Simulation-based Evaluation Of Sampling Design For Different Fishery Groups In Estuary

The survey of fishery resources is the basis for the exploitation and protection of fishery resources and the premise for getting the dynamics of fishery resources.At present,in the Yangtze Estuary and adjacent waters,comparison studies on transection sampling of microbenthic species in the intertidal zone and evaluation studies on fixed-station sampling have been carried out,sampling designs with different sampling methods for different fishery groups in this area are still needed to put into practice.Therefore,it is necessary to carry out relevant research on the sampling designs of fishery resources in the Yangtze estuary,so as to meet the needs of scientific research,policy formulation and implementation under the background of the gradual implementation of the "great protection of the Yangtze river" policy.So,based on the data collected in the Yangtze Estuary by fixed-station survey from 2006 to 2017,the population density was chosen as the survey objective and predicted density of each fishery group in each season by GAM was assumed as the true value respectively.Then the sampling performance of three different sampling designs including simple random sampling(SRS),systematic sampling(Sy S)and stratified sampling(St S)were evaluated by computer simulation sampling for total species,fishes,shrimps and crabs in each season,and accuracy and precision of the optimal sampling design under continuous number of sampling stations were analyzed.Meanwhile,transections were set according to the salinity gradient in the Yangtze Estuary and cluster sampling was used in the simulation sampling survey of fishery resources.In the view of sampling in reality,the optimal combination number of transection-station which adapt to the assumed sampling efforts was analyzed.The main results are listed as follows.(1)Fishery species composition and environment characters in the Yangtze Estuary.In each year,the number of species of fish was the largest among the species caught in each season,followed by the number of crabs,shrimps,mollusks and other fishery resources.The water temperature in the Yangtze Estuary was highest in Summer,followed by spring,autumn and winter.In spring,the water temperature distributed at the range of 17 to 24°C with a mean of 21°C.In summer,the range of water temperature was 26 to 31°C,and the mean was 28°C.In autumn,the water temperature followed a range of 10 to 20°C with a mean of 14°C.In winter,the water temperature distributed at the range of 6 to 11°C with a mean of 7°C.In each season from 2006 to 2017,the salinity in the Yangtze Estuary did not show a trend of gradual increase from west to east,but in the west of 121°48′E,the salinity was 0,while in the east of 121°48′E,the salinity varied greatly.Taking the observed salinity in the four seasons in 2017 as an example,in the waters east of 121°48′E,the waters with high salinity was mainly distributed in the north of the study water aera,while the waters of the south branch of the Yangtze Estuary and its natural extension had lower salinity.(2)Spatial and temporal distribution characteristics of population density of fishery groups in the Yangtze Estuary.For total species,the population density in autumn of 2017 was the highest,followed by spring,summer and winter.In terms of spatial distribution in each season,there was a relatively low value of population density in the natural extension waters of the south branch of the Yangtze Estuary,while the relatively high value of the population density appears in the research waters north of 31.5 ° N.Except that in winter,the population density of north branch was slightly higher than that of south branch.In each season,the predicted value of fishery population density was relatively higher in the study waters east of 121° 50′E and the population density in the outer waters of the Yangtze Estuary was higher than that in the inner waters of the Yangtze Estuary.Taking fishes as the research object,in summer of 2017,the density of fish species in the Yangtze Estuary was the highest,while the density of fish species in the other three seasons was relatively low.In terms of spatial distribution in each season,there was a relatively lower value of fish population density in the natural extension waters of the south branch of the Yangtze Estuary,and a relatively higher value of fish population density in the waters outside Chongming island and its natural extension area to the sea.Except that in winter,the population density of fishes in the north branch was slightly higher than that in the south branch.Taking shrimps as the research object,the population density of shrimps in the Yangtze Estuary present the highest value in autumn in 2017,and showed the lowest value in winter.In terms of spatial distribution in each season,there was a relatively lower value population density in the south branch and its natural extension waters,while the relatively higher value population density occur in the research waters north of 31.5 ° N.Except that in winter,the population density of shrimps in the north branch was slightly higher than that in the south branch.In each season,the predicted value of population density of shrimps was relatively higher in the study waters east of 121° 50′E.For crabs,in the autumn of 2017,the population density of crabs in the Yangtze Estuary was the highest,while in the other three seasons,the population density was relatively low.The relative lower value of population density of crabs existed in the natural extension waters of the south branch in each season,while the relative higher value of population density of crabs was found in the research waters north of 31.5 ° N.Except that in winter,the population density of crabs was slightly higher in north branch than that in south branch.(3)Comparison results of sampling designs in the Yangtze Estuary.Taking population density as survey objective,Relative error(REE),relative Bias(RB)and design effect(deff)were used to evaluate the accuracy and precision of five sampling designs under three different sampling efforts(12,15,and 18 stations).The five sampling designs were SRS(Design1),Sy S(Design2),and St S with three different sampling allocation schemes which were proportional allocation,average allocation,and optimal allocation criteria of sampling costs(Design3,Design4,and Design5,respectively).In general,when estimating the population density mean of each fishery group,the deff order of the five sampling designs was Design2> Design3>Design4>Design5>Design1,i.e.Sy S>St S>SRS,but the deff of Sy S was slightly lower than that of St S when estimating the population density mean of shrimps in autumn,crabs in spring and autumn.In general,Design2 had relatively higher precision and accuracy in estimating the population density mean of each fishery group(REE value was relatively smaller),followed by Design3,Design4,Design1 and Design5.However,the precision and accuracy of Design2 were slightly lower than that of Design3 when estimating the population density mean of shrimps in autumn,crabs in spring and autumn.For different survey objectives,RB values of Design1 and Design3 were symmetrically distributed around 0,and there was no overestimation or underestimation of the estimated values.For total species and fishes,the quartile interval of RB for Design2 mostly distributed in the range of-5% to 2%,which indicated that small underestimation was easily occur.For shrimps and crabs,the quartile interval of RB for Design2 distributed in the range of-7% to 0%,which indicated that underestimation was easily occur.In summary,the optimal sampling design for different fishery groups in different seasons was different.It is suggested to implement dynamic survey design according to the survey objects and seasons.(4)The optimal sampling effort adapt to the optimal sampling design for different fishery groups in the Yangtze Estuary.For total species,fishes,shrimps in spring,summer and winter,crabs in summer and winter,Sy S was selected as the optimal sampling design.In general,the variation of REE with sampling effort(n)could be divided into three intervals: 8<n<15,15<n<20,and 20<n<30.When 8<n<15,REE decreased to 10% gradually with increase of n.When 15<n<20,REE was basically stable at about 10% with the increase of n.When 20<n<30,REE decreased gradually with the increase of n.The variation trend of REE in the range of 15<n<20 indicated that the sampling accuracy of the Sy S was stable and relatively efficient with the increase of number of stations.For shrimps in autumn,crabs in spring and autumn,St S with proportional allocation method was selected as the optimal sampling design.Also the variation of REE with sampling effort(n)could be divided into three intervals in general.When 8<n<20,REE value decreased quickly with the increase of n.When 20<n<25,REE value kept at a stable value with the increase of n.When 25<n<30,REE value was approaching 5% or 7.5%.In summary,the number of 15 to 20 stations can be used as the reference range of the optimal sampling efforts for current fishery-independent survey in the Yangtze Estuary.However,under the influence of the actual conditions such as weather,sea conditions and budget costs,fewer number of stations than the reference number is accepted for relative enough sampling accuracy could be attained.(5)Results of transect sampling in the simulation sampling of different fishery groups in the Yangtze Estuary.Based on salinity gradient,31 transects was set,and the combination number of transect-station suit for different fishery groups in different seasons were different.The simulation sampling results showed that the sampling accuracy of cluster sampling as high and stable when 15 sections were sampled.For different fishery groups in different seasons,REE and RB values showed different variation rules with the increase of the number of transects and stations,and the appropriate combination number of transect-station required for sampling were also different.A hypothetic continuous number(10 to 30)of sampling stations was set to get the accordingly combination number of transect-station,and the results showed that the transect-station changes with the target research species.Additionally,if we divided the hypothetic continuous number into two ranges(10 to 20 and 20 to 30),the results indicated that at least 14 stations should be sampled for the hypothetic sampling number 10 to 20,and at least 24 stations should be sampled for the hypothetic sampling number 20 to 30,according to the different combination number of transect-station for different fishery groups in different seasons.It is difficult to determine an optimal sampling scheme to meet all the sampling objectives in a survey design.Therefore,it is necessary to comprehensively consider different sampling objectives in the survey and first meet the most important sampling objectives on the premise that the cost is feasible.