Sampling effort required to obtain repeatable average size estimates of juvenile fish

Abstract

Larval and juvenile fish cohorts often have skewed size distributions due to interindividual variation in growth rates. A resampling analysis was used to investigate how much sampling effort is required to achieve repeatable average size estimates in populations of juvenile yellowtail kingfish Seriola lalandi Valenciennes. One hundred juveniles (30 d post-hatch) were measured for total length and wet weight from 9 cohorts reared in captivity. From an analysis of cohort size distribution the mean was determined to be a suitable measure of the average for length, while the median was best for weight. One thousand resamples of mean length and median weight were performed for each cohort data set, with sample sizes ranging from 2 to 100 individuals. The point at which increasing sample size resulted in only a negligible increase in precision (variance between resamples < 0.1%) corresponded to n = 19 for mean length and n = 30 for median weight. Since yellowtail kingfish display a magnitude of size heterogeneity similar to other marine species with small pelagic larvae reared in captivity, the sample sizes determined in the present study can act as a general guideline for larval fish studies. KEY WORDS: Sampling effort · Sample size · Size heterogeneityResale or republication not permitted without written consent of the publisher

Issues of aquatic experimentation in research on marine finfish larval production in Europe and North America

“…() found that of 21 growth trials with three to five replicates, spanning five species and various developmental stages, 16 had a post hoc power below the usually desired level of 80%. Additionally, based on experimental data from yellowtail kingfish Seriola lalandi Valenciennes Moran () recommended minimum sampling per tank of 19 and 30 fish as a general guideline for analysis of total length and wet weight, respectively. However, some of the above‐mentioned studies indicate even larger sample sizes to reach 80% power, depending on the variability of data, tank replication and the magnitude of treatment effects to detect.…”