The study area in the current study is the Changjiang River Estuary-Hangzhou Bay and the adjacent area. Fishery resources are abundant in this sea area, and it is traditional fishing ground for fishermen in Zhejiang, Jiangsu, Fujian and Shanghai as well as Taiwan Province (Department of Fisheries, Ministry of Agriculture, Animal Husbandry and Fisheries, 1987; Zheng, 2003). The pilot water for the CMQ of the P. trituberculatus is right in this area (Fig. 1).
The distribution and density data of P. trituberculatu obtained from the bottom trawl survey in the Changjiang River Estuary-Hangzhou Bay and its adjacent waters (29°–32°N, 120°–125°E) in 2007 were used to simulate the “true” situation. The survey was carried out quarterly (February, May, August and November), and a total of 34 stations were set (Fig. 1). The bottom trawl surveys were conducted using a single otter trawl vessel with main engine power of 50 kW. The towing speed was 2 knots, and hauled 1 h for each station. The effective open width of the sampling net was 15 m.
The abundance of P. trituberculatus in each season was calculated using the sweep area method based on the data collected at 34 stations in the Changjiang River Estuary-Hangzhou Bay and its adjacent area in 2007. Kriging interpolation method was used to interpolate the unknown elements in the whole study area and get the dynamic base map of resource distribution (Cressie, 1993; Rivoirard et al., 2008; Liu, 2012), which was used as the base of the next sampling design (Petitgas, 2010; Pokhrel et al., 2013). In the current study, the survey area was divided into 850 sampling grids of 6′×6′. The abundance of P. trituberculatus varies greatly among different seasons for feeding, breeding, overwintering, etc. (Wu et al., 2016) (Fig. 2).
Additionally, the location of the P. trituberculatus individual is not fixed in the actual situation, it will keep moving for different purposes such as predation, avoiding predators and the stimulation of external environment (ocean, current, temperature, salinity, etc.) (Cheng et al., 2012; Ding et al., 2014; Sun, 2018). The resource abundance in different seasons and different regions within a season are different. The abundance in each season and each sampling unit was unchanged, but the location of each individual in each sampling unit was constantly changing randomly.
Four different sampling methods (FS, SR, SFS and SRS) were selected for the simulation study. Three numbers of stations (9, 16 and 24) were set to evaluate the influence of station number on the estimate results (Fig. 3).
Figure 3. Layout of 9 (a), 16 (b), 24 (c) stations in fixed-station sampling and 16 stations (d) in stratified fixed-station sampling (Strata A, B and C).
(1) Fixed-station sampling (FS): 9, 16 and 24 stations out of the 850 potential sampling stations were selected (Li et al., 2015). Due to the special environment of hydrology and ecological in the estuary, one sampling station was set in this area to ensure that the sampling station is better representative of the study area.
(2) Simple random sampling (SR): the locations of 9, 16 and 24 stations were selected randomly at all potential stations, and without replacement in each simulation.
(3) Stratified fixed-station sampling (SFS): three strata (A, B and C) were divided mainly based on resource distribution and isobaths. Only the situation of 16 stations was selected in the stratified sampling as a control group. The station number in each stratum was allocated according to the stratum’s area and the variance of resource density in each stratum. The number of sampling station for Strata A, B and C was 10, 4 and 2, respectively.
(4) Stratified random sampling: only the situation of 16 stations was conducted and the locations of sampling station (10, 4 and 2) in each stratum (A, B and C) was selected randomly.
Three reaction distances (1.5 m, 3 m and 5 m) were also assumed to evaluate the escape ability and stress response for different ages and sizes of P. trituberculatus. We found no relevant literature and independent experiments about the reaction distance of P. trituberculatus as support. Therefore, experts and fishermen were consulted to determine such a range (1.5 m to 5 m) and then three numbers (1.5 m, 3 m and 5 m) were selected for simulation, and reasonable results should also be within this range. The above sampling designs took all three reaction distances into account. Totally 96 scenarios were assumed in the simulation study (Table 1).
Scenario number Sampling design Station number Reaction distance/m S1 fixed-station sampling 9 1.5 S2 fixed-station sampling 9 3 S3 fixed-station sampling 9 5 S4 fixed-station sampling 16 1.5 S5 fixed-station sampling 16 3 S6 fixed-station sampling 16 5 S7 fixed-station sampling 24 1.5 S8 fixed-station sampling 24 3 S 9 fixed-station sampling 24 5 S10 simple random sampling 9 1.5 S11 simple random sampling 9 3 S12 simple random sampling 9 5 S13 simple random sampling 16 1.5 S14 simple random sampling 16 3 S15 simple random sampling 16 5 S16 simple random sampling 24 1.5 S17 simple random sampling 24 3 S18 simple random sampling 24 5 S19 stratified fixed-station sampling 16 1.5 S20 stratified fixed-station sampling 16 3 S21 stratified fixed-station sampling 16 5 S22 stratified random sampling 16 1.5 S23 stratified random sampling 16 3 S24 stratified random sampling 16 5
Table 1. Different scenario schemes for the simulation study (each of the following scenario includes four seasons, so a total of 96 (24×4) scenarios were considered in the simulation study)
We built an algorithm based on probability, which was used to simulate the abundance of P. trituberculatus that can be caught in the trawl area. The schematic diagram of the bottom trawl catching P. trituberculatus shows the fishing principle (Fig. 4). The formula is as follows:
where P is the probability that each P. trituberculatus could be caught; S is the furthest distance that P. trituberculatus can move after being stimulated by the trawling drag; x is the effective opening width of trawl. The bottom trawl is 15 m wide and is divided into three areas, when the individual’s location is in the range of 0 to S, Eq. (1F) should be selected; when the range is from S to 15-S, Eq. (1M) should be used; when the range is from 15-S to 15, then Eq. (1L) should be selected.
A simulation framework for sampling design was developed (Fig. 5). “True” values of each season were calculated according to the survey data and then made the dynamic base map of resource distribution in the study area. Four sampling methods (FS, SR, SFS and SRS) with three number of sampling stations (9, 16 and 24) and three reaction distances (1.5 m, 3 m and 5 m) composed 96 scenarios. Equation (1) was used to estimate the resource abundance of all 96 scenarios and each scenario repeated 1 000 times. The performance indices (REE, RAB) were calculated to compare the performance of each scenario and choose the reasonable sampling design. At last, the reasonable sampling designs were chosen form the scenarios.
Figure 5. The flowchart of the simulation study summarizing the simulation framework for the reasonable sampling designs for fishery-independent survey using bottom trawl.
In a simulated sampling process, all the calculated P values were added to obtain the nominal probability of P. trituberculatus captured by this sample scenario, and the above process was repeated 1 000 times for each scenario in Table 1.
The relative estimation error (REE) and relative absolute bias (RAB) were calculated to evaluate the results of abundance for different sampling scenarios (Eqs (2) and (3)).
where Yiestimated is the ith estimated value of abundance, Ytrue is the “true” abundance value of each season used to make the dynamic resource base map, and R is the number of simulations. REE is used to measure the accuracy and precision of simulation results (Cochran, 1977; Chen, 1996), while RAB is used to compare the deviations of the estimators.
Impacts of the sampling design on the abundance index estimation of Portunus trituberculatus using bottom trawl
- Received Date: 2019-06-02
- Accepted Date: 2019-09-11
- Available Online: 2020-12-28
- Publish Date: 2020-06-25
- sampling design /
- number of sampling station /
- Portunus trituberculatus /
- abundance estimation /
- bottom trawl
Abstract: In the survey of fishery resources, the sampling design will directly impact the accuracy of the estimation of the abundance. Therefore, it is necessary to optimize the sampling design to increase the quality of fishery surveys. The distribution and abundance of fisheries resource estimated based on the bottom trawl survey data in the Changjiang River (Yangtze River) Estuary-Hangzhou Bay and its adjacent waters in 2007 were used to simulate the “true” situation. Then the abundance index of Portunus trituberculatus were calculated and compared with its true index to evaluate the impacts of different sampling designs on the abundance estimation. Four sampling methods (including fixed-station sampling, simple random sampling, stratified fixed-station sampling, and stratified random sampling) were simulated. Three numbers of stations (9, 16 and 24) were assumed for the scenarios of fixed-station sampling and simple random sampling without stratification. While 16 stations were assumed for the scenarios with stratification. Three reaction distances (1.5 m, 3 m and 5 m) of P. trituberculatus to the bottom line of trawl were also assumed to adapt to the movement ability of the P. trituberculatus for different ages, seasons and substrate conditions. Generally speaking, compared with unstratified sampling design, the stratified sampling design resulted in more accurate abundance estimation of P. trituberculatus, and simple random sampling design is better than fixed-station sampling design. The accuracy of the simulated results was improved with the increase of the station number. The maximum relative estimation error (REE) was 163.43% and the minimum was 49.40% for the fixed-station sampling scenario with 9 stations, while 38.62% and 4.15% for 24 stations. With the increase of reaction distance, the relative absolute bias (RAB) and REE gradually decreased. Resource-intensive area and the seasons with high density variances have significant impacts on simulation results. Thus, it will be helpful if there are prior information or pre-survey results about density distribution. The current study can provide reference for the future sampling design of bottom trawl of P. trituberculatus and other species.
|Citation:||Chunyang Sun, Yingbin Wang. Impacts of the sampling design on the abundance index estimation of Portunus trituberculatus using bottom trawl[J]. Acta Oceanologica Sinica, 2020, 39(6): 48-57. doi: 10.1007/s13131-020-1607-z|