Calibration of fish biomass estimates from gillnets: Step towards broader application of gillnet data

Abstract

Fish are an important component of aquatic ecosystems, thus representative and reliable assessments of their population variables are essential for a variety of ecological applications, management and conservation. Determining Fish Density per actual Spatial Unit (volume or area, FDSU) as a measure of absolute fish quantity is of particular interest. Gillnets are undoubtedly one of the most common and important methods for assessing fish populations in large lentic waters. However, direct calculating of FDSU from gillnet catches is impossible because of the passive nature of this method, and to date there is no reliable model for calculating FDSU from gillnet catches. This weakness largely limits the use of gillnet data for applications requiring FDSU estimates.

The aim of this study was to calibrate gillnet catches using FDSU obtained by active methods (beach seine nets and hydroacoustics) to develop a tool for assessing FDSU from gillnet catches. To achieve this goal, we compared gillnet biomass to fish biomass estimated from the active methods, both of which cover similar spatiotemporal niches. This comparison was performed using a statistical approach based on the recognition of non-negligible random measurement error in both the explanatory (active methods) and response (gillnets) variables.

We found a strong positive linear relationship between fish biomasses sampled with gillnets and with active methods. The slope of the fitted linear model was similar when comparing gillnets with the two active methods. The statistical method used allowed for the inclusion of error in the biomass estimates with gillnets and active methods, refining the credible intervals of the estimated relationship. The effect of gillnet effort on model accuracy was simulated to show how increased effort narrows the credible interval. Finally, comparison with previously published relationships revealed a large but explainable discrepancy between our model and previous models.

Our study showed that conversion of gillnet biomass to biomass per actual spatial unit is possible. The effective sampling area of one square meter of gillnet was determined to be 8 m2 of waterbody surface area when European standard 12 mesh-sizes gillnets are used, and 5 m2 when four larger meshes are added to the European standard gillnets. Our model further stressed the impact of increased sampling effort on reducing estimation variability and shows that the model may be dependent on the fish community.