Optimization Of Sample Sites And Sizes For Fishery Resources Survey In Dianshan Lake

A basic requirement for a regular fisheries stock assessment and management is an estimate of fish biological characteristics,species richness and biomass distribution.Data are usually obtained through well-designed fishery-independent survey programs.Sample size plays a significant role in determining accuracy and precision for estimates of fish population dynamics or fish communities.For fish population with low abundance and aggregated distribution in a lake ecosystem,high intensity bottom trawl survey may result in extra mortality and disturbance to benthic community,imposing unnecessarily large negative impacts on the populations and ecosystem.Optimization of sampling design is necessary to acquire cost-effective sample size.We developed a simulation approach to evaluate and optimize sampling size for lake fishery resources survey with multiple goals including estimation of fish composition,abundance indices of individual species and fish group,species diversity indices,and fish length frequency distribution.Monthly surveys were conducted from July,2010 to June,2011,at 24 survey stations in Dianshan Lake resulting in the collection of 45 fish species from 14 families.A resampling technique for survey was used to estimate the optimal survey times and number of survey stations for the Dianshan Lake fish composition.Results showed a negative relationship between the number of stations per survey and survey times per year.To maintain a 90% fixed percentage of fish species detection and 95% fixed probability of detection within the fishery survey,the required sample sizes were 19 and 21 stations per survey,and 12 and 11 survey times per year,correspondingly.This study suggests that,because of monthly variation of fish species composition in the Dianshan Lake,10 or more survey than 10 times done in a year can reflect the true composition of fish species in this lake.For the case where the objective of study was to collect data for the estimation of fish abundance indices of individual species and fish groups.Results showed that,for each month,the REE for all indices decreased when the sample size increased.The first set of increases in sample size reduced the REE much more than the subsequent increases.For all the indices,a higher CV of the key indices tend to lead to a higher REE of the indices.The REE had different levels of values for each index in different seasons.The REE was relatively low for the species diversity indices compared to other indices,indicating that reduced sample size could still yield a relatively high precision and accuracy of estimation for the species diversity indices.For the case where the objective of study was to estimate Lake Dianshan's fish length frequency distribution.Results showed that,the relationship between sample size and MSE followed a negative exponential model and became asymptotic to the x-axis.In general,smaller species,smaller populations,and simpler length statistic required fewer samples.Similarly,the sample size required to adequately estimate length distributions was 7 or 8 stations or 400 to 600 fishes.The sample size was effected by the number of length intervals,length-frequency shape,and maximum fish size.This study suggests that sample sizes in a lake fishery resources survey can be reduced while still achieving relatively high precision and accuracy for most indices.This study also shows that optimal sample size may vary with survey objectives.Thus,we can reduce the sample size in Dianshan Lake fishery resources survey appropriately according to different survey objectives.This study provides a framework for determining the optimal sampling effort in the similar studies at other locations.