distribution of fishery resources changes and individuals are not distributed in the entire sea area because of the influence of environmental and human factors (Zhang et al., 2017). Temporal factors are closely associated with resource production (Wang et al., 2012). In spring, summer, and autumn, P. trituberculatus has obvious resource-intensive areas, especially in autumn. In winter, the resources of P. trituberculatus are evenly distributed. Therefore, a certain proportion of zero value and high catch value appeared in the survey month, which affected the mean value and variance of the estimation results and then affected the accuracy of resource estimation.

The abundance of P. trituberculatus in autumn was much higher than that in other seasons but much lower than the "true" value in autumn (Fig. 9). This finding can be attributed to the fact that as a passive fishing gear, crab pot has the upper limit of its catching capacity. The external crabs will not enter the crab pot when the number of crabs in the pot reached the capture capacity (Tang and Wu, 1990). In addition, the freshness of the bait and the competitive behavior to the territory will also affect the catching ability of the crab pot (Wu, 1996). Therefore, when using crab pots for fishery resources survey, it is necessary to attach great importance to seasons with high resources or resource-intensive areas. In these cases, the limitations of crab pots may cause underestimation of resources.

The previous relevant studies neither used statistical methods nor involved the catches based on the probability algorithm. In the current study, we established a statistical algorithm to calculate the probability that each individual can be caught in the study area, which is more consistent with the actual situation. In addition, some scholars reported that fishery-specific factors, such as pot design, escape gaps, freshness of bait, and agonistic interactions among conspecifics inside and at the entrance of pots, influence the catchability of crabs in the pot fishery (Miller, 1979; Brown, 1982; Smith and Jamieson, 1989). Some of these factors will be considered in our future works.

5 DATA AVAILABILITY STATEMENT

All data generated and/or analyzed during this study cannot be shared at this time as the data also forms part of an ongoing study.

6 ACKNOWLEDGMENT

The authors are grateful for all scientific staff and crew for their assistance with data collection during all the surveys.

References
Anonymous. 2018. Piloting marine limit fishing and promoting the total control of fishery resourcesjIntroduction to the pilot work of marine quota fishing in the five provinces. China Fisheries, 9: 2-4. (in Chinese with English abstract)
Barbraud C, Rolland V, Jenouvrier S, Nevoux M, Delord K, Weimerskirch H. 2012. Effects of climate change and fisheries bycatch on southern Ocean seabirds: a review. Marine Ecology Progress Series, 454: 285-307. DOI:10.3354/meps09616
Bennett D B. 1974. The effects of pot immersion time on catches of crabs, Cancer pagurus L. and lobsters, Homarus gammarus (L.). ICES Journal of Marine Science, 35(3): 332-336. DOI:10.1093/icesjms/35.3.332
Brown C G. 1982. The effect of escape gaps on trap selectivity in the United Kingdom crab (Cancer pagurus L.) and lobster (Homarus gammarus (L.)) fisheries. ICES Journal of Marine Science, 40(2): 127-134. DOI:10.1093/icesjms/40.2.127
Cao J, Chen Y, Chang J H, Chen X. 2014. An evaluation of an inshore bottom trawl survey design for American lobster (Homarusamericanus) using computer simulations. Journal of Northwest Atlantic Fishery Science, 46: 27-39. DOI:10.2960/J.v46.m696
Chapman C J, Smith G L. 1978. Creel catches of crab, Cancer pagurus L. using different baits. ICES Journal of Marine Science, 38(2): 226-229. DOI:10.1093/icesjms/38.2.226
Chen Y, Sherman S, Wilson C, Sowles J, Kanaiwa M. 2006. A comparison of two fishery-independent survey programs used to define the population structure of American Lobster (Homarus americanus) in the Gulf of Maine. Fishery Bulletin, 104(2): 247-255.
Chen Y. 1996. A Monte Carlo study on impacts of the size of subsample catch on estimation of fish stock parameters. Fisheries Research, 26(3-4): 207-223. DOI:10.1016/0165-7836(95)00447-5
Cheng G B, Shi H L, Lou B, Mao G M, Zhan W, Xu D D, Xue B G. 2012. Biological characteristics and artificial propagation, culture technique for Portunustrituberculatus. Hebei Fisheries, (4): 59-61. (in Chinese with English abstract)
Cochran W G. 1977. Sampling Techniques. 3rd edn. John Wiley and Sons, New York.
Conners M E, Schwager S J. 2002. The use of adaptive cluster sampling for hydroacoustic surveys. ICES Journal of Marine Science, 59(6): 1314-1325. DOI:10.1006/jmsc.2002.1306
Cressie N A C. 1993. Statistics for Spatial Data. John Wiley and Sons, New York.
Ding R F, Yu L F, Yan Y M. 1987. Investigation and Division of Fishery Resources in the East China Sea. East China Normal University Press, Shanghai. (in Chinese)
Ding Z N, Xu Y J, Lin J H, Li J, Zhang G H. 2014. Effects of salinity on feeding behavior and growth characteristics of swimming crab Portunus trituberculatus. Ecological Science, 33(5): 899-903.
Dorner B, Holt K R, Peterman R M, Jordan C, Larsen D P, Olsen A R, Abdul-Aziz O I. 2013. Evaluating alternative methods for monitoring and estimating responses of salmon productivity in the North Pacific to future climatic change and other processes: a simulation study. Fisheries Research, 147: 10-23. DOI:10.1016/j.fishres.2013.03.017
Editorial Committee of China Fishery Statistical Yearbook. 2017. China Fishery Statistics Yearbook (2017). China Agriculture Press, Beijing. 224p. (in Chinese)
Editorial Committee of China Fishery Statistical Yearbook. 2019. China Fishery Statistical Yearbook (2019). China Agriculture Press, Beijing. (in Chinese)
Gavaris S, Smith S J. 1987. Effect of allocation and stratification strategies on precision of survey abundance estimates for Atlantic Cod (Gadus morhua) on the eastern Scotian Shelf. Journal of Northwest Atlantic Fishery Science, 7: 137-144. DOI:10.2960/J.v7.a16
Gou P H. 2005. Some considerations about stratified sampling. Statistical Research, (11): 16-17.
Han Q P, Shan X J, Wan R, Guan L S, Jin X S, Chen Y L, Wu Q. 2019. Spatiotemporal distribution and the estimated abundance indices of Larimichthys polyactis in winter in the Yellow Sea based on geostatistical delta-generalized linear mixed models. Journal of Fisheries of China, 43(7): 1603-1614.
Hong M J, Su X H, Feng S. 1999. A study on the structureal feature and the fishing operations of the fish traps. Journal of Fujian Fisheries, (3): 74-78.
Huang J L, Huang S L. 2002. A study on feasibility of implementing total allowable catch in Chinese EEZ. Modern Fisheries Information, 17(11): 3-6.
Jardim E, Ribeiro Jr P J. 2007. Geostatistical assessment of sampling designs for Portuguese bottom trawl surveys. Fisheries Research, 85(3): 239-247. DOI:10.1016/j.fishres.2007.02.014
Jiao Y, Chen Y, Schneider D, Wroblewski J. 2004. A simulation study of impacts of error structure on modeling stock-recruitment data using generalized linear models. Canadian Journal of Fisheries and Aquatic Sciences, 61(1): 122-133. DOI:10.1139/f03-149
Jin Y J, Du Z F, Jiang Y. 2008. Sampling Technique. 2nd edn. China Renmin University Press, Beijing. (in Chinese)
Lai H L, Kimura D K. 2002. Analyzing survey experiments having spatial variability with an application to a sea scallop fishing experiment. Fisheries Research, 56(3): 239-259. DOI:10.1016/S0165-7836(01)00323-X
Li B, Cao J, Chang J H, Wilson C, Chen Y. 2015. Evaluation of effectiveness of fixed-station sampling for monitoring American lobster settlement. North American Journal of Fisheries Management, 35(5): 942-957. DOI:10.1080/02755947.2015.1074961
Li J C. 2010. Application of Sampling Techniques. 2nd edn. Science Press, Beijing. (in Chinese)
Lin H B, Zheng J, Wang Y M. 2013. Investigation on the fishing gears and methods of crab pot in Daishan county of Zhejiang province. Journal of Zhejiang Ocean University (Natural Science), 32(4): 319-322.
Liu Y, Chen Y, Cheng J H. 2009. A comparative study of optimization methods and conventional methods for sampling design in fishery-independent surveys. ICES Journal of Marine Science, 66(9): 1873-1882. DOI:10.1093/icesjms/fsp157
Liu Y. 2012. Theoretical Study on the Sampling Methods of Survey for Fishery Stock Estimation. East China Normal University, Shanghai. (in Chinese English abstract)
Manly B F J, Akroyd J A M, Walshe K A R. 2002. Two‐phase stratified random surveys on multiple populations at multiple locations. New Zealand Journal of Marine and Freshwater Research, 36(3): 581-591. DOI:10.1080/00288330.2002.9517114
Miller R J. 1979. Saturation of crab traps: reduced entry and escapement. ICES Journal of Marine Science, 38(3): 338-345. DOI:10.1093/icesjms/38.3.338
Miller T J, Skalski J R, Ianelli J N. 2007. Optimizing a stratified sampling design when faced with multiple objectives. ICES Journal of Marine Science, 64(1): 97-109. DOI:10.1093/icesjms/fsl013
Öndes F, Emmerson J A, Kaiser M J, Murray L G, Kennington K. 2017. The catch characteristics and population structure of the brown crab (Cancer pagurus) fishery in the Isle of Man, Irish Sea. Journal of the Marine Biological Association of the United Kingdom, 99(1): 119-133.
Paloheimo J E, Chen Y. 1996. Estimating fish mortalities and cohort sizes. Canadian Journal of Fisheries and Aquatic Sciences, 53(7): 1572-1579. DOI:10.1139/f96-077
Petitgas P. 2001. Geostatistics in fisheries survey design and stock assessment: models, variances and applications. Fish and Fisheries, 2(3): 231-249. DOI:10.1046/j.1467-2960.2001.00047.x
Pokhrel R M, Kuwano J, Tachibana S. 2013. A kriging method of interpolation used to map liquefaction potential over alluvial ground. Engineering Geology, 152(1): 26-37. DOI:10.1016/j.enggeo.2012.10.003
Rivoirard J, Simmonds J, Foote K G et al. 2008. Geostatistics for estimating fish abundance. John Wiley and Sons, Oxford, UK.
Robertson W D. 1989. Factors affecting catches of the crab Scylla serrata (Forsk?l) (Decapoda: Portunidae) in baited traps: soak time, time of day and accessibility of the bait. Estuarine, Coastal and Shelf Science, 29(2): 161-170. DOI:10.1016/0272-7714(89)90005-X
Simmonds E J, Fryer R J. 1996. Which are better, random or systematic acoustic surveys? A simulation using North Sea herring as an example. ICES Journal of Marine Science, 53(1): 39-50. DOI:10.1006/jmsc.1996.0004
Smith B D, Jamieson G S. 1989. A model for standardizing Dungeness crab (Cancer magister) catch rates among traps which experienced different soak times. Canadian Journal of Fisheries and Aquatic Sciences, 46(9): 1600-1608. DOI:10.1139/f89-204
Smith S J, Gavaris S. 1993. Improving the precision of abundance estimates of Eastern Scotian Shelf Atlantic Cod from bottom trawl surveys. North American Journal of Fisheries Management, 13(1): 35-47. DOI:10.1577/1548-8675(1993)013<0035:ITPOAE>2.3.CO;2
Song H T, Yu C G, Xue L J, Yao G Z. 2006. East China Sea Economic Shrimp and Crabs. Ocean Press, Beijing. (in Chinese)
Stevens B G, Vining I, Byersdorfer S, Donaldson W. 2000. Ghost fishing by Tanner crab (Chionoecetes bairdi) pots off Kodiak, Alaska: pot density and catch per trap as determined from sidescan sonar and pot recovery data. Fishery Bulletin, 98(2): 389-399.
Su J L. 2005. Chinaos Offshore Hydrology. Ocean Press, Beijing. (in Chinese)
Sun C Y, Wang Y B. 2019. Impacts of the sampling design on the abundance index estimation of Portunustrituberculatus using bottom trawl. Acta Oceanologica Sinica. DOI:10.1007/s13131-020-1607-z
Sun J. 2018. The prediction model of recuitment of Portunus trituberculatus in northern of Zhejiang fishing ground. Zhejiang Ocean University, Danzhou. (in Chinese with English abstract)
Tang Y M, Wu C W. 1990. Fishing experiment by crab pot for exploitation in coastal waters. Journal of Zhejiang College of Fisheries, 9(1): 1-5. (in Chinese with English abstract)
Wang J Q, Tian S Q, Gao C X, Dai X J. 2018. Optimization of sample size for lake fish resources survey. Journal of Shanghai Ocean University, 27(2): 265-273. (in Chinese with English abstract)
Wang Y B, Jiao Y. 2015. Estimating time-based instantaneous total mortality rate based on the age-structured abundance index. Chinese Journal of Oceanology and Limnology, 33(3): 559-576. DOI:10.1007/s00343-015-4112-z
Wang Y B, Zheng J, Wang Y, Zheng X Z. 2012. Spatiotemporal factors affecting fish harvest and their use in estimating the monthly yield of single otter trawls in Putuo district of Zhoushan, China. Chinese Journal of Oceanology and Limnology, 30(4): 580-586. DOI:10.1007/s00343-012-1215-7
Watanabe T, Yamasaki S. 1999. Catch variation and the soak time of gear in the pot fishery for the red queen crab Chionoecetes japonicus. Nippon Suisan Gakkaishi, 65(4): 642-649. DOI:10.2331/suisan.65.642
Wu C W. 1996. Preliminary tests on a kind of crab pot for protecting resources. Marine Tieheries, (3): 114-116. (in Chinese with English abstract)
Wu Q, Wang J, Chen R S, Huang J X, Zuo T, Luan Q S, Jin X S. 2016. Biological characteristics, temporal-spatial distribution of Portunus trituberculatus and relationships between its density and impact factors in Laizhou Bay, Bohai Sea, China. Chinese Journal of Applied Ecology, 27(6): 1993-2001. (in Chinese with English abstract)
Xu B D, Ren Y P, Chen Y, Xue Y, Zhang C L, Wan R. 2015. Optimization of stratification scheme for a fishery-independent survey with multiple objectives. Acta Oceanologica Sinica, 34(12): 154-169. DOI:10.1007/s13131-015-0739-z
Yang L. 1999. Study on mesh size selectivity of American blue crab pot. Journal of Aquatic Technology, (3): 41-42. (in Chinese)
Ye X M. 2001. Crab cage operation and lifting equipment. Chinese Journal of Fishery Modernization, (4): 33-34: 33-34, 44. (in Chinese)
Yuan X W, Liu Y, Cheng J H. 2011. Error analysis on stratified sampling and its application in fishery statistics. Marine Fisheries, 33(1): 116-120. (in Chinese with English abstract)
Zeng J R. 2016. Analysis on the operation characteristics and safety of crab pot fishing vessels. China Water Transport, 16(2): 6-7, 78. (in Chinese)
Zhang H L, Lin X P, Xu H X. 2010a. Design and experiment of selective shuttle crab pot. Fishery Modernization, 37(4): 49-53. (in Chinese)
Zhang H L, Xu H X, Huang H L, Zhou Y D. 2010b. Selectivity of crab pot for Portunus trituberculatu in the East China Sea region. Journal of Fisheries of China, 34(8): 1277-1284. (in Chinese with English abstract)
Zhang J, Wang Z Q, Guan W B. 2015. Fishery biology of two crab species caught by offshore pots. Journal of Dalian Fisheries University, 30(5): 540-545. (in Chinese with English abstract)
Zhang Q Y, Hong W S, Chen S X. 2017. Population changes of large yellow croaker and Japanese hairtail in China's coastal waters and their conservation measures. Journal of Applied of Oceanography, 36(3): 438-445. (in Chinese)
Zhao J, Zhang S Y, Lin J, Zhou X J. 2014. A comparative study of different sampling designs in fish community estimation. Chinese Journal of Applied Ecology, 25(4): 1181-1187. (in Chinese with English abstract)
Zheng Y J, Chen X Z, Cheng J H, Wang Y L, Shen X Q, Chen W Z, Li C S. 2003. East China Sea Continental Shelf Biological Resources and Environment. Shanghai Science and Technology Press, Shanghai. (in Chinese)