Till startsida
To content Read more about how we use cookies on gu.se

Sampling fish

Sampling fish using sampling underwater detonationSampling fish using underwater detonation. Photo: Göran Sundblad

The choice of method for sampling fish depends on what species of fish and life-stage you want to map, and in which habitat. Compared to sampling for direct mapping, predictive mapping put somewhat different demands on the method.

The distribution of fish is highly variable and depends on whether the species is pelagic or demersal, occupies coastal or open water, or prefers fresh or saline water. In addition, differences may appear within species as different life-stages, such as eggs, larvae, juveniles and adults, often occupy different ecological niches and habitats. The distributions of adult fish may also differ with season as they often utilise specific habitats for foraging and spawning. Therefore, in sampling fish, choice of method and gear type depends on what habitat, species and life-stage is targeted.

Data from a wide-range of habitats
To attain data for modelling the distribution and/or abundance of fish it is important to choose a sampling method that can be used in the whole potential distributional range of the species, not only in its optimum habitats. In this regard, sampling for modelling differs from sampling for direct mapping, see below. The reason is that predictive performance of most modelling techniques is dependent on information from samples taken also outside the optimal conditions for the speices in question. This requires more general sampling methods that are applicable in a range of habitat types; such as gillnets and fyke nets for juvenile and adult fish, as well as hydroacoustics. For earlier life-stages (eggs, larvae and young-of-the-year) diving/snorkelling, small underwater detonations and to some extent, white plate and scoop can be used in more heterogeneous habitats.

Sampling in points increases precision
Furthermore, we urge potential modellers to consider type of sampling (i.e. transect or point/quadrate) so that the statistical assumptions of the modelling technique are not violated. To date, most predictive modelling of fish in the Baltic Sea coastal regions have been done using active methods that yield point estimates, as this is generally considered the most appropriate approach. Passive methods (e.g. nets) and nested transect data can of course be useful for predictive mapping, but usually require more advanced statistical techniques.

For example, most basic modelling techniques, such as linear regression, GLM and GAM, assume that samples are independent and identically distributed. Violations of this assumption may result in biased precision of the estimates and ultimately wrongly specified models, while tree-based techniques, e.g. RandomForest, are not as restricted by this assumption.

Direct mapping requires more specific sampling
For direct mapping of the distribution and/or abundance of fish, the choice of survey method should be adapted for sampling in the main habitats of the target species and life-stage. Specifically tailored methods are more likely to yield high precision estimates compared to methods that are not optimal for the habitat of interest. For example, if the aim of a survey is to map and compare fish densities in two different areas, the choice of method should be made in relation to the type of habitat.

Methods should be comparable
An important thing to consider if areas are sampled with different methods, is that the methods have comparable catchability. Since there are ususally differences in catchability between gears, it is often difficult to pool data collected with different techniques, regardless if the aim of the study is direct mapping or predictive modelling.

Methods for sampling fish
Method Main habitat Life-stage Type of sampling Measurement unit
Bottom trawl Benthic/demersal, depth >30m Adult Transect Ind/effort or ind/m2
Midwater trawl Pelagic, depth >30m Adult Transect Ind/effort or ind/m3
Hydroacoustic w. pelagic trawl Pelagic Young-of-the-year/juvenile/adult Transect Tonnes/volume
Bongo sampler Pelagic Larvae Transect Ind/volume
Gulf (Olympia) sampler Pelagic Larvae  Transect Ind/volume
Nordic coastal multimesh gillnet Coastal area 0-20m depth Juvenile/adult Point Ind/effort
Fyke nets Coastal area, 0-20m depth Juvenile/adult Point Ind/effort
Beam trawl Sandy/muddy, 0-6m depth Juvenile  Transect Ind/effort
Drop trap Near shore, Sandy/Muddy, 0-6m depth Young-of-the-year Point Ind/m2
Pushnet Near shore, sandy/muddy, 0-6m depth Young-of-the-year Transect Ind/effort
Diving/snorkelling survey Near shore, all habitats, (snorkelling limited to 0-6m depth) Egg Transect/point Presence/absence
White plate and scoop Near shore, (partly) vegetated, 0-2m depth  Larvae Transect/point Presence/absence  
 Beach seine Near shore, partly vegetated, 0-6m depth, mainly sandy/muddy Young-of-the-year Point Ind/effort
Small underwater detonations Near shore, all habitats, 0-6m depth Young-of-the-year Point Ind/m2

Methods for sampling fish in different habitats and life-stages. Bold indicates methods included in Prehab. Modified from Bergström et al 2007.


More information about fish sampling methods and their application in the Baltic Sea are found here.

Photo: Ulf Bergström

© University of Gothenburg, Sweden Box 100, S-405 30 Gothenburg
Phone +46 31-786 0000, About the website

The University of Gothenburg uses cookies to provide you with the best possible user experience. By continuing on this website, you approve of our use of cookies.  What are cookies?