Effect of different sampling designs and methods on the estimation of secondary production: A simulation

Abstract

IntroductionIn marine studies, the secondary production of macrobenthic populations can be estimated in a number of ways. Classic techniques can be broadly classified as cohort-and size-based. Prominent in the first group are the Allen-curve, incrementand removal-summation, and instantaneous-growth methods. The second group consists mainly of the size-frequency, massspecific growth rate, and mass-specific mortality rate methods.These estimation techniques, and numerous variants thereof, have been described, e.g., by Benke (1984), Crisp (1984), andRigler andDowning (1984). They have been extensively compared through simulations, most notably by Cushman et al. (1978), Lapchin andNeveu (1980), andMorin et al. (1987). These and other authors examined how various estimates were affected by different assumptions on the growth and mortality curves, as well as by sampling effort. Their hypothetical populations, which were alternately assumed to be synchronous or not, were modeled on freshwater species having a lifespan of 1 year or less.This article revisits the issue and expands on the comparison of the above production estimates, and several of their variants introduced in the literature, using a hypothetical population of mussels characterized by a number of realistic factors. Most prominent among them are the simultaneous presence of different cohorts, quadrat-dependent population density, seasonal growth oscillation, gradual recruit arrival, and random individual variation both in survival and in weight gain.The simulations presented herein are also novel in that several aspects of the sampling design are considered, namely the number of sampling occasions (either regularly spaced or concentrated in the growing season), quadrats sampled per occasion, and use of two different sieves. Following Cushman et al. (1978), the bias and variance of the various estimators are compared to the exact production obtained by summing all individual body-mass gains, rather than to a theoretical Allen curve which may or may not reflect reality. Effect of different sampling designs and methods on the estimation of secondary production: A simulation AbstractThis article reports the results of a simulation study designed to investigate the effect of several sampling design factors on the accuracy and precision of various estimates of secondary production. Whereas most previous studies of this sort were concerned with freshwater fauna (e.g., insects), the hypothetical population used here reflects the characteristics of marine mussels from cold-temperate and subarctic regions. It features the simultaneous presence of different cohorts, gradual recruit arrival, seasonal growth oscillation, and quadratdependent population density, as well as random individual variation both in survival and in weight gain. For this population, the percentage relative bias (PRB) and relative root mean squared error (RRMSE) of 4 classic cohort-based methods, 3 size-based methods, and several variants thereof were computed as a function of sampling frequency...

Have Invasive Mysids (Mysis diluviana) Altered the Capacity of Osoyoos Lake, British Columbia to Produce Sockeye Salmon (Oncorhynchus nerka)?

“…On the other hand, simulation studies using "asynchronous" cohorts, i.e. not born all on the same day (as is true of the species examined here), have shown the sizefrequency method to be biased low [55,58].…”

“…The first of these is the well-known increment summation method, which sums the growth increments in biomass during a cohort's lifespan. The version we used was developed by Cusson et al [55] and is described as their method "IS-3." The second is the size-frequency method [56], which sums the loss of biomass between successive size classes and weights them by the number of days between surveys.…”

Evaluation of Brey’s production/biomass model on the basis of a long-term data set on a clam population

“…This method of assessing annual production by summing seasonal production estimates is superior to calculations using only once-per-year estimates of numbers and individual weights (as explained by Dekker & Beukema 2007). Sampling more frequently than twice annually would certainly have resulted in more precise estimates of annual P, as shown in a simulation study by Cusson et al (2006). They show ed that both bias as well as random errors declined with sampling frequency.…”

“…As a consequence of some assumptions (e.g. on the course of the decline of numbers of cohort members between samplings), the precision of direct estimates is not perfect, but will be satisfactory in most cases (Cusson et al 2006). Direct methods are relatively simple and straightforward if separate cohorts can be distinguished.…”

Seasonality in coastal macrobenthic biomass and its implications for estimating secondary production using empirical models

“…Several studies investigated effects of sampling design on estimation of secondary production (Cushman et al ; Morin et al ; Cusson et al ). They showed that the choice of sampling period and frequency affect the estimation, and inappropriate choices can lead to high bias in production estimates.…”